Monetary Policy and the Economy Q4/20 – Q1/21
Call for applications: Klaus Liebscher Economic Research Scholarship
Please e-mail applications to email@example.com by the end of October 2021. Applicants will be notified of the jury’s decision by end-November.
The Oesterreichische Nationalbank (OeNB) invites applications for the “Klaus Liebscher Economic Research Scholarship.” This scholarship program gives outstanding researchers the opportunity to contribute their expertise to the research activities of the OeNB’s Economic Analysis and Research Department. This contribution will take the form of remunerated consultancy services.
The scholarship program targets Austrian and international experts with a proven research record in economics and finance, and postdoctoral research experience. Applicants need to be in active employment and should be interested in broadening their research experience and expanding their personal research networks. Given the OeNB’s strategic research focus on Central, Eastern and Southeastern Europe, the analysis of economic developments in this region will be a key field of research in this context.
The OeNB offers a stimulating and professional research environment in close proximity to the policymaking process. The selected scholarship recipients will be expected to collaborate with the OeNB’s research staff on a prespecified topic and are invited to participate actively in the department’s internal seminars and other research activities. Their research output may be published in one of the department’s publication outlets or as an OeNB Working Paper. As a rule, the consultancy services under the scholarship will be provided over a period of two to three months. As far as possible, an adequate accommodation for the stay in Vienna will be provided. 1
Applicants must provide the following documents and information:
- a letter of motivation, including an indication of the time period envisaged for the consultancy
- a detailed consultancy proposal
- a description of current research topics and activities
- an academic curriculum vitae
- an up-to-date list of publications (or an extract therefrom)
- the names of two references that the OeNB may contact to obtain further information about the applicant
- evidence of basic income during the term of the scholarship (employment contract with the applicant’s home institution)
- written confirmation by the home institution that the provision of consultancy services by the applicant is not in violation of the applicant’s employment contract with the home institution.
1 We assume that the coronavirus crisis will abate in the course of 2021. We are also exploring alternative formats to continue research cooperation under the KLERS program for as long as we cannot resume visits due to the pandemic situation.
in English and German
Nontechnical summaries in English
Monitoring the economy in real time with the weekly OeNB GDP indicator: background, experience and outlook
Gerhard Fenz, Helmut Stix
Similar to most other industrialized countries, the COVID-19 pandemic triggered a deep and abrupt slump in economic activity in Austria. As traditional economic indicators are only available with a certain time lag, indicators that are a lot more timely were called for in this special situation. Against this background, we developed the new, experimental OeNB GDP indicator (OeNB GDP-I).
The OeNB GDP-I relies on a demand-side approach on estimating GDP. It (1) provides weekly estimates of economic activity in Austria, (2) provides estimates of the major GDP components, (3) focuses on year-on-year changes and (4) considers shifts from cash to noncash consumer spending, which makes it possible to obtain a relatively accurate estimate of consumption growth. The results of the OeNB GDP-I have been published on a regular basis since early May 2020, thus providing real time information on the state of the Austrian economy.
In this study, we present the OeNB GDP-I and its main results, explain how it was constructed, discuss its pros and cons and draw some (preliminary) lessons from more than half a year of weekly nowcasting.
Compared with Austrian GDP figures that have been published so far, the OeNB GDP-I has proven to be a valid and informative instrument suitable for capturing developments in the current economic crisis. It differs from well-known international economic indicators in that it is data driven (for reasons of data availability), while most other economic indicators rely on time series models. The unavailability of longer time series restricts the extent of possible validation. This means that the OeNB GDP-I is, and will remain, an instrument to be used in times of crises and not for observing economic developments in “normal” times. In contrast, a number of real-time (sub)indicators (in particular payments data) that have been employed in economic analyses at the national and international level since the outbreak of the COVID-19 pandemic will be of use also in economically calmer times. These indicators have turned out to be highly informative and can provide important insights, e.g. when studying the consumption response to a fiscal stimulus.
Austrian tourism sector badly hit by COVID-19 pandemic
Gerhard Fenz, Helmut Stix, Klaus Vondra
The tourism sector is an important pillar of the Austrian economy, accounting for almost 7½% of Austrian GDP. By European standards, this is an above-average contribution. We use weekly data on payment card spending and monthly data on overnight stays to analyze the strong impact of the COVID-19 crisis on Austrian tourism. During the lockdown in spring 2020, overnight stays by tourists dropped by almost 100%. While domestic tourists returned quickly after accommodation facilities reopened, foreign tourists (mainly from continental Europe) took a few weeks longer; most overseas tourists have not returned at all since the COVID-19 pandemic broke out in Austria in March 2020. Over the summer of 2020, tourism activity in Austria recovered strongly, backed by domestic and German tourists. Still, it remained clearly below 2019 levels. In October 2020, the renewed increase in the number of COVID-19 infections caused another severe downturn in Austrian tourism (total overnight stays: –49.3%, domestic tourists: –13.7%, foreign tourists: –66.8%), as several neighboring countries posted travel warnings for Austria. On November 2, 2020, a second lockdown was imposed on Austria. Accommodation establishments and restaurants were closed. Basing our estimations on payment card data, we expect a decline of 93% in overnight stays (domestic tourists: –90%, foreign tourists: –95%) for November 2020 compared to November 2019. According to the new rules communicated by the Austrian government on December 2, 2020, Austrian accommodation establishments will not open before January 2021. Moreover, travel warnings by major countries of origin (especially Germany) remain in place at least until the end of the year. Therefore, we expect overnight stays to drop by 95% in December. For the full year 2020, we expect a 36% decrease in total overnight stays, mainly because of the strong decline in overnight stays by foreign tourists (–41%). Overnight stays by domestic tourists, by contrast, will go down by 23%. Had the lockdowns fallen into the high tourist season, the overall decline in overnight stays might have been far stronger. If containment measures and travel warnings remain in place in the first months of 2021, severe losses in the Austrian accommodation and winter tourism industry are very likely. This would also have a strong impact on total Austrian tourism in 2021.
Prices and inflation in Austria during the COVID-19 crisis – an analysis based on online price data
Christian Beer, Fabio Rumler, Joel Tölgyes
The COVID-19 pandemic and the accompanying policy measures have affected both the demand and supply side of the Austrian economy, and consequently also consumer prices, in multiple ways. Apart from affecting prices, the COVID-19 pandemic also made it difficult to collect price data for inflation measurement. Statistical offices had to resort to imputations and the use of scanner data when price data could not be collected directly from shops. To gain insights into price developments during the first stage of the COVID-19 pandemic in Austria, the OeNB has collected price data from several online shops via webscraping, i.e. automatic daily downloads of large amounts of online price data, since the beginning of April 2020.
Based on these webscraped data, we analyzed price developments of those product categories that became especially relevant during the COVID-19 crisis, i.e. food and beverages, medical products, IT equipment, personal care products and delivered meals. Our observation period for most products is from April to August 2020. Our results suggest that, contrary to what the media occasionally reported, prices for food and nonalcoholic beverages showed an – albeit rather small – decline over the observation period while prices for alcoholic beverages and medical products did not show a clear upward or downward trend. For personal care products and IT equipment, we find a price increase in the first half of the observation period followed by a somewhat more pronounced price decline in the second half. In contrast, for meals delivered by a meal delivery service provider, we observe a steady – but rather small – price increase over the observation period (in this case, from mid-June to end-August 2020).
A comparison of the results derived from online data and official figures from the Harmonized Index of Consumer Prices (HICP) for Austria shows similar price developments for food and personal care products, but some differences for the remaining product categories. The latter may be attributable to conceptual differences in product and store coverage.
Have mitigating measures helped prevent insolvencies in Austria amid the COVID-19 pandemic?
Claus Puhr, Martin Schneider
In this study, we assess the impact of the COVID-19 pandemic on companies in Austria. Using a novel insolvency model, we estimate their risk of becoming insolvent. Our model reflects companies’ balance sheets as well as profit and loss statements. The economic impact of the pandemic varies strongly among industries. This is why we implemented the model for 17 economic sectors. As a result of the pandemic and lockdown measures, economic activity has fallen sharply. As a consequence, many companies and households have been facing an existential threat. Government and private mitigating measures have helped cushion the blow. Support for companies includes grants and subsidies (e.g. fixed cost grants and short-time work), deferrals of short- and long-term payment obligations, credit guarantees and changes to the insolvency law. In this analysis, we considered measures until August 31, 2020.
Our model shows that COVID-19 leads to a marked increase in corporate insolvencies. Without mitigating measures, the number of insolvencies in 2020 would have increased sixfold compared with previous years. But the mitigating measures in place helped reduce this number by two-thirds. The insolvency rates we predict based on our model should be interpreted with caution. Most importantly, our model allows us to compare and rank the mitigating measures. We find, for instance, that credit guarantees appear most effective, followed by fixed cost support and short-time work. In the short term, delayed filing for insolvency is most efficient. Yet, this effect is set to reverse itself in 2021, when public institutions are likely to return to their usual practice.
How has COVID-19 affected the financial situation of households in Austria?
Nicolas Albacete, Pirmin Fessler, Fabian Kalleitner, Peter Lindner
This study discusses the potential effects of the COVID-19 crisis on household finances in Austria. We use data from the Austrian Corona Panel Project carried out by the University of Vienna as well as data from the Eurosystem Household Finance and Consumption Survey for Austria.
In the first part of the study, we illustrate that different individuals and households have been exposed to the COVID-19 crisis in very different ways and to varying degrees. Households with a small living space, such as larger households with children, households with single parents or households living in densely populated areas, are more exposed to income shocks stemming from COVID-19. Income from pensions and other public transfers serve as an important buffer for poorer households against potential impacts of the COVID-19 crisis, as these income sources have not (yet) been exposed to the effects of the crisis. Furthermore, we find that the median household might be able to compensate for financial losses for a relatively long time by drawing on its liquid assets such as savings. Thus, putting the focus on those households who are not able to make up for losses incurred during the COVID-19 crisis, such as single-parent households or households with unemployed household members, seems warranted.
In the second part of the study, we analyze potential impacts of the COVID-19 crisis. Our analysis suggests that households’ income losses averaged about 12% during the first lockdown in April 2020; this percentage would double if one-third of employees on short-time work became unemployed. Moreover, tenants suffered particularly large income losses. Although households’ attitudes toward consumption were negatively affected at the onset of the COVID-19 crisis, they have improved over time. Uncertainties remain high, however. Saving attitudes were also surrounded by high uncertainties, but we find some weak evidence of increasingly positive attitudes for high-income households over time.
Support measures should mainly target those households who were in a difficult social, economic and financial situation already before the COVID-19 crisis and who suffered the largest income losses during the pandemic to protect them from further financial and social harm.
The effects of the monetary policy response to the COVID-19 pandemic: preliminary evidence from a pilot study using Austrian bank-level data
Claudia Kwapil, Kilian Rieder
The Eurosystem’s monetary policy response to the economic impact of the COVID-19 pandemic has been swift and powerful. Its policy package contained both extensions of existing unconventional monetary policy measures and new instruments geared to address the extraordinary economic challenges resulting from the COVID-19 pandemic. This pilot study analyzes the effects of one important building block of the monetary policy rescue package – the targeted longer-term refinancing operations (TLTROs) – on Austrian banks’ credit supply. In spring 2020, the conditions of the latest generation of TLTROs (TLTRO III) were relaxed substantially in view of the COVID-19 pandemic: volumes were expanded, interest rates were lowered and collateral requirements were reduced. We analyze whether those banks that borrowed more funds in the June 2020 TLTRO III (i.e. after the above-mentioned relaxation) did in fact extend more loans to customers in July, August and September 2020. Using data on Austrian banks and applying an instrumental variable strategy, we approximate the causal relationship between TLTRO take-up and banks’ credit supply. We find evidence for an unambiguously positive effect of TLTRO participation on new lending in Austria. The estimated elasticity of credit supply ranges between 0.26 and 1.00, depending on the period and credit categories covered.
Unprecedented fiscal (re)actions to ease the impact of the COVID-19 pandemic in Austria
Austria’s public finances have played a major role in mitigating the effects of the COVID-19 pandemic on the economy. First, automatic stabilizers have cushioned parts of the economic downturn. Second, unprecedented active fiscal policy measures were taken both at the national and the EU level to further support the economy.
In Austria, fiscal policy measures adopted during the lockdown periods in spring and November/December 2020 were mainly aimed at ensuring that the health care system remains fully operational and at supporting businesses (fixed cost grant, net turnover compensation) and households (short-term work scheme, hardship funds). Compensating businesses and households for income losses suffered because of the containment measures has helped maintain the economy’s production capacity. The latter would have been lost if viable firms and jobs had been permanently destroyed.
