Stayin’ alive? Government support measures in Portugal during the Covid-19 pandemic

Mateus, Márcio; Neugebauer, Katja Article — Published Version Stayin’ alive? Government support measures in Portugal during the Covid-19 pandemic Portuguese Economic Journal Provided in Cooperation with: Springer Nature Suggested Citation: Mateus, Márcio; Neugebauer, Katja (2025) : Stayin’ alive? Government support measures in Portugal during the Covid-19 pandemic, Portuguese Economic Journal, ISSN 1617-9838, Springer, Berlin, Heidelberg, Vol. 24, Iss. 3, pp. 335-372, https://doi.org/10.1007/s10258-025-00271-2 This Version is available at: https://hdl.handle.net/10419/330641 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Vol.:(0123456789) Portuguese Economic Journal (2025) 24:335–372 https://doi.org/10.1007/s10258-025-00271-2 ORIGINAL ARTICLE Stayin’ alive? Government support measures inPortugal duringtheCovid‑19 pandemic MárcioMateus2 · KatjaNeugebauer1 Received: 13 July 2023 / Accepted: 4 January 2025 / Published online: 18 March 2025 © The Author(s) 2025 Abstract During the Covid-19 crisis, the Portuguese government has provided a plethora of different support measures for firms. These included state-guaranteed loans and a public moratorium for existing loans. This paper examines the access to and uptake of these measures. What were the characteristics of firms being granted state-guaranteed loans? Were they different for firms accessing the moratorium? Did state-guaranteed loans potentially lead to an increase in zombie lending? We try to answer these questions using highly granular bank-, firmand loan-level data for Portugal. We find that guaranteed loans went mostly to firms operating in the sectors most severely hit by the pandemic and to firms that previously had a credit relation and/or benefitted from a state guarantee. Furthermore, the Portuguese public guarantee scheme seems to mainly have supported lower-credit-risk firms. In addition to that, riskier firms also paid higher interest rates and obtained smaller guaranteed loans than more viable firms. However, in contrast to our results for the state guarantees, we find that riskier firms were more likely to benefit from the public moratorium. Keywords Zombie lending· Zombie firms· Credit misallocation· Evergreening· State aid JEL G30· G38· G21 1 Introduction In 2020, the world was hit by the Covid-19 pandemic. Unlike previous crises, this one hit countries completely unexpectedly and to a rarely seen extent, leading to the adoption of large-scale state-support measures to keep economies afloat. * Márcio Mateus [email protected] Katja Neugebauer [email protected] 1 BancodePortugal, Portugal 2 European Systemic Risk Board Secretariat, Frankfurt, Germany 336 M.Mateus, K.Neugebauer Governments around the globe implemented emergency actions, ranging from social-distancing measures, testing and quarantining policies, to income and liquidity-support measures, to help households and firms. Against this background, many countries relied on the financial system to provide government-backed liquidity to support firms in dealing with the effects of lockdowns, which had led to an abrupt and, in many cases, sustained drying up of income. In the case of Portugal and most other European countries, these liquidity-support measures involved state guarantees for new loans and a debt moratorium for existing ones. These measures have been essential to support firms in the most acute phase of the crisis, providing liquidity at reduced costs in a context of an abrupt increase in the level of risk. In this paper, we look at the characteristics of the firms that have benefitted from these support measures. Did loans with state guarantees only go to firms that were viable before the onset of the pandemic, or did they also lead to an increase in the credit granted to unproductive and high-risk firms? Were there any significant differences between the risk profiles of firms accessing the moratorium and those accessing the public guarantees, since access to the latter was much stricter? We try to answer these questions in this paper using detailed loan-level data from Banco de Portugal’s Central Credit Register, matched with both firm and bank balance-sheet data. Over the last years, public credit-guarantee schemes have gained popularity worldwide as a tool to increase the availability of loans to financially constrained firms, typically SMEs or start-ups. Most of the literature on public credit-guarantee schemes focuses on the existence of asymmetric information between the lender and the borrower, frequently associated with the lack of adequate collateral, as a justification for government interventions in the credit market (Berger and Udell 2006 and Beck etal. 2010). The absence of government interventions might otherwise result in the undersupply or rationing of lending (Mankiw 1986; Gale 1990, 1991). Compared with direct lending by a public institution, loans guaranteed by government-backed institutions but distributed by banks present several advantages. First, a Covid-19 state guarantee is only considered as public debt if and when the guarantee is called, which means that it entails much lower costs as compared to direct lending. This is particularly important for Portugal, one of the countries with the highest levels of public debt in the euro area. Second, since screening and monitoring of borrowers is left to private institutions, the risk of politically-induced lending is being mitigated (Khwaja and Mian 2005). Furthermore, as state guarantees usually do not cover the full loan amount, banks also bear some credit risk, thus limiting moral-hazard concerns. Whereas economic theory tells us that capital should go where it yields the highest return and therefore banks should allocate it accordingly, the existence of a public-guarantee scheme creates a new set of incentives. On the one hand, if access conditions to guaranteed loans are very strict, to avoid high costs of the programme for public finances ex-post, only those firms that would obtain bank credit anyway will be benefitting from state guarantees. In this case, the benefit of public guarantees would be restricted to providing credit to firms at lower interest rates, without improving the overall access of firms to credit. On the other hand, if access conditions are too generous, state guarantees might lead to adverse selection, attracting 337 Stayin’ alive? Government support measures inPortugal during… riskier borrowers and thus downgrading