The measures enacted since the summer 2020 had a twofold purpose. First, restarting the economy by taking classic stimulus measures (cut in income taxes and VAT for certain sectors, one-off social payments) was key after the lockdown periods. These measures were meant to stimulate consumer demand, in particular from liquidity-constrained households. Second, initiatives were taken to promote private investment (carryback of 2020 losses, accelerated depreciation, investment premium) and public investment (federal cofinancing of local government investment, higher investment budgets). Ideally, these investments promote long-term objectives, such as the decarbonization and greening of the economy. In doing so, they support the transition to new technologies and ways of working, put the economy on a sustainable footing and thereby increase its long-term growth potential.
However, given the high uncertainty surrounding the economic outlook, measures might be less effective than during normal times. Households and businesses might just “wait and see” rather than consume and invest. Moreover, policy measures must be unwound with caution to avoid crisis legacy issues that might hamper the economic recovery.
The costs associated with the unprecedented fiscal measures and automatic stabilizers have left their mark on Austria’s public finances. In 2020, Austria is likely to see the largest budget deficit since 1995. Nevertheless, the sustainability of Austria’s public finances should not be at risk, as Austria went into the crisis with a sound fiscal position. However, as low interest rates might not stay around forever, the high debt ratio should be reduced in a socially and environmentally sustainable way.
Nontechnical summaries in German
Wirtschaftsbeobachtung in Echtzeit mit dem wöchentlichen OeNB-BIP-Indikator: Hintergrund, Erfahrungen, Ausblick
Gerhard Fenz, Helmut Stix
Wie in vielen anderen industrialisierten Staaten hat die COVID-19-Pandemie auch in Österreich zu einem abrupten und tiefen Einbruch der wirtschaftlichen Aktivität geführt. Da traditionelle wirtschaftliche Indikatoren erst mit einer gewissen zeitlichen Verzögerung verfügbar sind, bestand der Bedarf nach Indikatoren die wesentlich zeitnäher zur Verfügung stehen. Vor diesem Hintergrund wurde der neue experimentelle OeNB-BIP-Indikator entwickelt.
Der OeNB-BIP-Indikator basiert auf einer Messung der nachfrageseitigen Komponenten des BIP. Er (i) bietet eine Schätzung der wirtschaftlichen Aktivität auf wöchentlicher Basis, (ii) bietet Schätzungen des Wachstumsbeitrags der Hauptkomponenten des BIP, (iii) stellt die Entwicklung im Jahresvergleich dar und (iv) berücksichtigt Verschiebungen zwischen baren und unbaren Konsumausgaben, was eine relativ genaue Einschätzung der Konsumentwicklung zulässt. Die Ergebnisse des OeNB-BIP-Indikators wurden sein Anfang Mai 2020 regelmäßig veröffentlicht. Damit konnten die wirtschaftspolitischen Akteure und die Öffentlichkeit zeitnah informiert werden.
Im vorliegenden Beitrag präsentieren wir den Indikator sowie die Hauptergebnisse, erläutern seine Konstruktion, diskutieren seine Vor- und Nachteile und ziehen eine (vorläufige) Bilanz nach mehr als einem halben Jahr wöchentlichen Nowcastens.
Im Großen und Ganzen hat sich der BIP-Indikator bisher als sehr nützliches und valides Instrument in Zeiten der aktuellen wirtschaftlichen Krise erwiesen. Er unterscheidet sich von anderen bekannten internationalen Indikatoren, in dem er – aus Gründen der Datenverfügbarkeit – „datengetrieben“ ist, während die meisten anderen Indikatoren auf Zeitreihenmodellen beruhen. Dies schränkt das Ausmaß der möglichen Validitätsprüfungen ein. Insofern ist und bleibt der OeNB-BIP-Indikator ein Instrument, das krisenbezogen eingesetzt wird und nicht zur Konjunkturbeobachtung in „normalen“ Zeiten dient. Im Gegensatz dazu werden etliche Echtzeitindikatoren, die national und international seit Ausbruch der COVID-19-Pandemie zur Wirtschaftsanalyse verwendet werden, insbesondere Zahlungsverkehrsdaten, auch in wirtschaftlich ruhigeren Zeiten eingesetzt werden. Sie haben sich als ausgesprochen informativ erwiesen und eröffnen neue Möglichkeiten der Wirtschaftsanalyse.
Österreichischer Tourismussektor von COVID-19-Pandemie stark betroffen
Gerhard Fenz, Helmut Stix, Klaus Vondra
Der Tourismussektor stellt eine wichtige Stütze der österreichischen Wirtschaft dar. Rund 7½% des österreichischen BIP entfallen auf diesen Sektor. Im europäischen Vergleich ist dieser Wert überdurchschnittlich hoch. Die beträchtlichen Auswirkungen der COVID-19-Krise auf den österreichischen Tourismus werden in diesem Beitrag auf Basis wöchentlich erhobener Kartenzahlungsdaten und monatlich erhobener Nächtigungszahlen analysiert. Während des Lockdowns im Frühjahr 2020 gingen die Nächtigungen im österreichischen Fremdenverkehr um beinahe 100% zurück. Während die Anzahl inländischer Touristen nach der Wiedereröffnung der Beherbergungsbetriebe rasch wieder anstieg, dauerte dies bei den ausländischen Touristen (in erster Linie aus Kontinentaleuropa) einige Wochen länger; die meisten Touristen aus Übersee sind seit dem Ausbruch der COVID-19-Pandemie in Österreich im März 2020 ausgeblieben. Über die Sommermonate 2020 verzeichnete der österreichische Tourismus eine starke Erholung, die hauptsächlich auf inländische und deutsche Gäste zurückzuführen war. Dennoch blieben die Nächtigungen deutlich unter dem Vorjahrsniveau. Im Oktober 2020 führte die neuerlich ansteigende Zahl an COVID-19-Infektionen zu einem weiteren starken Rückgang im heimischen Fremdenverkehr (Nächtigungen insgesamt: –49,3%; inländische Touristen: –13,7%; ausländische Touristen: –66,8%). Am 2. November 2020 wurde in Österreich ein zweiter Lockdown verhängt. Beherbergungsbetriebe und die Gastronomie wurden geschlossen. Auf Basis von Schätzungen anhand der Kartenzahlungsdaten ist im November 2020 im Vorjahrsvergleich mit einem Rückgang von 93% bei den Nächtigungen zu rechnen (heimische Touristen: –90%; ausländische Touristen: –95%).
Gemäß den von der österreichischen Bundesregierung am 2. Dezember 2020 verlautbarten neuen Regelungen werden Beherbergungsbetriebe in Österreich nicht vor Jänner 2021 wieder öffnen. Darüber hinaus werden Reisewarnungen der wichtigsten touristischen Herkunftsländer (insbesondere Deutschlands) zumindest bis Jahresende in Kraft bleiben. Insgesamt ist somit im Dezember ein Rückgang der Nächtigungen von 95% zu erwarten. Für das Gesamtjahr 2020 wird von einem 36-prozentigen Rückgang der Nächtigungen ausgegangen, was in erster Linie auf das deutliche Minus bei den Nächtigungen ausländischer Touristen zurückzuführen ist (–41%). Der Nächtigungsrückgang bei inländischen Touristen hingegen wird 23% ausmachen. Wären die Lockdowns in die Hochsaison gefallen, so wären die Nächtigungszahlen vermutlich noch stärker zurückgegangen. Wenn die Eindämmungsmaßnahmen und Reisewarnungen in den ersten Monaten des Jahres 2021 aufrecht bleiben, sind hohe Verluste in der österreichischen Beherbergungs- und Wintersportindustrie sehr wahrscheinlich. Dies würde auch den Gesamttourismus 2021 in Österreich stark beeinträchtigen.
Preis- und Inflationsentwicklung in Österreich während der COVID-19-Krise – eine Analyse anhand von Online-Preisdaten
Christian Beer, Fabio Rumler, Joel Tölgyes
Die COVID-19-Pandemie und die Maßnahmen zu ihrer Bekämpfung wirken sich sowohl nachfrage- als auch angebotsseitig auf vielfache Weise auf die österreichische Wirtschaft, und somit auch auf die Verbraucherpreise, aus. Abgesehen von ihrer Auswirkung auf die Preise erschwert die COVID-19-Pandemie auch die Erhebung von Preisdaten zur Inflationsmessung. So mussten die statistischen Ämter mitunter auf Schätzungen und Daten von Supermärkten zurückgreifen, da eine Erhebung von Preisdaten vor Ort in den Geschäften nicht immer möglich war. Um Aufschluss über die Preisentwicklung während der ersten Phase der COVID-19-Pandemie in Österreich zu erhalten, erhebt die Oesterreichische Nationalbank (OeNB) seit Anfang April 2020 Preisdaten unterschiedlicher Online-Geschäfte. Dabei kommt die Methode des Webscraping zum Einsatz, d. h. der automatische tägliche Download großer Mengen von Online-Daten.
Auf Basis dieser Daten wurde die Preisentwicklung jener Produktkategorien analysiert, die während der COVID-19-Krise besonders ins Blickfeld gerückt sind: Nahrungsmittel und Getränke, medizinische Produkte, IT-Ausrüstung, Körperpflegeprodukte und Essenszustellungen. Der Beobachtungszeitraum für die meisten Produkte umfasst die Monate von April bis August 2020. Die Analyseergebnisse legen nahe, dass die Preise für Nahrungsmittel und alkoholfreie Getränke – entgegen teils anders lautenden Medienberichten – im Beobachtungszeitraum zurückgegangen sind (wenn auch nur leicht), während die Preise für alkoholische Getränke und medizinische Produkte keinen klaren Auf- oder Abwärtstrend erkennen ließen. Für Körperpflegeprodukte und IT-Ausrüstung zeichnete sich in der ersten Hälfte des Beobachtungszeitraums ein Preisanstieg ab, auf den in der zweiten Hälfte des Beobachtungszeitraums ein etwas deutlicherer Preisrückgang folgte. Bei Essenszustellungen durch einen Lieferservice hingegen konnte ein stetiger, wenn auch geringer, Preisanstieg festgestellt werden (Beobachtungszeitraum: Mitte Juni bis Ende August 2020).
Ein Vergleich der aus den Online-Daten abgeleiteten Ergebnisse mit den offiziellen Ergebnissen des Harmonisierten Verbraucherpreisindex (HVPI) für Österreich zeigt ähnliche Preisentwicklungen bei Nahrungsmitteln und Körperpflegeprodukten, jedoch einige Unterschiede in den übrigen Produktkategorien. Diese Abweichungen könnten auf konzeptionelle Unterschiede in der statistischen Erfassung von Produkten und Geschäften zurückzuführen sein.
Der Beitrag von Hilfsmaßnahmen zur Vermeidung von COVID-19-bedingten Unternehmensinsolvenzen in Österreich
Claus Puhr, Martin Schneider
Wir beleuchten in dieser Studie, wie sich die COVID-19-Pandemie auf Unternehmen in Österreich auswirkt. Mithilfe eines neu entwickelten Insolvenzmodells schätzen wir, wie hoch das Insolvenzrisiko der Unternehmen aufgrund der wirtschaftlichen Folgen der Pandemie sein wird. Unser Modell basiert auf Unternehmensdaten und bildet die Bilanz sowie die Gewinn- und Verlustrechnung ab. Da sich die wirtschaftlichen Folgen der Pandemie stark nach Branchen unterscheiden, haben wir das Modell für 17 Branchen implementiert. Die Pandemie und die Lockdown-Maßnahmen zur Eindämmung des Coronavirus haben zu einem starken Einbruch der wirtschaftlichen Aktivität geführt. Dies stellt für viele Unternehmen und private Haushalte eine existenzielle Bedrohung dar. Zur Abfederung dieser Folgen wurden von staatlicher und privater Seite Hilfsmaßnahmen ergriffen. Die Maßnahmen für Unternehmen umfassen Zuschüsse (z. B. Fixkostenzuschuss und Kurzarbeit), kurz- und langfristige Stundungen, Kreditgarantien sowie Änderungen im Insolvenzrecht. In der Analyse wurden Hilfsmaßnahmen bis zum 31. August 2020 berücksichtigt.
Unserem Modell zufolge führt die Pandemie zu einem starken Anstieg der Unternehmensinsolvenzen. Ohne Hilfsmaßnahmen würden 2020 im Vergleich zu den Vorjahren sechs Mal so viele Unternehmen insolvent werden. Die Hilfsmaßnahmen können diese Zahl jedoch um zwei Drittel reduzieren. Die mit unserem Modell berechneten Insolvenzraten sind mit einem hohen Ausmaß an Unsicherheit verbunden. Die Stärke des Modells liegt denn auch insbesondere in der Abschätzung der Wirksamkeit der einzelnen Hilfsmaßnahmen. Kreditgarantien scheinen beispielsweise die Maßnahme mit der größten Wirkung zu sein, gefolgt vom Fixkostenzuschuss und von der Kurzarbeit. Auf kurze Sicht ist die Aussetzung der Insolvenzantragspflicht am effizientesten; allerdings wird sich der Effekt dieser Maßnahme 2021 umkehren, wenn die Maßnahme annahmegemäß ausläuft.
Auswirkungen der COVID-19-Pandemie auf die finanzielle Situation der privaten Haushalte in Österreich
Nicolas Albacete, Pirmin Fessler, Fabian Kalleitner, Peter Lindner
In dieser Studie werden die potenziellen Auswirkungen der COVID-19-Pandemie auf die finanzielle Situation der privaten Haushalte in Österreich untersucht. Sie stützt sich auf Daten aus dem Austrian Corona Panel Project der Universität Wien sowie aus dem Household Finance and Consumption Survey des Eurosystems in Österreich.
Im ersten Teil der Studie wird aufgezeigt, dass Einzelpersonen und Haushalte auf sehr unterschiedliche Weise und in unterschiedlichem Ausmaß von der COVID-19-Krise betroffen sind. So sind Haushalte, die relativ wenig Wohnfläche zur Verfügung haben – etwa größere Haushalte mit Kindern, Alleinerzieherhaushalte oder Haushalte in dicht besiedelten Gebieten –, den durch COVID-19 verursachten Einkommensschocks stärker ausgesetzt. Für einkommensschwächere Haushalte stellen Pensionseinkommen und sonstige staatliche Transferleistungen einen wichtigen Puffer gegen potenzielle Auswirkungen der COVID-19-Krise dar, da diese Einkommensquellen von den Folgen der Krise bislang (noch) nicht betroffen waren. Darüber hinaus zeigt sich, dass der Medianhaushalt finanzielle Verluste über einen relativ langen Zeitraum hinweg kompensieren könnte, indem er auf liquide Mittel, wie Ersparnisse, zurückgreift. Es scheint daher geboten, sich mit jenen Haushalten zu befassen, die während der COVID-19-Krise erlittene Verluste nicht wettmachen können, wie etwa Alleinerzieherhaushalte oder von Arbeitslosigkeit betroffene Haushalte.
Der zweite Teil der Studie analysiert die potenziellen Folgen der COVID-19-Pandemie. Die Ergebnisse deuten darauf hin, dass private Haushalte während des ersten Lockdowns im April 2020 Einkommenseinbußen von durchschnittlich 12 % verzeichneten; dieser Prozentsatz würde sich verdoppeln, wenn ein Drittel der von Kurzarbeit betroffenen Arbeitnehmerinnen und Arbeitnehmer ihren Arbeitsplatz verlieren würde. Besonders hohe Einkommensverluste verzeichneten zudem Mieterhaushalte. Auf die Konsumabsichten privater Haushalte hatte sich die COVID-19-Pandemie zunächst negativ ausgewirkt, seither ist die Ausgabenneigung allerdings wieder gestiegen. Beträchtliche Unsicherheiten bleiben dennoch bestehen. Auch die Einstellung zum Sparen ist mit hoher Unsicherheit behaftet; doch fanden sich immerhin schwache Hinweise auf eine im Zeitverlauf zunehmend positive Sparneigung einkommensstarker Haushalte.
Unterstützungsmaßnahmen sollten insbesondere mit Blick auf jene Haushalte gesetzt werden, die sich bereits vor der COVID-19-Pandemie in sozialen, wirtschaftlichen und/oder finanziellen Schwierigkeiten befunden hatten und die während der Pandemie die höchsten Einkommenseinbußen zu verzeichnen hatten. Nur so kann man weiteren finanziellen und sozialen Benachteiligungen vorbeugen.
Auswirkungen der geldpolitischen Reaktion auf die COVID-19-Pandemie: vorläufige Ergebnisse einer Pilotstudie auf Basis österreichischer Einzelbankdaten
Claudia Kwapil, Kilian Rieder
Die geldpolitische Reaktion des Eurosystems auf die wirtschaftlichen Folgen der COVID-19-Pandemie erfolgte rasch und in großem Umfang. Das Politikpaket umfasste sowohl Erweiterungen bestehender unkonventioneller Maßnahmen als auch neue Instrumente, um den außerordentlichen wirtschaftlichen Herausforderungen durch die Pandemie zu begegnen. Diese Pilotstudie analysiert die Auswirkungen eines wichtigen Bausteins des geldpolitischen Rettungspakets – nämlich die gezielten längerfristigen Refinanzierungsgeschäfte (TLTRO) – auf das Kreditangebot österreichischer Banken. Im Frühjahr 2020 wurden in Reaktion auf COVID-19 die Bedingungen der jüngsten Generation an TLTROs (TLTRO III) erheblich gelockert, indem das Volumen ausgeweitet, der Zinssatz gesenkt und die Anforderungen an die Sicherheiten heruntergeschraubt wurden. Wir untersuchen, ob jene Banken, die im Juni 2020 (nach der oben genannten Lockerung) den TLTRO III stärker in Anspruch genommen haben, in den Folgemonaten Juli und August vermehrt Kredite an ihre Kundinnen und Kunden vergeben haben. Dabei stützen wir uns auf österreichische Bankdaten und benutzen eine Instrumentalvariablenstrategie, um den kausalen Zusammenhang zwischen der TLTRO-Mittelaufnahme und dem Kreditangebot der Banken zu untersuchen. Wir finden Hinweise auf einen eindeutig positiven Effekt der TLTRO-Beteiligung auf das Kreditangebot in Österreich. Die geschätzte Elastizität des Kreditangebots liegt je nach berücksichtigtem Zeitraum bzw. je nach berücksichtigten Kreditkategorien zwischen 0,36 und 1,97.
Beispiellose fiskalische Maßnahmen zur Bekämpfung der Auswirkungen der COVID-19-Pandemie in Österreich
Österreichs Fiskalpolitik kommt bei der Abmilderung der wirtschaftlichen Folgen der COVID-19-Pandemie eine wichtige Rolle zu. So federn zum einen automatische Stabilisatoren den Wirtschaftsabschwung teilweise ab und zum anderen tragen neue, proaktive fiskalpolitische Maßnahmen sowohl auf nationaler als auch auf europäischer Ebene zur weiteren Stützung der Wirtschaft bei. Die fiskalpolitischen Maßnahmen, die in Österreich während der Lockdowns im Frühjahr und im November/Dezember 2020 getroffen wurden, zielten insbesondere auf die Sicherstellung eines funktionsfähigen Gesundheitssystems und die Unterstützung von Unternehmen (Fixkostenzuschuss, Umsatzersatz) sowie privaten Haushalten (Kurzarbeit, Härtefallfonds) ab. Die Entschädigung von Unternehmen und Haushalten für Einkommensverluste aufgrund von Eindämmungsmaßnahmen trug dazu bei, Produktionskapazitäten zu retten. Letztere wären verloren gegangen, wenn rentable Unternehmen um ihre Existenz gebracht und Arbeitsplätze dauerhaft vernichtet worden wären. Die seit dem Sommer 2020 verabschiedeten Maßnahmen verfolgten zwei Ziele: Erstens hatte der Neustart der Wirtschaft durch klassische Impulse (Einkommenssteuersenkung, Senkung der Mehrwertsteuer für bestimmte Sektoren, einmalige Sozialleistungen) nach der Aufhebung der Lockdowns oberste Priorität. Die damit verbundenen Maßnahmen sollten die Konsumnachfrage – insbesondere von Haushalten mit Liquiditätsengpässen – ankurbeln. Zweitens wurden Initiativen ergriffen, um private Investitionen (Verlustvortrag 2020, degressive Abschreibung, Investitionsprämie) ebenso wie öffentliche Investitionen (Bundeszuschuss für kommunale Investitionen, Aufstockung der Investitionsbudgets) anzuregen. Idealerweise sollen diese Investitionen auch langfristige Ziele, etwa die Abkehr von fossilen Brennstoffen und die Ökologisierung der Wirtschaft, vorantreiben. Dies wiederum erleichtert den Übergang zu neuen Technologien und Arbeitsweisen, stellt die heimische Wirtschaft auf eine tragfähige Basis und steigert damit das langfristige Wachstumspotenzial. Angesichts der weiterhin höchst unsicheren konjunkturellen Perspektiven könnten diese Maßnahmen jedoch weniger wirksam als unter normalen Voraussetzungen sein. So könnten private Haushalte und Unternehmen einfach eine abwartende Haltung einnehmen, anstatt zu konsumieren und zu investieren. Auch sollten die Maßnahmen mit Bedacht zurückgenommen werden, um etwaige Nachwirkungen der COVID-19-Krise, die die wirtschaftliche Erholung hemmen könnten, zu vermeiden. Die mit den neuen fiskalischen Maßnahmen und den automatischen Stabilisatoren verbundenen Kosten bleiben nicht ohne Folgen für das österreichische Budget. So wird Österreich 2020 voraussichtlich das höchste Budgetdefizit seit 1995 verzeichnen. Dies sollte die Tragfähigkeit der öffentlichen Finanzen dennoch nicht gefährden, da Österreich zu Beginn der COVID-19-Pandemie gesunde öffentliche Finanzen aufwies. Da jedoch das derzeit niedrige Zinsniveau nicht von Dauer sein könnte, sollte der hohe öffentliche Schuldenstand auf sozial verträgliche und ökologisch nachhaltige Weise reduziert werden.
Monitoring the economy in real time with the weekly OeNB GDP indicator:
background, experience and outlook
Gerhard Fenz, Helmut Stix
Referee: Philipp Wegmüller, State Secretariat for Economic Affairs (SECO), Switzerland
This study presents the OeNB’s new weekly indicator of economic activity, which is based on a demand-side approach to measuring GDP and which relies on real-time data. The weekly OeNB GDP indicator (1) tracks economic development in Austria on a weekly basis; (2) provides estimates of the contributions of the main demand components of GDP; (3) focuses on seasonally adjusted year-on-year changes; and (4) considers shifts from cash to noncash consumer spending, thus taking into account behavioral changes in the use of payment instruments.
The OeNB has published weekly GDP estimates since early May 2020 and has thus provided policymakers and the public with important and timely information on the state of the Austrian economy. First benchmarking results indicate that the weekly OeNB GDP indicator generated rather accurate results for aggregate economic activity in the first two quarters after the outbreak of the COVID-19 pandemic in Austria.
We describe the construction and the main features of the weekly OeNB GDP indicator, present its results for the period from March to December 2020, discuss the strengths and shortcomings of our approach and draw some lessons from more than eight months of weekly nowcasting with real-time data.
Indicator updates will continue to be released during the COVID-19 pandemic at https://www.oenb.at/Publikationen/corona/bip-indikator-der-oenb.html .
JEL classification: C53; E01; E27
Keywords: GDP, nowcasting, COVID 19, real-time data, payments data
As in most other industrialized countries, the COVID-19 pandemic triggered a deep and abrupt slump in economic activity in Austria. Timely estimates of the economic contraction following the March 2020 lockdown, the subsequent gradual recovery of the Austrian economy and the renewed contraction in November and December 2020 present economic research with substantial challenges. Traditional economic indicators are typically not available on a timely basis given their monthly or quarterly publication schedule. Moreover, the performance of traditional forecasting models might be suboptimal in this special case, as some econometric relationships that are reliable in normal times may have broken down during this severe contraction, e.g. because of sudden behavioral changes and/or nonlinearities.
Against this background, the Oesterreichische Nationalbank (OeNB) developed a weekly economic indicator based on economic data that are measured at high frequency. This indicator estimates real GDP via the expenditure approach. The weekly OeNB GDP indicator (1) tracks economic developments in real time 3 ; (2) provides estimates of the contributions of the main demand components of GDP; (3) looks at seasonally adjusted year-on-year changes; and (4) incorporates behavioral shifts as e.g. its consumption estimate encompasses cash and noncash expenditure and thus takes account of the surge in the use of payment cards during the COVID-19 pandemic. As such, the indicator accounts for some major points of criticism that have been raised against real-time economic indicators. 4 Overall, the weekly OeNB GDP indicator generated accurate results for aggregate economic activity in the first two quarters after the outbreak of the COVID-19 pandemic in Austria while traditional nowcasting models performed rather poorly. Since the weekly OeNB GDP indicator is available on a weekly basis, it provides policymakers and the public with important and timely information on the state of the Austrian economy, which is particularly important given the rapid changes in economic activity caused by renewed lockdowns, stay-at-home orders or travel restrictions.
The aim of this study is to describe the construction and main features of the weekly OeNB GDP indicator. Also, we present its results for the period from March to December 2020 and discuss some early tests on its validity. The weekly OeNB GDP indicator aligns with a series of international economic indicators based on real-time data. We will therefore also put it in an international perspective and (briefly) compare its main features with those of other approaches. Finally, we discuss the lessons that we draw from more than nine months of nowcasting using high-frequency data, and in particular the strengths, limitations and potential of this approach. 5
Before we proceed, we would like to point out that the weekly OeNB GDP indicator is based on an experimental approach and represents a “living project,” i.e. we continuously work on improving it and implementing additional data. This means that results may be revised, also retrospectively.