the quality of the pool of borrowing firms. Finally, a moral-hazard problem can arise as financial institutions may also have fewer incentives to monitor the borrowers after granting the loan, since the largest part of these loans is guaranteed by the state. Despite the positive short-term impact of state-guaranteed loans in the context of the Covid-19 pandemic, the mediumto long-term impact can be problematic if it contributes to the survival of unproductive or very risky firms. In fact, the literature suggests that the survival of unproductive firms—often called zombie firms—may hamper the growth of their more productive competitors through congestion effects, as zombie firms retain a certain market share and use scarce productive inputs (Caballero etal. 2008; McGowan etal. 2018). Zombie lending can also be considered a cause of competitive distortions, with a negative impact on healthy firms (see, e.g., Acharya etal. 2019, Acharya etal. 2024, Blattner etal. (2023) forthcoming, Banerjee and Hofmann 2020, McGowan etal. 2018, Schmidt etal. 2021). Furthermore, it can potentially prevent the "creative destruction" of firms à la Schumpeter (1942). The relationship between credit allocation and productivity is of crucial importance for Portugal, as productivity growth has been subdued since the turn of the millennium, and the aggregate productivity level is relatively low compared to other EU countries. The lower average level of qualification of Portuguese workers and managers and the smaller scale of Portuguese firms, when compared to other EU countries, may certainly explain part of this phenomenon but there is also evidence of inefficient credit allocation between 2008 and 2016 (Azevedo etal. 2022). The effectiveness of state-support measures has gained renewed attention in the wake of the Covid-19 pandemic, and several papers have tried to gauge the impact of the state-support measures that have been put into place. Using an accounting exercise, Schivardi and Romano (2020) find that, without government support, a large number of Italian companies would have become illiquid very early on in the crisis. However, they also show that government support measures in the form of credit guarantees could cover more or less all the liquidity needs of Italian firms. Furthermore, they argue that, during exceptional periods like the Covid-19 pandemic, it might be worthwhile throwing potential zombie firms a lifeline in order to protect the functioning of production or value chains they might be involved in. During economically stressful times, the loss of a link in the production chain (i.e., the closure of a zombie firm) might lead to missing inputs for other firms in the same value chain, potentially causing a domino effect. Core and De Marco (forthcoming) use Italian loan-level data to test how efficient the private sector has been in allocating public guarantees. They find that larger banks and banks with better information technology played an important role in the efficient allocation of these loans. They also find a positive impact of relationship lending in Italy. A particularly interesting finding of their analysis is that riskier firms had higher take-up rates for state-guaranteed loans. Cascarino etal. (2022) also examine the effect of the Italian public loan guarantee programme on credit additionality during the Covid-19 pandemic. They find that public guarantees were effective in increasing credit supply, especially for smaller and riskier firms. Furthermore, the findings suggest that bank capital played a fundamental role in supporting higher lending through guaranteed 338 M.Mateus, K.Neugebauer loans in the face of the exceptional liquidity shock. In addition, guaranteed loans were not granted more frequently to ex-ante riskier firms, and the credit additionality of the guarantees did not substantially depend on firm characteristics, with only a slightly higher substitution of existing credit for guaranteed loans to riskier firms. Bighelli etal. (2021) use cross-country micro-level data in order to examine the potential consequences of the Covid-19 pandemic on productivity in Croatia, Finland, Slovakia and Slovenia. In doing so, the authors also analyse which firmlevel characteristics had an impact on the likelihood of benefitting from government subsidies. They mainly look at wage subsidies that allowed companies to continue paying their employees. They find that the more productive firms in Croatia, Slovakia and Slovenia had a higher likelihood of receiving state aid, although the amount of the subsidies was lower for those firms. In the case of Finland, productivity did not seem to have an impact on the allocation of government subsidies. The authors also find that only a small share of the subsidies went to zombie firms. Altavilla etal. (2021) examine whether state-guaranteed loans actually led to an increase in the credit supply to firms, or whether they merely were used as a substitute for nonguaranteed loans. Using credit-register as well as supervisory data for four large euro-area countries (France, Germany, Italy, Spain), they find that there was some substitution taking place but that government guarantees largely contributed to an extension of credit. Furthermore, this new credit went mainly to small and largely creditworthy firms in the sectors most affected by the pandemic. The banks extending these guaranteed loans were predominantly large, liquid and well-capitalised. Jiménez etal. (2024), using data from the Spanish credit register for the Covid-19 period, find that public loan guarantees were more likely to be granted to riskier firms and small and medium enterprises. Additionally, public loan guarantees were more likely to be extended by banks with higher non-performing loan ratios, lower capital ratios, and lower profitability. In contrast, the same firm and bank characteristics had the opposite effect on the likelihood of obtaining loans without a public loan guarantee. The above-mentioned results present mixed evidence concerning a future risk of zombiefication, as it seems that economically viable as well as unviable firms have profitted from these state-support measures. Given the strict access requirement to state-guaranteed loans in Portugal during the crisis, it will be interesting to see whether results are more clear-cut in the Portuguese case. Our analysis starts out by mapping the evolution of zombie firms in Portugal. In a next step, we then analyse whether zombie firms were more likely than nonzombie firms to take out government guaranteed loans. However, the Portuguese government had imposed strict rules on the eligibility for government guarantees, and therefore the uptake by zombie firms should have been largely prevented. Furthermore, firms could also make use of a public moratorium, to which the access conditions were much less strict. Therefore, we add to the existing literature by not only looking at the access of zombie firms to government-guaranteed loans but also their take-up of the public moratorium. Indeed, we expect to find a larger share of zombie and quasi-zombie firms to have made use of the moratorium as compared to the state guarantees. This might imply risks to financial stability now that the moratorium has ended and firms need to continue paying back their loans. 