1 Austrian GDP growth during the COVID-19 pandemic
Chart 1 presents the key results of the weekly OeNB GDP indicator for the period from March to December 2020. 6 The red line shows the change in Austrian real GDP per week against the comparable week of 2019. The value of –5.1 recorded in calendar week (CW) 34 which started on August 17, for example, indicates that real GDP in this week in 2020 was 5.1% below real GDP in calendar week 34 of 2019. 7
The first lockdown in Austria of March 2020 led to a sudden and steep slump in economic activity – at rates not seen in Austria since the World War II. 8 Our estimates suggest that two weeks after the lockdown was in full effect (CW 14), Austrian GDP was 26% below 2019 values. 9 This gloomy state continued until shops were reopened (smaller shops in CW 16 and larger shops at the end of CW 18). The reopening of restaurants and hotels also supported economic recovery. However, GDP growth remained negative over the summer of 2020 (July began in CW 27) and in early fall.
New restrictions were imposed in response to the renewed surge in COVID-19 cases in October 2020. First, Austria issued travel warnings for other countries. Then, other countries issued travel warnings for Austria. In early November 2020, a partial lockdown was imposed in Austria, essentially shutting down restaurants, hotels, cinemas, fitness studios, etc. These measures were tightened on November 17, 2020, with nonessential retail shops and schools being temporarily closed (second lockdown). On December 7, 2020, retail shops and personal service shops were allowed to reopen.
While short-term economic developments are discussed regularly in the reports published on the OeNB’s website www.oenb.at/Publikationen/corona/bip-indikator-der-oenb.html , in this study, we focus on the broader results that emerge from the OeNB GDP indicator estimates:
- In the early stages of the COVID-19 crisis, consumption, nontourism exports and tourism exports contributed most to the decline in Austria’s economic activity. This changed over time, as nontourism exports have returned toward normal levels. For the second lockdown period of November 2020, we find that consumption and tourism exports drove the economic downturn.
Our estimates show that consumption remained subdued even after the containment measures were lifted in May 2020. This development reflects elevated unemployment rates, a partial continuation of short-term work at lower incomes, increased economic uncertainty and precautionary savings as well as, possibly,
a certain extent of spending restraint motivated by fears of contracting the coronavirus.
- Exports also remained subpar for an extended period of time; this shows how important international developments are for a small open economy. At the beginning of October (CW 40), growth rates in non-tourism exports started to turn positive, which was a positive signal in an overall gloomy economic environment.
- Tourism exports, which contribute around 7.5% to GDP in Austria, almost completely collapsed during the first and second lockdowns and gradually recovered over the summer months of 2020, mainly by virtue of domestic and German tourists (Fenz et al., 2020).
- While recovery was fast after the first lockdown in March 2020, over the summer and fall the slope of the recovery increasingly took the shape of a check mark with an increasingly flatter right arm, with real GDP levels ranging between 2% and 5% below 2019 levels.
- The second lockdown has caused a renewed decline in economic activity in Austria. However, the drop is less sharp than in spring, mainly because production and exports have remained largely unaffected.
- Altogether, COVID-19-related GDP losses in Austria (measured as the difference to 2019 GDP levels) are substantial. During the first lockdown, losses amounted to about EUR 2 billion per week. Over the weeks of fall (before the second lockdown), losses came to about EUR 0.5 billion. The fall lockdown led to a renewed increase in weekly GDP losses to about EUR 1 billion. Overall, GDP losses accumulated to EUR 27 billion from March 16 to December 13, 2020. If we also consider the level of GDP that was forecast before the outbreak of the COVID-19 pandemic, losses amount to EUR 31 billion or 7.8% of 2019 GDP.
2 Methodological background
In this section, we discuss why the COVID-19 crisis required the development of new economic activity indicators. Then, we discuss our approach of measuring the demand components of GDP with real-time data.
2.1 The case for new real-time indicators
What forecasters usually like to do, especially in the course of nowcasting exercises, is the following: feed the data into the model, run the model, take the results at face value (after some cross-checking) and – this last point is typically less popular – write a forecast report. As the models are typically highly sophisticated and well validated, this procedure usually leads to results with high nowcasting accuracy (in the sense that they only deviate slightly from final GDP data).
While this well-established and well-tested procedure works well in normal times, it tends to fail in times of severe crises. Most models (factor models, time series models, bridge equations, vector autoregressive models, etc.) are estimated on the basis of historical data and are validated in a pseudo out-of-sample way, with the most recent data being used for validation but not for estimation. This approach to modeling may be quite appropriate as long as there are no structural breaks and the assumption can be maintained that the data follow the same stochastic process during the entire sample period. The problem is that the data sample typically contains no, or at best only a few, episodes of severe economic crises. Furthermore, each crisis has its unique features. Therefore, autoregressive terms, which tend to increase the forecast accuracy in nowcasting models in normal times, can give rise to substantial forecast errors in crises times. Moreover, nonlinearities may not matter much in normal times but can be crucial in crisis times, e.g. if economic agents change their behavior in response to unprecedented events.
Another, and even more important, weakness of traditional approaches to nowcasting is their low time frequency. The typical target variable is GDP, data on which are available only on a quarterly basis, and many short-term economic indicators used in traditional nowcasting models are published with a considerable time lag. The flash estimate for GDP is available 30 days after the respective quarter; the publication lag of foreign trade variables or industrial production is up to two months. If economic activity plunges within a few days or weeks, quarterly models cannot meet policymakers’ high demand for timely information.
Thus, the extraordinary circumstances of the COVID-19 pandemic generated an urgent need for short-term indicators that meet the following requirements:
- They are measured at high frequency (daily or weekly) and are available without much of a delay (almost in “real-time”);
- They are not prone to behavioral changes, not biased by fiscal or monetary policy measures or other measures taken to contain the crisis;
They exhibit a direct (linear) relationship to one of the main components of
- They are available for a period of time long enough to account for seasonal patterns and to apply standard econometric tools.
The availability of data and their characteristics determine the nowcasting method that can be applied. If researchers observe enough indicators for a sufficiently long time period, time series approaches like principal components analysis can be applied and are the prime choice (e.g. for the Weekly Economic Index by the Federal Reserve Bank of New York or Aprigliano et al. 2019). If – as in our case – most of the real-time indicators are only available for a short time period, a more data-driven approach seems appropriate.
Table A1 lists – without any claim to completeness – a set of possible economic indicators that are available for Austria. Our real-time data set of weekly or daily indicators comprises detailed information on labor market developments at a regional and a sectoral level and broken down by socioeconomic characteristics; daily mobility data – for Austria as a whole and for the individual provinces; data on freight volumes, at a detailed regional level, and flight data; weekly debit and credit card transaction data according to country of issue and/or use and spending category; information on cash shipments, ATM withdrawals and bank transfers; various financial market data; information on electricity consumption, air pollution and internet activity.
Not all of these indicators fulfill all the requirements listed above. For example, data on financial markets are currently biased by fiscal or monetary policy decisions; air pollution data are strongly influenced by idiosyncratic events like wind and weather conditions, which are difficult to control for; the higher number of people working from home affects internet activity data, etc. We therefore do not use these data in estimating weekly economic activity.
Beyond that, and this is of particular importance, many of the available real-time indicators cover only a short period of time – often less than two years. This limits the possibilities of applying standard econometric tools. Therefore, we pursue a more “data-driven” approach, for which we use only a few particularly informative real-time indicators. As most of these indicators are directly linked to one of the main demand components of GDP, we estimate weekly economic activity via the expenditure approach.
Another obstacle when using daily or weekly data is seasonal adjustment. The standard statistical tools currently do not support the seasonal adjustment of daily or weekly data, although new procedures are being developed (Ollech, 2018). However, these new procedures require sufficiently long time series. Given that the available indicators are only observed for a short period of time, we have to seasonally adjust the data “by hand.”
Specifically, care is required in choosing the appropriate reference week of the previous year when calculating year-on-year growth rates, i.e. for moving holidays, beginning-of-the-month effects, etc. In
this sense, seasonal adjustment is truly “hand-made” and involves considerable judgment.
In the next sections, we present detailed information on how we estimate weekly growth in private consumption and exports as these are the two most important demand components. In contrast, we will only briefly touch upon the other demand components and the aggregation of all subcomponents to overall GDP.
Private consumption is, next to exports, the single most important expenditure-side component of Austrian GDP (accounting for a share of 51%, see table 1).
As in many other international approaches, our nowcasting estimate of private consumption rests on measuring consumer spending via payment card expenditure (e.g. Andersen et al., 2020; Aprigliano et al. 2019; Baker et al., 2020; Bounie et al., 2020; Brown et al., 2020; Chetty et al., 2020; Carvalho et al., 2020; González Mínguez et al., 2020; INSEE, 2020; Kraenzlin et al., 2020). Given
the importance of cash in Austria, we also account for a broad estimate of cash expenditure (see box 1). 11
The sum of (estimated) cash and (measured) payment card spending by Austrian residents in Austria comprises about 55% of consumer spending, as derived from national accounts data. About 25% of private consumption in Austria refer to expenditure for housing and insurance. Travel expenses abroad account for almost 5% of total consumer spending. Our estimate of Austrian private consumption relies on the 55% of “discretionary” (cash and payment card) consumer spending we observe on a weekly basis. We assume that consumption expenditure for housing and insurance remains constant relative to the previous year. Travel expenses abroad (i.e. tourism imports) are estimated on the basis of payment card information (see section 2.4 for more details). For the remaining share of consumer expenditure (about 15%), for which we do not have any proxy variable, we assume a growth rate similar to the one observed for the above-mentioned 55% of “discretionary” spending. Overall, once we have an estimate of weekly cash and card transactions, the estimate of private consumption is obtained from a simple summation and extrapolation. The essential task is to estimate the weekly value of cash and card transactions.
Estimating “discretionary” weekly consumer spending requires information on the following components:
- the value of domestic face-to-face debit and credit card spending of Austrian residents;
- the value of domestic cash spending of Austrian residents;
- the value of remote (online) transactions of Austrian residents; These transactions can be conducted via credit or debit card, by online transfers via online banking accounts, by ordinary bank transfers, by cash, by gift cards, by mobile phone bills, etc.
As regards domestic face-to-face payment card transactions, we have data on close to 100% of the weekly value of spending via debit cards issued by Austrian banks. Also, we receive data from several credit card-issuing banks in Austria that, taken together, have a dominant market share. We use information on market shares to compute projections for overall credit card spending.
As the weekly OeNB GDP indicator rests on a year-on-year comparison, we could derive annual changes in consumption only from annual changes in payment card spending if the payment cards-cash ratio remained constant. However, in Austria – like in many other countries – the COVID-19 pandemic has caused behavioral changes in the use of payment instruments (see box 1), which are motivated, inter alia, by the fear of contagion, by merchants promoting the use of payment cards or by changes in consumption baskets. Neglecting changes in the use of cash would result in a biased consumption estimate.
The main problem in measuring weekly cash consumption is that it is unobserved and can only be estimated indirectly, e.g. via the value of cash shipments or cash withdrawals at ATMs. If merchants or banks receive cash, it will be shipped to cash-handling companies or to the OeNB. As the organization of the cash cycle in Austria is rather centralized, it takes a relatively short period of time for a banknote to be shipped back to the OeNB (Schautzer and Stix, 2019). For this institutional peculiarity, our estimate relies heavily on data on the weekly return flow of cash to the OeNB. 12
This means that we make the somewhat heroic assumption of a velocity of one, meaning that each banknote is only used once before it is returned to the OeNB, when estimating the absolute value of cash transactions in Austria per week (e.g. to derive the percentage of overall private consumption paid in cash). To assess year-on-year changes in consumption, which is required for estimating the weekly OeNB GDP indicator, we must make the somewhat weaker assumption of a constant velocity. Actual velocity will be somewhat higher than one, e.g. because automated cash recycling machines can check banknote fitness and put banknotes into recirculation without them being delivered back to the OeNB or because merchants directly use their cash receipts for consumption. On the one hand, we will thus underestimate cash consumption. On the other hand, we will overestimate it because cash shipments to the OeNB comprise banknotes that were not used for consumption. This is the case, for instance, when a person receives a cash payment and deposits the respective amount with a bank and the bank ships back this amount of cash to the OeNB or when people reduce their hoarded stocks of cash. Despite these biases, we presume that cash deliveries to the OeNB are highly correlated with actual cash consumption in Austria. 13
It is not possible to provide a comprehensive test for these assumptions. Some degree of validation can be obtained by comparing data on ATM withdrawals, which can be considered a close proxy for cash consumption, with our measure of cash shipments. 14 Overall, the correlation of the value of weekly cash shipments and of ATM withdrawals has been very high in Austria since March 2020, with a correlation coefficient of above 0.9. Furthermore, we compute the implied share of cash from the total of cash, debit card and credit card expenditure, as shown in box 1. The implied share of cash obtained for the time before the COVID-19 pandemic is similar to the respective share of cash obtained in the payment diary study conducted in Austria in 2019 (European Central Bank, 2020; see box 1). Overall, these two checks suggest that our cash shipment measure provides for a reasonably appropriate measure of weekly cash spending in Austria. 15
An alternative to using information on banknote (return) shipments to the OeNB would be to use information on banknote shipments from the OeNB or to use data on ATM withdrawals. We consider banknote shipments from the OeNB less appropriate as these also comprise cash held for hoarding. This was of particular importance in the early weeks of the COVID-19 crisis when cash withdrawals soared (such a surge was also observed in other countries, see Deutsche Bundesbank, 2020; Goodhart and Ashworth, 2020). Data on the value of ATM withdrawals would be more promising but these are not available on a weekly basis. Moreover, hoarding could have similarly influenced ATM withdrawals in the early days of the COVID-19 crisis. 16 Cash shipments to the OeNB, in turn, which we consider, predominantly reflect cash spending (with some contribution from nonconsumptive purposes, e.g. dissolved hoarding stocks).