339 Stayin’ alive? Government support measures inPortugal during… We find that the number of zombie firms in Portugal has decreased over time. Descriptive results show that while only a small share of zombie and quasi-zombie firms have been granted state-guaranteed loans, their share is substantially higher when it comes to access to the public moratorium. Our regression results confirm these findings. On the one hand, the Portuguese public-guarantee scheme seems to mainly have supported lower-credit-risk firms. On the other hand, we find that riskier firms were more likely to apply for the public moratorium. Furthermore, our results also show that riskier firms paid higher interest rates and obtained smaller guaranteed loans than more viable firms. In addition to that, we find that firms that previously had a credit relation and/or benefitted from a state guarantee were more likely to also benefit this time. Finally, we find that firms in the sectors most affected by the pandemic were more likely to benefit from the state guarantees and the moratorium. Our paper is structured as follows. Section2 gives some institutional background on state-support measures in Portugal during the Covid-19 pandemic. Section 3 gives an overview of the data being used, the zombie methodology chosen, and presents some descriptive statistics. Section4 presents the regression results, and Sect.5 concludes. 2 Portuguese government‑support measures duringtheCovid‑19 pandemic Access to state-support measures during the Covid-19 pandemic differed across countries. This section gives an overview of the respective public-guarantee scheme and the moratorium in Portugal. 2.1 The Portuguese public guarantee scheme The Portuguese public guarantee scheme started its operation in 1994, when the first public entity entitled to grant public guarantees (SPGM—Sociedade de Investimentos) was created. Since then, four more public entities, commonly referred to as Sociedades de Garantia Mútua (SGM), have been given the right to grant public guarantees: Agrogarante, Garval, Lisgarante and Norgarante. However, this type of government support only experienced a major boost in Portugal, like in other European countries, in the aftermath of 2007–08 global financial crisis. Between 2007 and 2010, the outstanding amount of guarantees increased from €0.5 to €3.8 billion (Fig.1). In response to the Covid-19 pandemic, the Portuguese government strengthened the SGM capacity to issue guarantees. Several credit lines were created to support specific sectors of activity, particularly those most affected by the pandemic, or specific regions. Between March 2020 and July 2021, the stock of bank loans to nonfinancial corporations (NFC) with a state guarantee attached increased from about €5 billion, around 8% of the stock of bank loans to NFCs, to about €13 billion, about 17% of the stock (Fig.2). 340 M.Mateus, K.Neugebauer The features of the Covid-19 guaranteed credit lines complied with the European Commission’s Temporary Framework for state-aid measures to support the economy during the COVID-19 outbreak.1 Most of the credit lines have a maximum maturity of six years, although some can reach up to 10years (some consider a grace period between 1 and 2years). The maximum amount of the guarantees provided varies according to the credit line. In some cases, a maximum amount per firm is set but usually it is proportional to the number of employees or the firm size. In all credit Fig. 1 Annual amount of guarantees issued and end-of-year amounts outstanding. Source: Annual income statements of SPGM—Sociedade de Investimento and Banco Português de Fomento Fig. 2 Stocks and shares of bank loans to NFCs with a state guarantee. Source: Banco de Portugal’s Central Credit Register 1 https:// ec. europa. eu/ compe titionpolicy/ stateaid/ coron avirus/ temporary-framework_en. 341 Stayin’ alive? Government support measures inPortugal during… lines, the maximum amount per beneficiary is capped at one quarter of the sales in 2019 or twice the annual payroll of the beneficiary for 2019, or for the last year available. The guarantee level varies between 70 and 90% of the principal amount due. The spreads of the guaranteed loans are capped at 1%, 1.25% and 1.5% on loans with a maturity below 1year, between 1 and 3years, and between 3 and 6years, respectively. Additionally, firms also incur a guarantee fee that varies between 0.25% and 2%, depending on firm size and loan maturity. Finally, the vast majority of guaranteed lending is associated with genuinely new loans, not a replacement of existing credit facilities with the lender. The eligibility criteria are credit-line specific, and the main factor of differentiation is usually the sector of activity the line was targeted at. Nevertheless, there are many common features between the different credit lines. State guarantees cannot be granted to firms qualified as "undertakings in difficulty"2 on 31 December 2019, to firms with credit incidents pending within the banking system or the mutual guarantee scheme system, to firms whose tax or social security situation is not in order, or to firms with negative equity on the last approved balance sheet.3 Firms whose headquarters or management are located in countries, territories or regions with a clearly more favourable tax regime and large firms with more than 3,000 employees, operating in sectors other than the tourism sector,4 were also not eligible. Additionally, the access to some credit lines was also restricted to firms whose turnover dropped below a certain threshold, usually 25% or 40%, via-à-vis pre-Covid levels. Less frequently, firms were also required to maintain all permanent jobs up to 31 December 2020. 2.2 The Portuguese public moratorium One of the most frequent measures across Europe in response to the pandemic was a loan moratorium for borrowers. Against the initial background of an abrupt decline in firms’ turnover and the reduction in households’ disposable income within a very short time period, there was a significant risk of the borrowers not being able to fulfill their credit-related obligations. If aligned with European Banking Authority Guidelines (EBA/GL/2020/02), moratoria allowed lenders to not reclassify the loans benefitting from this measure, almost automatically, as forborne. This favourable prudential treatment for loans covered by the moratoria avoided an increase in capital costs and impairments that would have resulted from an almost automatic reclassification. Given the severity of the risks associated with the Covid-19 pandemic, the Portuguese government – as also observed in other European countries—established a 2 As defined in Article 2 (18) of the Commission Regulation (EU) No 651/2014 of 17 June 2014. 3 Companies with negative equity on the last approved balance sheet were allowed to access a guaranteed credit line if they presented this situation to be regularised in the interim balance sheet until the date of the respective application. 4 There is only one credit line for which large firms with more than 3,000 employees were eligible, which was Linha Covid – TURISMO (Médias e Grandes Empresas). The tourism sector, as defined by this credit line, includes the following NACE codes: 49,392, 55, 56, 77, 79, 82,300, 90, 91, 93 and 96,040. 342 M.Mateus, K.Neugebauer public moratorium regime.5 Initially in force between 27 March 2020 and 30 September 2020, it was first extended until 31 March 20216 and then, in the context of the worsening pandemic, until September 2021.7 With this latter amendment, firms were entitled to request the application of the public moratorium until 31 March 2021, and up to a maximum of nine months, for loans that did not benefit from this measure before. Only loans granted before the moratorium came into force (i.e., before 27 March 2020) were considered eligible for the moratorium. The large majority of firms joined the moratorium until June 2020 (Fig.3). The Portuguese moratorium regime introduced a set of measures. First, an extension of credit agreements with principal payment at the end of the contract (bullet loans), under the same terms and for a period equal to the duration of the moratorium. Second, the suspension of the payment of principal, income and interest with maturity scheduled until the end of that period, for the period during which the measure is in effect. Third, the prohibition of revoking credit-line agreements and loans granted for the amounts contracted at the date of entry into force of the Decree-Law (27 March 2020). Access to the moratorium depended on the cumulative compliance with the following requirements: i) firms are required to have a head-office and economic activity in Portugal; ii) not being part of the financial sector; iii) to have their tax and social security situation in order and; iv) the loan benefitting from the moratorium could not be more than 90days overdue. Loans granted to finance the acquisition of securities or positions in other financial instruments and credit cards for individual use were also not eligible. Fig. 3 Amounts and shares of NFC loans under the public moratorium. Source: Banco de Portugal’s Central Credit Register 5 Under Decree-Law No 10-J/2020 of 26 March 2020. 6 Under Decree-Law No 26/2020 of 16 June 2020. 7 Under Decree-Law No 107/2020 of 31 December 2020. 349 Stayin’ alive? Government support measures inPortugal during… Table 2 Descriptive statistics. This table presents summary statistics for the different samples used for estimation in this paper. The upper half of the table presents the state-guarantee sample. It contains data at the firm level, the loan level as well as regional and bank-level data. Descriptive statistics for the moratorium sample are presented in the lower half of the table State-guarantee sample N Mean Std.Dev 5th pct Median 95th pct Firm level data State-guarantee (dummy) 253,251 0.2 0.4 0 0 1 Most affected sector (dummy) 253,251 0.235 0.424 0 0 1 Previous credit relation (dummy) 253,251 0.513 0.5 0 1 1 Previous state guarantee (dummy) 253,251 0.124 0.33 0 0 1 Assets (€ millions) 253,251 1.247 12.5 0.015 0.177 3.715 Cash / Assets 253,251 0.237 0.261 0.001 0.133 0.824 EBITDA / Assets 253,251 0.106 0.765 −0.123 0.082 0.487 Probability of default (%) 253,251 2.554 3.354 0.237 1.386 9 Zombie 253,251 0.256 0.399 0 0 1 Negative GVA (dummy) 253,251 0.103 0.304 0 0 1 Loan level data Loan amount (€)—with state-guarantee 61,560 132,532 307,101 10,000 50,000 500,000 Loan amount (€)—without state-guarantee 231,235 61,386 476,707 1,189 15,000 191,482 Loan interest rate (%)—with state-guarantee 61,558 1.432 0.509 0.671 1.5 2.25 Loan interest rate (%)—without stateguarantee 230,465 3.583 3.714 0 2.5 10.471 Loan maturity (years)—with state-guarantee 61,560 5.379 1.111 3.833 6 6 Loan maturity (years)—without stateguarantee 231,235 1.58 2.559 0.083 0.333 6 Guarantee (other than state)—with stateguarantee 61,560 0.368 0.482 0 0 1 Guarantee (other than state)—without stateguarantee 231,235 0.576 0.494 0 1 1 Region level data Excess mortality 7 1.121 0.066 1.045 1.135 1.222 Bank-level data Bank assets (€ millions) 34 10,688 22,503 33 787 81,651 Bank own funds ratio (%) 34 31.083 27.663 13.033 19.275 95.808 Bank NPL ratio (%) 34 7.405 10.64 0.134 4.94 25.404 Bank ROA (%) 34 0.131 2.303 −5.804 0.551 2.791 Interbank funding / Assets (%) 34 26.6 35.88 0 6.61 95.18 Moratorium sample N Mean Std.Dev 5th pct Median 95th pct Firm level data Moratorium (dummy) 138,921 0.349 0.477 0 0 1 Most affected sector (dummy) 138,921 0.263 0.44 0 0 1 Previous state guarantee (dummy) 138,921 0.238 0.426 0 0 1 Assets (€ millions) 138,921 2.321 71.178 0.023 0.245 5.142 Cash / Assets 138,921 0.159 0.191 0.002 0.085 0.58 EBITDA / Assets 138,921 −0.038 9.603 −0.222 0.08 0.38 Probability of default (%) 138,921 3.085 4.183 0.181 1.641 10.875 350 M.Mateus, K.Neugebauer (the median of firms’ assets is slightly below €180,000), have a comfortable liquidity position (the median of the cash-to-assets ratio is around 13%), a low leverage ratio (the median of liabilities over assets is about 53%) and a low probability of default (half of the firms in the sample had a probability of default over a one-year horizon below 1.4%). The level of zombieness