The final ingredient needed for our consumption estimate is the value of remote (online) transactions. We use data from debit and credit card issuers as well as from providers of secure transfers via online banking accounts. Unpublished survey information shows that these means of settling online transactions comprise the major payment methods for online transactions. However, we have no information on the market shares of the various payment instruments used for online purchases to compute projections for the entire online payment market. An additional problem arises as not all online payments can be assigned to domestic and foreign purchases. Given this situation, we take a pragmatic approach and just record an unadjusted sum of remote transactions. This should nevertheless provide a reasonable estimate of year-on-year changes in online spending as long as there are no large changes in the market shares of the various payment instruments and as long as there is no large shift between domestic and foreign retailers. Moreover, these data limitations are not overly important – from a quantitative perspective – for our consumption estimate, as remote transactions still account for only a modest share in overall private consumption.
Cash use first declined but then recovered in Austria during the COVID-19 pandemic
Austria has a high cash intensity. According to the most recent payment diary study conducted in 2019 (European Central Bank, 2020), cash payments in Austria accounted for about 58% of the value of consumer purchases excluding recurrent payments, while card payments accounted for 28% and other payment instruments for 13%. Among payment cards, debit cards are most frequently used in Austria. European Central Bank (2020) does not report separate shares for debit and credit cards, but if we draw on the results of an earlier study on the situation in Austria, debit cards can be assumed to have a share of around 21% and credit cards a share of around 7% in consumer spending (Rusu and Stix, 2016). If we rebase the shares reported in European Central Bank (2020) and only consider transactions conducted using cash, debit cards and credit cards, cash transactions in Austria have a value share of 67%.
Chart B1 shows our estimate of the implicit shares of cash in point-of-sale (face-to-face) payment transactions before and during the COVID-19 pandemic in Austria. Although these estimates rely on strong assumptions and should therefore be treated as approximations only (see the discussion in the text), the implicit pre-COVID-19 shares are broadly comparable with the results obtained from the above-mentioned payment diary survey studies.
The share of cash in payment transactions declined strongly after the spring 2020 lockdown measures were imposed in Austria, from around 65% to slightly below 50% – which means that EUR 5 out of EUR 10 spent in domestic face-to-face transactions were paid in cash. The increase in card spending was driven by debit cards, and by contactless debit card payments in particular. For the latter, the limit for payments not requiring a PIN has been raised from EUR 25 to EUR 50. Credit cards temporarily lost ground in Austria during the March 2020 lockdown, given that a significant share of credit card spending is related to travel-related payments (Rusu and Stix, 2016), and recovered after the lockdown. Until the end of summer, cash use recovered to a share of about 55%, on average. In the most recent weeks, with the new lockdown in place in Austria since early November 2020, the share of cash payments in total payment transactions has remained roughly constant (if we abstract from short-run fluctuations).
Studies on the use of different payment instruments have shown that consumer behavior tends to change only slowly over time. Bearing this in mind, the swift change in cash use observed during the COVID-19 pandemic is indeed remarkable. However, the fact that cash use recovered also shows that some consumers have slowly returned to their pre-crisis payment behavior and/or that consumption behavior has returned to its pre-crisis state. The results of an OeNB survey conducted in the summer of 2020 indicate that the greatest drop in the use of cash, on average, can be observed for consumers who previously used cash a lot – mainly older persons, persons with lower incomes or persons who tended to not use digital banking or payment products. This survey also shows that 30% of the Austrian population (aged 14 years or older) were concerned about the possible transmission of the coronavirus via banknotes. 64% said they were not concerned and 6% answered that they did not know. Survey results are summarized in German at https://www.oenb.at/Presse/thema-im-fokus/bargeldnutzung-in-oesterreich.html (September 25, 2020).
Overall, our estimate of weekly private consumption expenditure shows the strongest decline at the beginning of the first lockdown in spring 2020 (–35% in CW 13 compared to the same week of 2019) and indicates that consumption remained subdued even after the restrictive measures were lifted. The second lockdown, which started in CW 47, triggered another slump in expenditure which so far has remained less significant than during the first lockdown. Given its large share in GDP, weekly consumption significantly shapes the overall GDP growth pattern.
2.3 Goods exports
Business activity is closely related to freight performance. The high correlation between freight growth and economic growth has been emphasized in numerous international studies (e.g. OECD, 2004; Fenz and Schneider, 2009), with the linkage being particularly evident in small open economies.
For Austria, weekly information on truck mileage and rail transport is available on a real-time basis. 17 Weekly data on air traffic (passengers and freight volume) are provided for analytical purposes but for reasons of confidentiality only monthly data are approved for publication. Among all means of transportation, truck mileage shows the closest relationship to export activity (see chart 3).
Fenz and Schneider (2009) document the good leading indicator properties of truck mileage data for goods exports in Austria. On the basis of their results, they developed the OeNB’s Export Indicator – a monthly indicator of export performance published regularly (in German only) on the OeNB’s website. 18
To calculate the weekly OeNB GDP indicator, we update the estimations presented in Fenz and Schneider (2009) to determine the growth contributions of real exports of goods and services excluding tourism. The complementary relationship between goods and services exports is an empirically well-supported fact (Ariu et al. (2020) and Walter (2017) for Austria). Tourism exports are analyzed separately here (see section 2.4) while for exports of other services we make the simplifying assumption that they are closely linked to goods exports.
For our estimation, we aggregate truck mileage data on a quarterly basis and use this variable as the only explanatory variable in a simple regression for real exports of goods and services excluding tourism according to the national accounts. Both variables are seasonally adjusted. We refrain from using autoregressive terms, which would increase the overall fit of the equation but would worsen the nowcasting and forecasting performance during crises, as experience from past crises has shown. The estimated coefficient of truck mileage is 1.18 and it is highly significant. To nowcast weekly export activity, we assume that these estimation results at the quarterly level also hold at the weekly level. 19 Alternatively, we could have estimated an unobserved component model in state space form but this is left to future work.
2.4 Tourism exports
We analyze the tourism component of Austrian exports (and imports) separately for several reasons. First, tourism has been hit particularly hard by the COVID-19 crisis. Moreover, tourism exports account for more than 7% of total value added in Austria, which is well above the EU average, and their import share (19%)
is significantly lower than that of total exports (45%). Finally, tourism services
are obviously less closely linked to transport activity than goods exports and nowcasting them should therefore be based on other indicators.
We use data provided by payment card service providers to separately estimate tourism expenditure by foreigners in Austria (tourism exports) and by domestic residents abroad (i.e. tourism imports, which are also part of private consumption expenditure). Information on the respective expenses has been available on a weekly basis since the beginning of 2019. Payment card data are broken down
by country of origin and several categories of goods and services. We use data
on cardholders’ expenses on typical tourist activities such as overnight stays, restaurants or traveling as a proxy for their overall tourism expenditure. Data are adjusted for moving holidays and inflation developments. Year-on-year changes are used to calculate the respective contributions to total GDP growth.
For a more detailed discussion of developments in Austrian tourism during the COVID-19 pandemic, see Fenz et al. (2020).
2.5 Other GDP components
The remaining demand components of GDP include investment activity (construction and nonconstruction investment), government consumption and changes in inventories (including the statistical discrepancy).
The development of construction investment is estimated using weekly data on the number of registered unemployed persons in the construction sector. When using weekly labor market data, the effects of short-term work schemes during the COVID-19 pandemic has to be taken into account. We do this on a judgmental basis since timely information is available only on the number of applications for short-term work schemes per economic sector but not on the actual utilization of these schemes – typically, actual utilization is substantially lower than the number of applications. Information about the actual utilization only becomes available with a considerable time lag. Other investment (nonconstruction investment) includes equipment investment and investment in R&D. 20 Since no suitable real-time indicator is available, we make the assumption that the weekly pattern of other investment follows the weighted average of the other demand components. 21 Public consumption is assumed to grow constantly at an annual growth rate of 1.5%. 22 Finally, the growth contributions to GDP by inventory changes (including the statistical discrepancy) are assumed to be zero for all weeks considered.
2.6 Putting all subcomponents together
To infer weekly GDP growth from the estimated weekly demand components, two more steps are needed. First, we adjust each demand component for its import content according to the latest input-output table for Austria. Import contents vary considerably from 11% for public consumption to 45% for exports. Specific subcomponents like investment in vehicles even reach import content shares of more than 80% (Fenz and Schneider, 2019). Second, each demand component is weighted with its share in GDP to derive the import-adjusted GDP shares shown in table 1. The sum of the import-adjusted demand components corresponds to total GDP.
|Share in GDP||Share in imports||
share in GDP
|of which: construction||11||22||8|
|of which: tourism exports||5||19||4|
|of which: tourism imports||3||x||x|
|Source: Statistics Austria, authors’ calculations.|
The import-adjusted GDP shares of the demand components we model in detail – private consumption, exports and construction investment – account for more than 75% of GDP. With an import-adjusted share of 37%, private consumption is the single most important GDP component. Possible changes in the import shares of the main demand components induced by the COVID-19 pandemic are taken into account at least partly by explicitly modeling tourism exports characterized by a below-average import share and tourism imports with an import share of 100%. 23
2.7 How does the weekly OeNB GDP indicator compare internationally?
Over the past months, a plethora of real-time indicators has been developed and analyzed. These indicators refer to consumption, industrial production, exports, economic sentiment and overall economic activity. These indicators have greatly contributed to an understanding of how the economy and specific economic sectors have evolved in response to the COVID-19 shock (see e.g. Indergand, Kemeny and Wegmüller, 2020).
As the weekly OeNB GDP indicator focuses on overall economic activity, we briefly put it into perspective with other real-time indicators that focus on GDP. Specifically, we focus on selected, publicly available indices and neglect proprietary sources.
The Weekly Economic Index (WEI) of the Federal Reserve Bank of New York measures real economic activity on a weekly basis (Lewis, Mertens and Stock, 2020a, 2020b). The WEI is based on a principal component analysis of ten high-frequency series, which is scaled to annual GDP growth. As mentioned above, such approaches have the advantage that they can provide nowcasts for a longer time period – back until 2008 in the case of the WEI. This makes it possible to conduct robustness tests that enhance the credibility of such indices. The downside of this principal component approach is that it does not provide information on the subcomponents of GDP. The WEI displays a sharp recession in the USA that reaches its lowest value in a –11.5% drop in real GDP (as of end-April 2020).
The nowcasts of the French National Institute of Statistics and Economic Studies (e.g. INSEE, 2020) are based on detailed and comprehensive assessments of the subcomponents of French GDP. This approach is based on the production side of GDP as well as on a broad range of real-time indicators and provides a disaggregated sectoral analysis. The respective nowcasting report is updated (at irregular intervals) about every second month, and GDP estimates refer to a monthly period. The deepest slump was recorded in April 2020 at about –30%, and the shape of the recovery is very similar to that observed in Austria: first, a strong recovery, and then a prolonged period featuring subpar GDP levels. We note that the estimate by INSEE (2020) of the size of the decline in household consumption as well as in GDP during the first lockdown in spring 2020 is rather similar to our estimate.
There are other informative and interesting nowcasting indicators of GDP that are based on time series methods and the extraction of a trend component, e.g. GDPNow of the Federal Reserve Bank of Atlanta (Higgins, 2014) and the Weekly activity index for the German economy (WAI) published by the Deutsche Bundesbank (Eraslan and Götz, 2020). GDPNow is a “running estimate” of real GDP growth in a specific quarter. It uses newly available data to update the forecasts
of the current quarter’s GDP growth. GDPNow provides information on the subcomponents of GDP. The WAI is based on a principal components analysis of high-frequency indicators, including pedestrian activity, Google search terms, etc., and presents changes over 13-week averages. The WAI does not reveal information on the state of the economy in the most recent weeks and is thus not directly comparable to the other indices. Finally, we note that the Austrian Institute of Economic Research WIFO developed a weekly GDP indicator for Austria based on a time series approach. This indicator has been computed since March 2020 but was published only in October 2020 (Baumgartner et al., 2020). Therefore, we do not discuss this informative indicator in more detail. The GDP growth path of this indicator is rather similar to that of the OeNB’s. In November 2020, the Swiss State Secretariat for Economic Affairs (SECO) released a new weekly indicator that is broadly comparable to the other indicators and that is also based on a time series approach (SECO, 2020).