is relatively low, and about 10% of the firms recorded a negative gross value added in 2019 Table3 compares the descriptive statistics of our sample of firms eligible for the state guarantee with those of non-eligible firms. About 133,000 firms, which corresponds to around 27% of the Portuguese firms, were non-eligible for a state guarantee. The share of firms operating in the sectors most affected by the pandemic is higher for the non-eligible firms (32.6% versus 23.1%), which suggests that a non-negligible part of the firms operating in those sectors were excluded from the guaranteed credit lines. The share of firms with a credit relation or a state guarantee before the beginning of the pandemic was significantly lower for the non-eligible firms (40.8% and 3.9%, versus 51.5% and 12.3%, respectively). Looking at the median, we can observe that eligible firms are significantly larger (€180,000 versus €33,000), have more liquidity (13% versus 6.1%), are more profitable (8.2% versus −3.6%) and markedly less leveraged (53% versus 161%) than non-eligible firms. Moreover, eligible firms are much less likely to enter into default and record a lower level of zombieness (1.4% and 0 versus 7.1% and 0.85, respectively). Finally, the share of firms with a negative GVA is almost three times higher for the non-eligible Table 2 (continued) Zombie 138,921 0.29 0.406 0 0 1 Negative GVA (dummy) 138,921 0.066 0.247 0 0 1 Table 3 Descriptive statistics: eligible vs. non-eligible firms for state-guarantees. This table presents summary statistics of firm-level explanatory variables for eligible and noneligible firms with respect to access to the state guarantees Mean 5th pct. Median 95th pct. Eligible Noneligible Eligible Noneligible Eligible Noneligible Eligible Noneligible Most affected sector (dummy) 0.231 0.326 0.000 0.000 0.000 0.000 1.000 1.000 Previous credit relation (dummy) 0.515 0.408 0.000 0.000 1.000 0.000 1.000 1.000 Previous state guarantee (dummy) 0.123 0.039 0.000 0.000 0.000 0.000 1.000 0.000 Assets (€ millions) 1.249 2.048 0.015 0.001 0.177 0.033 3.714 1.385 Cash / Assets 0.237 0.190 0.001 0.000 0.133 0.061 0.824 0.892 EBITDA / Assets 0.106 −10.685 −0.123 −5.348 0.082 −0.036 0.487 0.340 Leverage 0.513 268.010 0.037 0.564 0.527 1.611 0.955 27.339 Probability of default (%) 2.555 9.734 0.237 1.020 1.386 7.094 9.001 27.599 Zombie 0.256 0.537 0.000 0.000 0.000 0.846 1.000 1.000 Negative GVA (dummy) 0.104 0.309 0.000 0.000 0.000 0.000 1.000 1.000 351 Stayin’ alive? Government support measures inPortugal during… firms (30.9% versus 10.4%). These results corroborate the increased financial fragility of firms not eligible for the state guarantees 3.4.2 Moratorium sample The uptake of the moratorium outperformed the one of the state-guarantee scheme, as about 35% of eligible firms made use of it (Table2). As with the state-guarantee scheme, around one quarter of eligible firms were in one of the sectors most affected by the pandemic. The share of firms in our moratorium sample that benefitted from a state guarantee before the pandemic was 24%. Firm size, leverage, probability of default and level of zombieness is larger in the moratorium sample, when compared with the state-guarantee sample, whereas the share of liquid assets and the share of firms with a negative GVA is somewhat lower. Overall, the descriptive statistics show an increased riskiness of the moratorium sample 3.4.3 Zombie firms inPortugal Figure5 shows the development in the level of zombieness as well as the share of zombie firms according to the McGowan etal. (2018) and Storz etal. (2017) definitions. Results for these definitions indicate a decline in the share of Portuguese zombie firms over time. Using the McGowan etal. (2018) definition, the share of zombie firms in Portugal in 2019 stood at 6.9%, a decline by 4 p.p. from its peak in 2014. Looking at the binary zombie definition by Storz etal. (2017), results indicate a zombie share of only 4.8% in 2019 (down from 10.4% in 2012). However, by making this definition fuzzy, results get much richer. Figure 5 also plots the share of firms with a zombie score of 0.5 or higher. In 2019, 28.0% of Portuguese Fig. 5 Share of zombie and quasi-zombie firms in Portugal | in %. Source: Central Balance-Sheet Database and authors’ calculations 352 M.Mateus, K.Neugebauer firms fell into this range. Breaking this further down shows that 14.1% of Portuguese firms were in the range between 0.9 < Z < 1 in 2019, i.e., they were relatively close to being full zombies (Z = 1). Most firms in this range fail to be classified as full zombies according to the binary zombie definition only because they recorded zero net investment. 9.2% of the firms had a zombie score between 0.5 and 0.9 in 2019. As was the case for the binary definition, the share of zombies using the fuzzy definition has declined steadily since its peak in 2013, where it stood at 36.1%. This decline is also observed when looking at the employmentand asset-weighted shares of full zombie firms, obtained by weighting the full zombies by their respective shares of employment and assets within the overall population of firms. The share of employment-weighted zombie firms stood at 8.8% in 2013 and declined to 3.1% in 2019, whereas the share of asset-weighted zombies declined from 7.7% in 2013 to 2.9% in 2019. Comparing these weighted shares with the unweighted ones indicates that Portuguese zombie firms are smaller than the average firm, accounting for less employment and total assets 3.4.4 Zombie firms andtheir uptake ofthePortuguese state‑support measures Between March 2020 and June 2021, 54.8% of the amount of the new loans with state guarantees went to non-zombie firms (firms’ level of zombieness was assessed based on the data available up to 2019, i.e., before the pandemic), whereas this share was 52.9% for new loans without state guarantees (Fig.6). Only 0.9% of the amount of new loans with a state guarantee was granted to full zombie firms. The proportion was almost identical for new loans without a state guarantee, about 0.8%. 