3 Plausibility and benchmarking checks
3.1 Plausibility checks with alternative real-time indicators
Some real-time indicators that are not directly included in the estimation of the weekly OeNB GDP indicator, such as data on electricity consumption, mobility behavior, short-term work and financial market variables, are used for plausibility checks. The Google mobility index (calculated as the average of the Google subindices “supermarket and pharmacy”, “public transport”, “workplaces”,
“retail and recreation”) and the OeNB’s GDP indicator move almost completely in parallel. Even if the extraordinarily high correlation of 0.95 results from the specific features of the COVID-19 crisis and the government’s containment measures, it nevertheless indicates that these new indicators will play an increasingly important role in economic monitoring in the future. The link to electricity consumption and financial indicators, on the other hand, is recognizable, but much less pronounced.
3.2 First benchmarking results are promising
After individual data announcements, such as overnight stays and production indices, confirmed results for some subcomponents of the weekly OeNB GDP indicator , it successfully passed its first real “elk test” when national accounts (NA) data for the second quarter of 2020 were released. The results of the latest release of NA data (published on September 28, 2020) show that Austrian GDP fell by 14.5% (real, seasonally and working-day adjusted) in the second quarter of 2020 compared to the same period in 2019, which is remarkably close to the estimate provided by the weekly OeNB GDP indicator of 14.4%. Regarding Austrian GDP in the third quarter of 2020, the first release of NA data (published on November 30, 2020) suggests that it was 4.2% lower than in the third quarter of 2019. Our estimate based on the weekly OeNB GDP indicator was a GDP growth rate of –4.4%. 24
|Second quarter of 2020|
OeNB GDP indicator
(July 10, 2020)
New release of NA data (September
First release of NA data
(August 28, 2020)
NA flash estimate
(July 30, 2020)
|Change on comparable quarter of 2019|
|of which: construction||–12.0||–8.1||–9.6||x|
|Third quarter of 2020|
OeNB GDP indicator
(October 10, 2020)
First release of NA data (November
NA flash estimate
(October 30, 2020)
|Change on comparable quarter of 2019|
|of which: construction||–3.9||–1.8||x|
|Source: Statistics Austria, WIFO, OeNB.|
|Note: NA = national accounts, x = data not available.|
Table 2 shows that the estimates of the individual demand components (for the second quarter of 2020) were also quite accurate, albeit less so than the estimate for overall GDP. The deep slump in private consumption expenditure, notably, was predicted well (second quarter of 2020: –14.9% according to the weekly OeNB GDP indicator versus –15.8% according to the NA; third quarter: –4.3 versus –4.7%), which is reassuring given that we had to make many assumptions when constructing our consumption index. Our estimates for exports and investment, by contrast, deviate farther from the preliminary NA figures. In general, we must note, however, that the assessment of the individual demand components is hampered by the significant negative contribution of the statistical discrepancy to GDP growth in the second quarter of 2020 (–2.3 percentage points) in the latest release of NA data. 25 Past experience indicates that this signals future revisions mainly of the investment and foreign trade components.
4 Summary and conclusions
Each crisis has its very specific and unique features, drivers and transmission channels. The COVID-19 crisis and the ensuing containment measures triggered simultaneous supply and demand shocks, had very heterogenous sectoral impacts and caused the economic downturn to proceed at an unprecedented speed. This extraordinary situation generated the need for real-time information on various economic sectors that is typically not provided by traditional nowcasting models or short-term forecasting models.
In response to this situation, we have developed an experimental weekly estimate of economic activity which focuses on seasonally adjusted year-on-year changes. The weekly OeNB GDP indicator, which has been published regularly since May 2020, has provided policymakers and the public with timely and reliable information on the state of the Austrian economy.
Our choice of an estimation approach was governed by the availability and characteristics of real-time indicators for the Austrian economy. As many indicators are directly linked to one specific demand component, we estimate economic activity via the expenditure side of GDP. Moreover, the experimental nature of many indicators and the fact that they cover only a short period of time made the application of traditional econometric methods impossible. We therefore opted for a data-driven approach rather than a more conventional model-based approach. Our approach requires a lot of qualitative assessments and adjustments, e.g. the treatment of moving holidays, working day adjustments or the identification of outliers in cash shipment data. These adjustments often require further analyses, in-depth expert discussions, etc. – all in all, an extra effort that would not be necessary when applying a purely model-based approach. Moreover, our data-driven approach relies on the availability of suitable high-frequency data and on some institutional peculiarities (e.g. with regard to cash logistics) which might limit its applicability to other countries (Matheson, 2013).
By publishing a weekly estimate of economic activity, we have entered new grounds. This always entails some risks. In particular, it was not possible to validate in advance the accuracy of the OeNB GDP indicator. Reassuringly, the indicator turned out to be very accurate in nowcasting aggregate quarterly economic activity for the second and third quarters of 2020 (while the performance of traditional models with regard to these two quarters was rather weak). But more observations are needed for a final assessment, and it remains to be seen whether the new indicator also performs in times when economic activity is closer to normal. The results for some subcomponents of the weekly OeNB GDP indicator, such as exports of goods or construction investment, can be assessed on a monthly basis using foreign trade data or production indices. 26 Some additional validation arises from comparisons with high-frequency indicators from other institutions and for other countries. For example, our estimates of private consumption and of GDP during the first lockdown in spring 2020 are rather similar to those reported for France (INSEE, 2020), where comparable stay-at-home orders and other protective measures had been imposed. Our consumption estimates for the weeks of the first lockdown period are rather close to estimates by Brown et al. (2020) for Switzerland or by Bank of Israel (2020) for Israel. The evolution of our weekly GDP estimates over the post-lockdown period is rather similar to those of the Austrian Institute of Economic Research (Baumgartner et al., 2020), which are based on a time-series approach. While these are (promising) bits and pieces, clearly a more systematic and profound validation analysis will need to be carried out.
The seemingly high accuracy of the weekly OeNB GDP indicator vis-à-vis traditional nowcasting models raises the question whether it should be a regular tool in nowcasting GDP. The answer is “probably not.” In normal times, traditional models have proven to be rather precise for nowcasting and short-term forecasting while high-frequency real-time data might only provide additional explanatory content in times of crises (Delle Chiaie and Perez Quiros, 2020). Thus, the presumably low marginal benefit in normal times needs to be weighed against the cost and effort of collecting and processing the necessary data on a weekly basis as well as carrying out the required qualitative assessments of the results. Clearly, more research will be necessary to assess the corresponding costs and benefits, taking into account the results of further validation analyses. While this is beyond the scope of this paper, our conjecture is that the main benefits from integrating real-time data into the existing model toolkit arise mainly from their contribution in times of larger economic downswings or outright crises.
In general, the economics profession has shown creativity and swiftness in utilizing real-time data to provide urgently needed empirical evidence. Our experience with alternative data on transport activity during the last crisis, the Great Recession of 2008/2009, led to the development of the OeNB’s monthly Export Indicator. During the COVID-19 crisis, it has been mainly real-time data on payment transactions which have created new possibilities in analyzing consumption and tourism activities. We think that these new data will be informative also in normal times, e.g. for policy analyses such as the assessment of the economic impact of fiscal transfers (see Chetty et al., 2020; Baker et al., 2020) or the change in (online) consumption patterns (Brown et al., 2020). In times of steady digital innovation, ever more information will be available, also at the disaggregated level and for small geographical areas. Clearly, this will open up new possibilities to economic modeling and forecasting. Apart from economic expertise, the increasing availability of new data will require new forms of collaboration, e.g. with data scientists. Economic institutions that run models and conduct forecasts will need to adjust to this development.
Andersen, A. L., E. T. Hansen, N. Johannesen and A. Sheridan. 2020. Pandemic, Shutdown and Consumer Spending: Lessons from Scandinavian Policy Responses to COVID-19. https://arxiv.org/pdf/2005.04630.pdf .
Aprigliano, V., G. Ardizzi and L. Monteforte. 2019. Using Payment System Data to Forecast Economic Activity. In: International Journal of Central Banking, October, 55–80.
Ardizzi, G., A. Nobili and G. Rocco. 2020. A Game Changer in Payment Habits: Evidence from Daily Data during a Pandemic. Mimeo.
Ariu, A., Mayneris, F. and M. Parenti. 2020. One way to the top: How services boost the demand for goods. In: Journal of International Economics, Vol. 123(C).
Baker, S. R., R.A. Farrokhnia, S. Meyer, M. Pagel and C. Yannelis, 2020. How Does Household Spending Respond to an Epidemic? Consumption During the 2020 COVID-19 Pandemic. In: NBER Working Papers 26949. https://ideas.repec.org/p/nbr/nberwo/26949.html .
Bank of Israel. 2020. Selected Rapid Indicators of the State of Economic Activity during the Coronavirus Crisis. https://www.boi.org.il/en/Pages/Indicators.aspx .
Baumgartner, J., Kaniovski, S., Bierbaumer-Polly, J., Glocker, C., Huemer, U., Loretz, S., Mahringer, H. and H. Pitlik. 2020. Die Wirtschaftsentwicklung in Österreich im Zeichen der COVID-19-Pandemie Mittelfristige Prognose 2020 bis 2024. In: WIFO Monatsberichte. 93(4), 239–265.
Bounie, D., Y. Camara and J. W. Galbraith. 2020. Consumers’ Mobility, Expenditure and Online-Offline Substitution Response to COVID-19: Evidence from French Transaction Data.
In: CIRANO Working Papers 2020s-28, CIRANO. https://ideas.repec.org/p/cir/cirwor/2020s-28.html .
Brown, M., Fengler, M., Lalive, R. and R. Rohrkemper. 2020. Monitoring consumption in Switzerland. https://monitoringconsumption.com/about/ .
Carvalho, V., Hansen, S., Ortiz, A., García, J. R., Rodrigo, T., Rodríguez Mora, J.V. and P. Ruiz (2020), Tracking the covid-19 crisis with high-resolution transaction data. In: CEPR Discussion Paper No. 14642.
Chetty, R., J. N. Friedman, N. Hendren, M. Stepner and the Opportunity Insights Team. 2020. Real-Time Economics: A New Platform to Track the Impacts of COVID-19 on People, Businesses, and Communities Using Private Sector Data. https://opportunityinsights.org/wp-content/uploads/2020/05/tracker_paper.pdf .
Deutsche Bundesbank. 2020. Cash withdrawals and payments in urban and rural areas.
In: Monthly Report. June 2020. 33-44.
Delle Chiaie, S. and G. Perez Quiros. 2020. Nowcasting economic activity and trade in times of COVID-19: are high frequency data useful? Mimeo ECB 2020.
Eraslan, S. and T. Götz. 2020. Weekly activity index for the German economy. Deutsche Bundesbank, available for download at https://www.bundesbank.de/wai .
European Central Bank. 2020. Study on the payment attitudes of consumers in the euro area. https://www.ecb.europa.eu/pub/pdf/other/ecb.spacereport202012~bb2038bbb6.en.pdf?05ce2c97d994fbcf1c93213ca04347dd .
Fenz, G. and M. Schneider 2019. Weak global trade darkens growth outlook for Austria – Economic outlook for Austria from 2019 to 2022 (December 2019). In: Monetary Policy & the Economy. Q4/2019.
Fenz, G. and M. Schneider 2009. A Leading Indicator of Austrian Exports Based on Truck Mileage. In: Monetary Policy & the Economy. Q1/2009.
Fenz, G., Stix, H. and K. Vondra. 2020. Austrian tourism sector badly hit by COVID-19 pandemic. In: Monetary Policy & the Economy. Q3/2020.
González Mínguez, J., Pérez García de Mirasierra, M. and A. Urtasun. 2020. Consumption in Spain during the state of alert: An analysis based on payment card spending, In: Banco de Espana Economic Bulletin. 3/2020. https://www.bde.es/f/webbde/SES/Secciones/Publicaciones/InformesBoletinesRevistas/ArticulosAnaliticos/20/T3/descargar/Files/be2003-art22e.pdf .
Goodhart, C. and J. Ashworth. 2020. Coronavirus panic fuels a surge in cash demand, Vox-EU, https://voxeu.org/article/coronavirus-panic-fuels-surge-cash-demand .
Higgins. P. 2014. GDPNow: A Model for GDP “Nowcasting”. In: Federal Reserve Bank of Atlanta Working Paper Series 2014-7.
Indergand, R., Kemeny, F. and Philipp Wegmüller. 2020. Konjunkturprognosen in Zeiten von Corona – ein Werkstattbericht. Die Volkswirtschaft. Plattform für Wirtschaftspolitik. https://dievolkswirtschaft.ch/de/2020/04/konjunkturprognosen-in-zeiten-von-corona-ein-werkstattbericht/ .
INSEE. 2020. Economic outlook 2020. https://insee.fr/en/statistiques/4473307 .
Kraenzlin, S., C. Meyer and T. Nellen. 2020. COVID-19 and regional shifts in Swiss retail payments. In: Swiss National Bank Working Paper No. 15/2020.
Lewis, D. J., K. Mertens and J. H. Stock. 2020a. Weekly Economic Index, https://www.newyorkfed.org/research/policy/weekly-economic-index .
Lewis, D. J., K. Mertens and J. H. Stock. 2020b. Monitoring Real Activity in Real Time: The Weekly Economic Index, In: Federal Reserve Bank of New York Liberty Street Economics, March 30.
Matheson, T. D. 2014. New indicators for tracking growth in real time. In: OECD Journal: Journal of Business Cycle Measurement and Analysis 2013/2. https://www.oecd-ilibrary.org/docserver/jbcma-2013-5jz741mh2czs.pdf?expires=1601378615&id=id&accname=id5760&checksum=F12D3E1D656190D0C523D96130B364B4 .
National Bank of Greece. 2020. Tracking Greek GDP in high frequency. In: GREECE Macro Flash, https://www.nbg.gr/en/the-group/press-office/e-spot/reports/greece-macro-flash-tracking-greek-gdp-in-high-frequency , June 2020.
Olech, D. 2018. Seasonal adjustment of daily time series. In: Deutsche Bundesbank Discussion Paper No 41/2018.
Prammer, D. 2020. Fiskalmaßnahmen in Österreich zur Bekämpfung der COVID-19-Krise und ihre Wirkungen. Forthcoming Monetary Policy & the Economy, Q3/2020.
Rusu, C. and H. Stix. 2017. Cash and card payments – recent results of the Austrian payment diary survey, available for download at https://ideas.repec.org/a/onb/oenbmp/y2017iq1-17b4.html . Originally published in German as: Von Bar- und Kartenzahlern – Aktuelle Ergebnisse zur Zahlungsmittelnutzung in Österreich. In: OeNB. Monetary Policy & the Economy Q1/17. 54–85.
Schautzer A. and H. Stix. 2019. Approaching 20 years of euro cash in Austria: What has changed, and what’s next? In: Monetary Policy & the Economy, Q1-Q2/2019. 99–112.
SECO (Swiss State Secretariat for Economic Affairs). 2020. Exkurs: Neuer Index zur wöchentlichen Wirtschaftsaktivität (WWA). https://www.newsd.admin.ch/newsd/message/attachments/63920.pdf .
Walter, P. 2017. Anatomy of Austria’s trade in services. In: Monetary Policy and the Economy Q1/17 p.33–51.
|Target variable||Use||Data source|
|Unemployed||d/w||Sectoral/ regional||< 1 week||< 1 year||
|d/w||Sectoral/regional||< 1 week||< 1 year||
|Vacancies||d/w||Sectoral/regional||< 1 week||< 1 year||
|d/w||Sectoral/regional||< 1 week||< 1 year||
|Short-term work||w||Sectoral/regional||< 1 week||< 1 year||
|d/w||Regional||< 1 week||< 1 year||
|Plausibility check||Google, Apple|
|Truck milage||d/w||Regional||< 1 week||> 5 years||
Exports, investment, general
|Railway||w||Regional||< 1 week||> 5 years||
|Flight data||w||< 1 week||> 5 years||
|Cash||d/w||< 1 week||< 2 years||
|Debit cards||d/w||Sectoral||< 1 week||< 2 years||
|Credit cards||d/w||Sectoral||< 1 week||< 2 years||
|Online transfers||w||< 1 week||< 2 years||Private consumption||Estimation||
|Bank transfers||w||< 1 week||< 2 ears||
|Financial market data|
Stock price indices,
yield curve, CDS,
risk premia, etc.
|d||> 5 years||
|Plausibility check||Various data providers|
|15 min||Sectoral||< 1 week||> 5 years||Industrial sector||Plausibility check||APG, E-Control|
|Gas consumption||d||< 1 week||> 5 years||Industrial sector||Not used|
|Air pollution data||d||Regional||< 1 week||> 5 years||
tweets, tag clouds
|d||< 1 week||> 5 years||
Tax data and/or social
|w/m||< 1 week||> 5 years||
|Webscraping||d||< 1 week||< 1 year||
|Internet activity||d||Regional||< 1 week||< 2 years||
|Note: d = daily; w = weekly; CDS = credit default swaps; AMS = Public Employment Service Austria; APG = Austrian Power Grid.|
|Calendar weeks in 2020|
|CW1||December 30 – January 5|
|CW2||January 6 – January 12|
|CW3||January 13 – January 19|
|CW4||January 20 – January 26|
|CW5||January 27 – February 2|
|CW6||February 3 – February 9|
|CW7||February 10 – February 16|
|CW8||February 17 – February 23|
|CW9||February 24 – March 1|
|CW10||March 2 – March 8|
|CW11||March 9 – March 15|
|CW12||March 16 – March 22|
|CW13||March 23 – March 29|
|CW14||March 30 – April 5|
|CW15||April 6 – April 12|
|CW16||April 13 – April 19|
|CW17||April 20 – April 26|
|CW18||April 27 – May 3|
|CW19||May 4 – May 10|
|CW20||May 11 – May 17|
|CW21||May 18 – May 24|
|CW22||May 25 – May 31|
|CW23||June 1 – June 7|
|CW24||June 8 – June 14|
|CW25||June 15 – June 21|
|CW26||June 22 – June 28|
|CW27||June 29 – July 5|
|CW28||July 6 – July 12|
|CW29||July 13 – July 19|
|CW30||July 20 – July 26|
|CW31||July 27 – August 2|
|CW32||August 3 – August 9|
|CW33||August 10 – August 16|
|CW34||August 17 – August 23|
|CW35||August 24 – August 30|
|CW36||August 31 – September 6|
|CW37||September 7 – September 13|
|CW38||September 14 – September 20|
|CW39||September 21 – September 27|
|CW40||September 28 – October 4|
|CW41||October 5 – October 11|
|CW42||October 12 – October 18|
|CW43||October 19 – October 25|
|CW44||October 26 – November 1|
|CW45||November 2 – November 8|
|CW46||November 9 – November 15|
|CW47||November 16 – November 22|
|CW48||November 23 – November 29|
|CW49||November 30 – December 6|
|CW50||December 7 – December 13|
|CW51||December 14 – December 20|
|CW52||December 21 – December 27|
|CW53||December 28 – January 3|
2 Oesterreichische Nationalbank, Economic Analysis Division, firstname.lastname@example.org, Economic Studies Division, email@example.com. We are indebted to the companies (some of which prefer to remain anonymous) that continuously provide the (anonymized and aggregated) data necessary to construct the weekly OeNB GDP indicator. We are grateful to the referee and to Ernest Gnan, Walpurga Köhler-Töglhofer, Doris Ritzberger-Grünwald, Martin Summer and Thomas Steiner (all OeNB) for very helpful comments, suggestions and support in developing the GDP indicator. Also, we thank Doris Prammer, Anton Schautzer, Martin Schneider, Alfred Stiglbauer and Patrick Thienel (all OeNB) for valuable support regarding data and methods. Opinions expressed by the authors of studies do not necessarily reflect the official viewpoint of the Oesterreichische Nationalbank or the Eurosystem.
3 We will use the term “real-time data” as synonymous with “almost or quasi real-time data,” meaning data that are available at a daily or weekly frequency without great delays in publication. Also, our use of the term real-time data differs from the term used in the context of forecast evaluation in the sense that our real-time data can be subject to revisions.
4 See, for example, “Why real-time economic data need to be treated with caution” (The Economist, July 23, 2020). https://www.economist.com/finance-and-economics/2020/07/23/why-real-time-economic-data-need-to-be-treated-with-caution .
5 The OeNB GDP indicator has been published on a regular basis since mid-May 2020 and its results have been made available on the OeNB’s website. Each publication comprises a data file (both in English and German) and a German summary of the results.
6 Cutoff date for data: December 13, 2020.
7 2019 is particularly suitable as a year of comparison as it was a normal business year with growth rates close to the long-term average and a closed output gap.
8 Containment measures were imposed from March 9, 2020, onward. Monday, March 16, 2020, (calendar week 12) was the first working day when shops remained closed. A timeline is provided at https://en.wikipedia.org/wiki/COVID-19_pandemic_in_Austria (retrieved on September 24, 2020), for example.
9 Table A2 in the annex lists calendar weeks and the corresponding calendar dates.
10 Some seasonal adjustment occurs by focusing on year-on-year changes. Beyond that, seasonalities are mainly introduced by moving holidays and beginning-of-the-month effects.
11 The importance of including cash into estimating consumption is also highlighted in Ardizzi et al. (2020) for Italy and Brown et al. (2020) or Kraenzlin et al. (2020) for Switzerland.
12 We exclude data on cash shipments by international wholesale cash dealers from our estimations.
13 Another subtle issue arises as cash shipments also comprise cash spending by tourists and cash exported by domestic residents when traveling abroad. We have attempted to estimate these components (as best we can, given available data and using ad hoc assumptions) and find that their quantitative importance is not large relative to the overall amount of cash that is shipped from or to the OeNB.
14 Unfortunately, we cannot use information on ATM withdrawals directly because we observe only a relatively small share of all ATM withdrawals in Austria.
15 Moreover, our interpretation is continuously vetted and discussed with experts in the OeNB’s Cashier’s Division and adjusted if necessary. For example, our year-on-year comparison of banknote shipments is biased in certain weeks as the issuance of the new EUR 100 banknote series in 2019 resulted in an above-average return flow of old EUR 100 banknotes. In such cases, we adjusted return flow data on a judgmental basis, utilizing data on changes in ATM withdrawals.
16 Hoarding behavior could, to a small extent, also exist for ATM withdrawals, e.g. around pay-day, when people replenish their cash reserves and store cash for later purchases (Brown et al., 2020b).
17 Road transport is by far the single most important means of freight transport. In 2018, road transport accounted for 75% of total freight performance in Austria when measured in tons and 56% when measured in million ton-kilometers. Rail transport is second-most important, accounting for a share of 14% and 24%, respectively. Air transportation (<1%) and transportation via waterways (1%) are much less important in Austria. Pipelines account for the remainder of the overall transportation volume (10% and 19%, respectively).
19 To refine our estimations further, we use regional truck mileage data. From the beginning of 2019, the Austrian highway authority ASFINAG has provided detailed information on the border sections of the Austrian highway system. These data should be even more closely linked to export activity. Our weekly estimates are adjusted for the differences between the growth rate of truck mileage on the whole highway network and the growth rate of truck mileage in the border sections.
20 Other investment (nonconstruction investment) comprises equipment investment, at a share of 60%, and investment in R&D, at a share of 40%. Equipment investment is very sensitive to the business cycle and is characterized by a high import share of almost 70%, while investment in R&D is less sensitive to the business cycle and is characterized by a relatively low import share of 20%.
21 This approach led to reasonable results for the weeks until the end of the second quarter of 2020. From the beginning of the third quarter of 2020, the estimated recovery in “other investment” has been assessed to be too positive, given the steep rise in corporate debt. The latter should cause investment activity to be dampened more strongly than overall economic activity. We therefore adjusted the weekly pattern of “other investment” activity judgmentally, in line with the OeNB’s June 2020 forecast.
22 This assumption follows the assessment of the OeNB’s fiscal experts in their biannual macroeconomic projection exercise.
23 Tourism imports are modeled not only as a subcomponent of imports but also as a subcomponent of private consumption. Changes in the consumption of tourism services abroad therefore have no direct effect on overall GDP as their import share amounts to 100%.
24 We compute quarterly growth rates by taking the average of the weekly growth rates of a given quarter, adjusting for endpoints if calendar weeks overlap with months.
25 Growth contributions of the statistical discrepancy are assumed to be zero in the weekly OeNB GDP indicator. Moreover, the latest NA data also show that changes in inventories made a significant negative contribution to GDP growth (–1.3%). Changes in inventories are not explicitly modeled in the weekly OeNB GDP indicator and are assumed to be growth neutral.
26 At a conceptual level, it will never be possible to validate the weekly estimates as GDP is measured only at a quarterly frequency. However, further validation tests are possible for some economic indicators that are published monthly.
Austrian tourism sector badly hit by COVID-19 pandemic 27
Gerhard Fenz, Helmut Stix, Klaus Vondra
Referee: Oliver Fritz, WIFO
Contributing 7.3% to Austrian value added, tourism is an important pillar of the Austrian economy. It has been hit particularly hard by the COVID-19 crisis. We analyze the impact of the crisis using high-frequency real-time data on payment card spending and monthly data on overnight stays. During the lockdown in spring 2020, overnight stays in Austria dropped by almost 100%. Over the summer, tourism activity recovered strongly, backed by domestic and German tourists. Nevertheless, it remained clearly below 2019 levels. In October 2020, the renewed increase in the number of COVID-19 infections led to another severe downturn in Austrian tourism, as several neighboring countries posted travel warnings. On November 2, 2020, a second lockdown started in Austria – accommodation establishments and restaurants were closed. Hence, we expect overnight stays to drop again by around 95% in November. As the Austrian government announced on December 2, 2020, Austrian accommodation establishments will not open before January 2021; on top of that, travel warnings by major countries of origin (especially Germany) will remain in place. Based on these assumptions, we estimate total overnight stays to decrease by 36% in 2020. This will be mainly attributable to a strong decline in overnight stays by foreign tourists (–41%), while overnight stays by domestic tourists will go down by only 23%. The overall decline in overnight stays could have been far stronger if the lockdown in spring 2020 and the recent shutdown had not fallen into the off-season but into the high season in winter or summer.
JEL classification: E23, L83
Keywords: tourism, COVID-19 pandemic, Austria
Tourism is an important pillar of the Austrian economy. According to data provided by the tourism satellite account (TSA), its direct and indirect value-added effects account for almost 7½% of GDP. By European standards, the Austrian tourism sector thus makes an above-average contribution to overall economic output. Almost 6% of total employment in Austria are directly attributable to main tourism activities like “accommodation and restaurants,” “transport” and “culture, sports and entertainment.”
Tourism was particularly strongly affected by the COVID-19 crisis. The first lockdown Austria imposed as of March 16, 2020, led to a sudden drop in revenues by almost 100% in many tourist areas – an economic downturn of unprecedented size and speed. Over the summer months of 2020, tourism in Austria recovered strongly. Therefore, the decline in overnight stays in Austria until fall 2020 was comparatively less pronounced than in Southern European countries like Greece, Spain or Portugal. However, containment measures to fight the COVID-19 pandemic, travel restrictions and travel warnings as well as fears and the perceived risk of COVID-19 infections continue to burden the tourism industry. Since mid-September 2020, these problems have intensified, and since November 2, 2020, the tourism industry has been suffering the consequences of a second lockdown.
Against this background, it is particularly important to closely monitor developments in tourism and to provide timely information about the COVID-19 pandemic’s effects. In the past, statistics on overnight stays were the main data source for tourism analysis. Data on overnight stays are available for several accommodation categories, at a detailed regional level and for all countries of origin of foreign tourists in Austria. This information is highly relevant but has two weaknesses. First, data are published with a time lag of one month. Second, data are only available at a monthly frequency. We therefore supplement our analysis with information gained from payment card service providers. The latter data are available almost in real time and on a weekly basis, thus enabling, on the one hand, a timely assessment of the latest developments and, on the other hand, a very precise chronological representation of containment measures and their impact on Austrian tourism.
This study is structured as follows: Section 1 presents stylized facts on the economic weight of tourism in Austria and its provinces, comparing Austria with other European countries. In section 2, we analyze the economic consequences of the COVID-19 crisis for Austrian tourism between March and November 2020, using two data sources: the number of overnight stays and data on payment card expenditure collected from payment card service providers. In section 3, we provide projections of the path of overnight stays in Austria until end-2020. We give an overview of the 2020 summer tourist season and predict developments in Austrian tourism for the full year 2020. We summarize our results in section 4, outlining potential risks for Austria’s winter tourist season 2020–2021.
1 Tourism is a key sector in the Austrian economy
1.1 The tourism industry generates 7½% of total value added in Austria
In Austria, the tertiary sector plays a dominant role in total economic activity, accounting for more than 70% of total value added (services: 70.2%; agriculture: 1.2%, manufacturing: 28.6%). Under the System of National Accounts (SNA), the tourism sector cannot be precisely separated from other sectors, but it can be approximated by the sum of NACE services sectors I (accommodation and food services) and R (arts, entertainment and recreation). In both sectors, however, it is impossible to distinguish activities of local residents from those of tourists, no matter whether they come from Austria or from abroad. Especially NACE sectors I56 (food and beverage service activities) and R (arts, entertainment and recreation) contain high shares of consumption by domestic nontourists. Nevertheless, the sum of the value added generated by these two sectors provides a first rough estimate of the significance of the tourism sector in Austria: Together, they accounted for 6.6% of total value added in Austria in 2019 (sector I: 5.4%; sector R:1.3%, see table 1). Given the statistical difficulties, the share of 6.6% overestimates the economic weight of the tourism sector. Then again, we might add other NACE sectors – like H50 (water transport), H51 (air transport) and N79 (travel agencies) – to the calculation, which would, in turn, increase the share.
|National account data – value added||EUR million||Share in value added in %|
|Accommodation and food service activities (NACE I)||19,141||5.4|
|Arts, entertainment and recreation (NACE R)||4,468||1.3|
|Sectors I and R||23,608||6.6|
|Tourism satellite accounts (TSA) – GDP||EUR million||Share in GDP in %|
|Direct value added excluding business trips||22,135||5.6|
|Direct value added including business trips||23,545||5.9|
|Direct and indirect value added||29,171||7.3|
|Source: Statistics Austria, Eurostat.|
Given the important role tourism plays in the Austrian economy and its inadequate representation in the SNA, Statistics Austria has been calculating a tourism satellite account (TSA) for Austria since 1999 – based on recommendations by Eurostat, the OECD
and the World Tourism Organization (UNWTO). The TSA uses both supply- and demand-side information, combining it with input-output tables, which makes it possible to more accurately quantify the direct and indirect value-added effects of the tourism sector. In
a narrow sense (i.e. considering only direct effects and excluding business trips), tourism in Austria contributed 5.6% to total GDP in 2019. In a broader sense (i.e. including indirect effects and business trips), its share was 7.3%. This proportion has remained almost unchanged since 2000. 29
Based on the TSA, Statistics Austria calculates tourism consumption expenditure by category on an annual basis. In 2018, accommodation accounted for just over one-third of total tourist spending, followed by expenses for food service activities, which accounted for just under one-fourth. The share of transport was not negligible, either – around 10% of total tourist expenditure was used for air travel and 7% for ground travel (by boat, rail or road). In contrast, tourist expenditure for culture, entertainment and other services made up less than 10% of the total 30 (see Fritz and Ehn-Fragner, 2020, for an in-depth analysis).
Moreover, tourism consumption expenditure can be broken down further into expenditure by foreign tourists and expenditure by domestic tourists. According to this breakdown, foreign tourists account for 54% and domestic tourists for 46% of total tourist expenses in Austria. Matching these figures with the tourism sector’s share of 7.3% in Austrian GDP implies that foreign tourist expenses account for about 4% of Austrian GDP, while domestic tourist expenses account for just above 3%. By comparison, foreign tourists have a 74% share in total overnight stays in Austria, while domestic tourists account for 26% (see table A1 in the annex). This, in turn, implies that day trips play a major role in domestic tourism.
1.2 Significance of tourism in Austria differs widely across regions
The tourism sector’s share in economic activity varies across Austria’s provinces. The first best method to evaluate these differences would be based on regional TSAs. By mid-2021, Statistics Austria will, for the first time, produce consistent regional TSAs for all Austrian provinces (except Vorarlberg). For the time being, we compare the role of tourism in Austria’s provinces on the basis of the share in value added and employment of NACE sector I (accommodation and food service activities) – despite the above-mentioned drawbacks of this data source. The left-hand panels of chart 2 show the shares the individual provinces have in total Austrian tourism (top row: share in value added; bottom row: share in employment). The right-hand panel of chart 2 shows the relative importance tourism has in each province (top row: relative importance for total value added per province; bottom row: relative importance for employment).
The largest contributions to Austria’s tourism come from Tyrol and Vienna. Measured by their share in value-added generated by NACE sector I in Austria, Tyrol ranks first, followed by Vienna. This ranking is reversed when we look at employment shares. The employment share of tourism is higher than the value-added share in Vienna, which reflects the fact that food service activities, which are more employment intensive, are of higher economic importance in Vienna than in Tyrol.
What is more useful in measuring the economic importance of tourism in Austrian regions is its share in value added or in employment (see chart 2, right-hand panel). According to both criteria, Tyrol has the largest tourism sector, with tourism accounting for a share of 15% in value added and of 13% in employment. Salzburg, Vorarlberg, Carinthia and Burgenland follow. Vienna, Styria, Lower Austria and Upper Austria are all below the Austrian average. In addition to the mere size of the tourism sector, other factors are also significant for assessing its vulnerability in the current crisis. For instance, the COVID-19 crisis affected cities and regions with a high proportion of tourists from distant countries with particular strength, as we show in section 2. Vienna’s tourism, in particular, is additionally affected by the COVID-19 crisis as several major international conferences had to be canceled.
1.3 Economic importance of Austrian tourism industry above EU average
The economic importance of the tourism sector varies substantially across European countries. A comprehensive comparison is difficult as comparable data for all European countries are not available. Chart 3 shows the results of the available TSAs, more specifically internal tourism consumption (sum of domestic and inbound (foreign) tourist expenditure) as a proportion of domestic supply (as measured in gross production value). Overall, the results have to be interpreted with caution as, on the one hand, survey years differ significantly across countries, ranging from 2010 (Malta) to 2018 (Netherlands), and, on the other hand, different methods or definitions were used in the calculations (see Eurostat, 2019). Bearing this in mind, we find that the economic importance of the tourism industry in Austria is above the EU average (see chart 3).
In addition, chart A1 in the annex illustrates the economic importance of NACE sector I (accommodation and food service activities) for value added, employment and hours worked in a comparison of European countries. The corresponding data are available for all EU countries, confirming the above assessment and in particular the fact that, by international standards, the tourism sector makes an above-average contribution to economic activity in Austria.
2 COVID-19 lockdown severely affected Austrian tourism
The measures taken to contain the COVID-19 pandemic have hit the tourism industry, like many other sectors of the economy, on both the supply and the demand side. Border controls and strict entry rules, quarantine regulations and the closing of accommodation establishments and restaurants are among the major supply-side shocks. The reduced demand for holiday travel given the risk of infection as well as the severe economic downturn are the most important demand-side shocks.
Our analysis of the economic impact of the COVID-19 crisis on Austrian tourism rests upon two data sources: first, the number of overnight stays as reported by Statistics Austria and second, data collected from payment card service providers on expenditure on accommodation and other tourism-related goods and services. Data on the number of overnight stays are available on a monthly basis up to and including September 2020; advance information for some (sub)categories is already available for October 2020. These data comprise information on overnight stays broken down by Austrian provinces and by accommodation categories. In both cases, a cross-classification according to tourists’ countries of origin is available as well. In contrast, data collected from payment card service providers are available on a weekly basis up to and including end-November 2020 and include information on expenditure by Austrian residents and nonresidents for several consumption categories. In our analysis, we focus on payment card expenditure
on accommodation (including hotels, holiday homes, private rooms, campsites, recreation facilities and other accommodation) to ensure comparability with information about overnight stays. The payment card data considered here cover almost the entire turnover of card transactions in Austria. Moreover, one provider with a substantial market share provides detailed information on the country of origin of cards used in Austria. The prompt availability of these data enables us to analyze changes in the tourism sector almost in real time. 31
As shown in table 2 and chart A2 in the annex, overnight stays and payment card expenditure on accommodation have followed a very similar course during the COVID-19 crisis. In April 2020, during the first lockdown in Austria, both indicators dropped by almost 100%. For the weeks and months after the lockdown, payment card data show a stronger recovery than overnight stays, especially with regard to domestic tourists. The shift in consumer preferences toward cashless means of payment (Fenz and Stix, 2020) seems to be the main reason for these differences. 32 In the following sections, we will use both data sources to describe in detail the developments in Austrian tourism during the COVID-19 crisis and to give an outlook for the remaining months of 2020.
|Overnight stays||Payment card expenditure for accommodation establishments|
|Total||Foreign tourists||Domestic tourists||Total||Foreign tourists||Domestic tourists|
|Annual change in %|
|Source: Statistics Austria, payment card service providers, OeNB.|
|Note: x = data not available yet.|
2.1 First COVID-19-induced lockdown hits Austrian tourism in the off-season
As payment card data on travel expenses are available on a weekly basis, they allow for a very precise chronological representation of the containment measures and their impact on the tourism sector in Austria. Following the lockdown imposed on March 16, 2020 (calendar week 12), card payments by foreign tourists fell by 100%, those by Austrian residents by almost 100%. With the reopening of accommodation establishments on May 29, 2020, payment card spending by residents recovered quickly and soon exceeded 2019 levels. This trend was supported by people’s strong preference for spending their vacation in Austria and not abroad and by a shift toward cashless means of payment. From June 4, 2020, onward, borders were gradually reopened, but expenses by foreign tourists in Austria recovered only slowly and did not start to rise significantly before the second half of June 2020. The sharp declines observed during the lockdown fell into the off-season: Together, the months of April and May account for only 10% of the annual number of overnight stays (see table A2 in the annex).