38.4% of new loans with a state guarantee were obtained by firms with scores below 1 and higher than or equal to 0.5. About half of this amount went to firms with a zombie score of 0.8 or higher but below 1. The fact that loans with state guarantees Fig. 6 Share of new loans granted to zombie and quasi-zombie firms. Source: Central Balance-Sheet Database, Central Credit Register and authors’ calculations 353 Stayin’ alive? Government support measures inPortugal during… predominantly went to non-zombie firms or firms with low zombie scores reflects the strict access conditions for these types of loans As detailed in Sect.2.2, access requirements were softer for the moratorium than those for state-guaranteed loans. Nevertheless, only 4.0% of the amount of the loans under the moratorium belongs to full zombie firms. The respective share is 53.1% for firms with zombie scores above (or equal to) 0.5 and below 1 (Fig.7). This indicates that the share of loans under the moratorium that belongs to lowerquality firms is larger than that of higher-quality firms. However, it is noteworthy that also a large share of loans under the moratorium belongs to non-zombie firms (37.3%) 4 Results In this section we analyse the impact of firm, bank and loan characteristics on the uptake of state-guaranteed loans and the moratorium during the pandemic. The starting point of our analysis is the regression setup used by Core and De Marco (2024), who look at state-support measures in Italy during the Covid-19 pandemic. First, based on firm-level data, we estimate the factors influencing the likelihood of firms getting a guaranteed loan. Second, using loan-level data, we analyse the variables influencing the interest rates and loan amounts of new loans granted during the pandemic. Third, we return to the firm-level data to analyse the factors influencing the moratorium uptake, comparing the risk profile of firms that benefitted from the moratorium with those that benefitted from the state guarantees. All variables presented in the regression tables have been normalised to have a mean of 0 and a standard deviation of 1. Thus, all the coefficients can be compared and read as the Fig. 7 Share of loans under the moratorium allocated to zombie and quasi-zombie firms. Source: Central Balance-Sheet Database, Central Credit Register and authors’ calculations 354 M.Mateus, K.Neugebauer effect of a one standard deviation increase. Estimations are done using Stata’s reghdfe (Correia 2017), which allows for multi-way fixed effects and clustering. 4.1 State guarantees duringthepandemic—firm‑level evidence In this section, we analyse the impact of firm, region and sector characteristics on the uptake of new state guarantees during the pandemic by estimating the following linear probability model: where State Guaranteef,r,s is a dummy variable that is equal to one if firm f, whose headquarters are located in region r and that operates in sector s has been granted a loan with a state guarantee, and zero otherwise. The control group consists of firms that were eligible but did not receive any state-guaranteed loan between April 2020 and June 2021. Most-affected sectors is a dummy variable that takes on the value of one for firms in those sectors that recorded a decrease in turnover of more than 40% in the second quarter of 2020, compared to the expectable situation in a scenario without the pandemic, and zero otherwise. Excess mortalityr is the ratio of the number of human deaths recorded between March 2020 − April 2021 and March 2019 − April 2020 in each region, and vector Xf contains a set of firm-specific controls, as detailed in Sect.3.2. All firmlevel variables refer to the end of 2019. Additionally, we also include three dummy variables that are equal to one if a firm recorded a negative GVA in 2019, accessed a state-guaranteed loan in 2019 or at the beginning of 2020 (before the pandemic) or had a credit relation prior to the onset of the pandemic, respectively, and zero otherwise. Finally, specification 1 also contains region, sector or region-sector fixed effects, thus controlling for unobserved heterogeneity. Standard errors are clustered at the sector level. The results for the guaranteed-loans uptake are presented in Table4. The estimates presented in column (1), using only region fixed effects, indicate that operating in one of the sectors that were most affected by the pandemic increases firms’ probability of accessing a state-guaranteed loan by 12.9 percentage points, i.e., there is a 65% higher probability compared with the mean take-up rate of 20%. The coefficient of this variable is large in comparison with the other explanatory variables. This is in line with expectations, since most of the state-guaranteed credit lines were specifically targeted at those sectors. Interestingly, the estimates in column (2) suggest that firms located in regions with a higher excess mortality have a lower probability of accessing a state-guaranteed loan. However, since we are not controlling for other region characteristic in this regression setup, this coefficient may be capturing other unobserved regional effects. The most important factors influencing the likelihood of firms getting a state guarantee, in all specifications, is having already had a credit relation or a stateguaranteed loan in the past. Column 4, our most saturated specification in terms (1) State Guarantee f,r,s= 𝛽 0+ 𝛽 1 Most-affected sector s+ 𝛽 2 Excess mortalityr +𝛽 3 X f +FE +𝜀 f,r,s 355 Stayin’ alive? Government support measures inPortugal during… of fixed effects, indicates that if a firm had any state-guaranteed loan before the beginning of the pandemic (between January 2019 and March 2020), it was 19 percentage points more likely to receive a state-guaranteed loan, compared to a firm without a previous state-guaranteed loan. In the same vein, firms with a credit relation prior to the onset of the pandemic were 17.6 percentage points more likely to receive a state-guaranteed loan vis-à-vis firms without a credit relation. These effects are large and also intuitive, as firms that were already in the credit market and firms that had successfully gone through the application Table 4 State-guaranteed loans (firm-level data). This regression table presents estimation results obtained by using a Linear Probability Model. The dependent variable is a dummy that is equal to one for eligible firms that received a government-guaranteed loan during the Covid-19 pandemic, and zero for eligible firms that did not. Variables have been normalised to have a mean of zero and a standard deviation of one. Standard errors are clustered at the sector level. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively Dependent variable State-guarantee dummy (1) (2) (3) (4) (5) Most affected sector 0.129*** (0.024) Previous credit relation 0.198*** (0.007) 0.174*** (0.008) 0.177*** (0.008) 0.176*** (0.008) 0.172*** (0.008) Previous state guarantee 0.206*** (0.01) 0.189*** (0.006) 0.192*** (0.006) 0.190*** (0.007) 0.189*** (0.007) Assets 0.070*** (0.018) 0.103*** (0.012) 0.098*** (0.012) 0.098*** (0.013) 0.094*** (0.012) Cash/assets −0.033*** (0.005) −0.048*** (0.003) −0.048*** (0.003) −0.047*** (0.003) −0.046*** (0.003) EBITDA/assets 0.006 (0.004) 0.004 (0.004) 0.005 (0.004) 0.005 (0.004) 0.002 (0.003) Leverage 0.109*** (0.013) 0.109*** (0.012) 0.109*** (0.011) 0.107*** (0.012) 0.099*** (0.011) Probability of default −0.040*** (0.006) −0.047*** (0.006) −0.047*** (0.006) −0.047*** (0.006) −0.040*** (0.006) Zombie −0.036*** (0.006) −0.027*** (0.005) −0.027*** (0.005) −0.026*** (0.005) Excess mortality −0.053*** (0.004) Negative GVA −0.054*** (0.009) Fixed effects Region YES NO YES NO NO Sector NO YES YES NO NO Region-Sector NO NO NO YES YES Number of observations 252,887 252,887 252,887 252,887 252,887 Adjusted R20.202 0.228 0.239 0.248 0.250 356 M.Mateus, K.Neugebauer Table 5 State-guaranteed loans (firm-level data) – deciles. This regression table presents estimation results obtained by using a Linear Probability Model. The dependent variable is a dummy that is equal to one for eligible firms that received a governmentguaranteed loan during the Covid-19 pandemic, and zero for eligible firms that did not. Variables have been normalised to have a mean of zero and a standard deviation of one. The first column of this table corresponds to the fourth column of Table4, our baseline specification. Standard errors are clustered at the sector level. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively Dependent variable State-guarantee dummy (1) (2) (3) Previous credit relation 0.176*** 0.174*** 0.174*** (0.008) (0.009) (0.008) Previous state guarantee 0.190*** 0.188*** 0.188*** (0.007) (0.006) (0.006) Assets 0.098*** 0.099*** 0.098*** (0.013) (0.013) (0.012) Cash/assets −0.047*** −0.045*** (0.003) (0.003) Cash/assets—2nd decile 0.037*** (0.005) Cash/assets—3rd decile 0.041*** (0.004) Cash/assets—4th decile 0.036*** (0.004) Cash/assets—5th decile 0.025*** (0.004) Cash/assets—6th decile 0.016*** (0.004) Cash/assets—7th decile 0.004 (0.004) Cash/assets—8th decile −0.002 (0.004) Cash/assets—9th decile −0.012** (0.004) Cash/assets—10th decile −0.010*** (0.003) EBITDA/assets 0.005 0.005 −0.003* (0.004) (0.004) (0.001) Leverage 0.107*** 0.105*** (0.012) (0.011) Leverage—2nd decile 0.008** (0.003) Leverage—3rd decile 0.020*** (0.004) Leverage—4th decile 0.034*** (0.004) Leverage—5th decile 0.046*** (0.005) Leverage—6th decile 0.064*** (0.005) Leverage—7th decile 0.078*** (0.006) 357 Stayin’ alive? Government support measures inPortugal during… process for a guarantee before would be expected to find it easier to apply for a new loan with a state guarantee during the pandemic. Looking at the other firm-level variables, the size of the firm has also played an important role in getting a state-guaranteed loan. The respective coefficient enters with a positive sign and is highly statistically significant, indicating that larger firms (that fulfilled the eligibility criteria) were more likely to receive a stateguaranteed loan than smaller firms. Firms with larger cash holdings were less likely to receive a loan with a state guarantee, while more leveraged firms were more likely to have participated in this programme. These results are also in line with economic intuition. On the one hand, firms with more cash at hand are less likely to need further financing. In fact, when we look at the deciles of the cash-toassets distribution (Table5), we observe it is not monotonic, i.e., the negative relation between cashto-assets and state guarantees is only visible for firms with a high level of liquidity (cash-to-assets ratio above the 7th decile). The level of capitalisation and liquidity of Portuguese SMEs has increased significantly since the European sovereign debt crisis, mostly through retained earnings. This trend has continued during the pandemic, thus possibly reducing the need for external funding. On the other hand, most indebted firms are more likely to be liquidity constrained due to the frequent need to refinance and pay back their debt. Indeed, the relation between leverage and the take-up of state guarantees seems to be monotonic, i.e., an increase in the decile of leverage is associated with a higher probability of accessing a state-guaranteed loan. This holds true up until the 8th decile, after which the effect levels out. Table 5 (continued) Dependent variable State-guarantee dummy (1) (2) (3) Leverage—8th decile 0.086*** (0.007) Leverage—9th decile 0.082*** (0.008) Leverage—10th decile 0.085*** (0.011) Probability of default −0.047*** −0.044*** −0.045*** (0.006) (0.006) (0.007) Zombie −0.026*** −0.026*** −0.028*** (0.005) (0.004) (0.004) Fixed effects Region NO NO NO Sector NO NO NO Region-Sector YES YES YES Number of observations 252,887 252,887 252,887 Adjusted R20.248 0.250 0.247 358 M.Mateus, K.Neugebauer Lastly—and most importantly in the context of our analysis—we look at the creditworthiness of the firms receiving state-guaranteed loans using two different indicators: a firm’s probability of default and the fuzzy zombie indicator. Table4 shows that both credit-risk-related coefficients are negative and highly statistically significant, indicating that higher-risk firms, i.e., firms with a higher probability of default and firms with a higher level of zombieness, were less likely to receive a state-guaranteed loan. It should be noted that although both variables are used as proxies for the creditworthiness of a firm, the probability of default and the level of zombieness of a firm are not quite the same. In fact, both variables are only very weakly correlated in our sample. Table 6 State-guaranteed loans (firm-level data)—firm size classes. This regression table presents estimation results obtained by using a Linear Probability Model. The dependent variable is a dummy that is equal to one for eligible firms that received a government-guaranteed loan during the Covid-19 pandemic, and zero for eligible firms that did not. Size categories are defined according to the Commission Recommendation 2003/361/EC of 6 May 2003. Micro firms: number of employees < 10, turnover and/ or annual balance-sheet total ≤ 2 million euros. Small firms: number of employees < 50, turnover and/ or annual balancesheet total not ≤ 10 million euros. Medium-sized firms: number of employees < 250, turnover ≤ 50 million euros and/or annual balance-sheet total ≤ 43 million euros. Large corporations: remaining cases. Variables have been normalised to have a mean of zero and a standard deviation of one. The first column of this table corresponds to the fourth column of Table4, our baseline specification. Standard errors are clustered at the sector level. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively Dependent variable State-guarantee dummy All firms Mirco Small Medium Large Previous credit relation 0.176*** (0.008) 0.187*** (0.01) 0.139*** (0.006) 0.121*** (0.015) 0.086*** (0.003) Previous state guarantee 0.190*** (0.007) 0.157*** (0.006) 0.194*** (0.007) 0.256*** (0.02) 0.054 (0.048) Assets 0.098*** (0.013) 0.042*** (0.01) 0.056*** (0.013) −0.003 (0.025) −0.073*** (0.005) Cash/assets −0.047*** (0.003) −0.046*** (0.003) −0.091*** (0.007) −0.075*** (0.015) −0.019*** (0.002) EBITDA/assets 0.005 (0.004) 0.006 (0.004) 0.010* (0.005) −0.026 (0.019) −0.005*** (0.001) Leverage 0.107*** (0.012) 0.100*** (0.012) 0.134*** (0.007) 0.066*** (0.017) 0.034*** (0.003) Probability of default −0.047*** (0.006) −0.032*** (0.005) −0.034*** (0.006) −0.038* (0.015) −0.031*** (0.007) Zombie −0.026*** (0.005) −0.039*** (0.004) 0.010 (0.007) 0.022 (0.018) −0.010** (0.003) Fixed effects Region NO NO NO NO NO Sector NO NO NO NO NO Region-Sector YES YES YES YES YES Number of observations 252,887 211,701 32,519 4,607 436 Adjusted R20.248 0.188 0.242 0.335 0.88 365 Stayin’ alive? Government support measures inPortugal during… Table 8 State-guaranteed loans (loan-level data) – Loan amount / total assets. This regression table presents estimation results of regressions, where the dependent variable in Panel A is the loan amount of loans with a state guarantee, scaled by the pre-Covid total asset of the firm. Panel B presents the results for loans without a state guarantee. Variables have been normalised to have a mean of zero and a standard deviation of one. Standard errors are clustered at the sector level. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively Panel A Dependent variable Loan/assets of loans with a state guarantee (1) (2) (3) (4) (5) (6) Previous credit relation −0.075*** (0.006) −0.196*** (0.009) −0.075*** (0.006) −0.075*** (0.006) −0.076*** (0.006) −0.075*** (0.006) Previous state guarantee −0.041*** (0.005) −0.184*** (0.012) −0.038*** (0.005) −0.028*** (0.004) −0.025*** (0.004) −0.028*** (0.004) Assets −0.557*** (0.021) −0.563*** (0.021) −0.613*** (0.022) −0.613*** (0.022) −0.618*** (0.023) Cash/assets 0.090*** (0.006) 0.089*** (0.006) 0.068*** (0.005) 0.068*** (0.005) 0.072*** (0.005) EBITDA/assets 0.035*** (0.01) 0.036*** (0.01) 0.026** (0.01) 0.027** (0.01) 0.032** (0.01) Leverage 0.099*** (0.008) 0.100*** (0.008) 0.094*** (0.007) 0.093*** (0.007) 0.090*** (0.007) Probability of default −0.067*** (0.01) −0.067*** (0.009) −0.079*** (0.009) −0.079*** (0.009) −0.082*** (0.009) Zombie −0.045*** (0.006) −0.046*** (0.006) −0.038*** (0.005) −0.038*** (0.005) Loan maturity (years) 0.077*** (0.006) 0.123*** (0.007) 0.073*** (0.006) 0.087*** (0.005) 0.088*** (0.005) 0.087*** (0.005) Guarantee (other than state) −0.015*** (0.004) −0.012* (0.006) −0.015** (0.005) −0.009* (0.004) −0.005 (0.004) −0.005 (0.004) Bank assets −0.025*** (0.007) 0.022*** (0.005) 0.029*** (0.005) Bank own funds ratio −0.022** (0.008) 0.011 (0.006) 0.009 (0.005) Bank NPL ratio −0.021 (0.013) 0.001 (0.009) 0.009 (0.008) Bank ROA 0.021 (0.013) −0.032*** (0.009) −0.019* (0.008) Interbank funding / Assets −0.021** (0.007) −0.009 (0.006) −0.003 (0.005) Negative GVA −0.007 (0.004) Fixed effects Region YES YES YES NO NO NO Sector NO NO NO NO NO NO Region-Sector NO NO NO YES YES YES Bank NO NO NO NO YES YES Number of observations 60,032 60,032 60,032 60,032 60,032 60,032 Adjusted R20.45 0.141 0.452 0.485 0.486 0.485 366 M.Mateus, K.Neugebauer Table 8 (continued) Panel B Dependent variable Loan/assets of loans without a state guarantee (7) (8) (9) (10) (11) (12) Previous credit relation −0.164*** (0.006) −0.227*** (0.008) −0.162*** (0.006) −0.143*** (0.006) −0.144*** (0.006) −0.143*** (0.006) Previous state guarantee −0.060*** (0.008) −0.113*** (0.011) −0.061*** (0.008) −0.050*** (0.009) −0.045*** (0.009) −0.044*** (0.009) Assets −0.325*** (0.018) −0.329*** (0.017) −0.419*** (0.031) −0.431*** (0.031) −0.434*** (0.031) Cash/assets 0.063*** (0.007) 0.063*** (0.006) 0.054*** (0.006) 0.054*** (0.006) 0.057*** (0.006) EBITDA/assets 0.041** (0.014) 0.042** (0.014) 0.037** (0.014) 0.037* (0.015) 0.043** (0.016) Leverage 0.041*** (0.008) 0.042*** (0.008) 0.035*** (0.009) 0.034*** (0.009) 0.036*** (0.009) Probability of default −0.033*** (0.008) −0.031*** (0.008) −0.035*** (0.01) −0.037*** (0.011) −0.041*** (0.011) Zombie −0.017* (0.007) −0.017** (0.007) −0.015 (0.008) −0.018* (0.007) Loan maturity (years) 0.330*** (0.013) 0.365*** (0.014) 0.330*** (0.013) 0.332*** (0.015) 0.339*** (0.016) 0.339*** (0.016) Guarantee (other than state) 0.049*** (0.007) 0.121*** (0.008) 0.054*** (0.007) 0.045*** (0.006) 0.045*** (0.006) 0.045*** (0.006) Bank assets 0.019* (0.008) 0.006 (0.01) 0.02 (0.012) Bank own funds ratio 0.005 (0.004) 0.028* (0.011) 0.023* (0.01) Bank NPL ratio −0.005 (0.006) −0.026* (0.011) −0.015 (0.012) Bank ROA 0.001 (0.008) −0.048*** (0.008) −0.039*** (0.008) Interbank funding / Assets −0.042*** (0.012) 0.000 (0.016) −0.013 (0.02) Negative GVA 0.023*** (0.006) Fixed effects Region YES YES YES NO NO NO Sector NO NO NO NO NO NO Region-Sector NO NO NO YES YES YES Bank NO NO NO NO YES YES Number of observations 229,003 229,003 229,003 229,003 229,003 229,003 Adjusted R20.415 0.311 0.416 0.479 0.487 0.487 367 Stayin’ alive? Government support measures inPortugal during… moratorium as dependent variable. The control group consists of firms whose loans were eligible for the moratorium but which decided not to make use of it. We include the same firm-level explanatory variables used in the previous specifications. We do not include bank-level variables in this setup, as access to the moratorium was not a choice of the banks but firms could simply opt for it if their loans were eligible. Results for the moratorium uptake are presented in Table 9. The estimates presented in column (1) indicate that being in one of the most affected sectors increases firms’ probability of accessing the moratorium, compared to firms operating in less affected sectors, by 11 percentage points, i.e., there is a 32% higher probability compared with the mean take-up rate of 35%. Interestingly, firms that benefitted from a state-guaranteed loan prior to Covid-19 pandemic were not only more likely to access the Covid-related state guarantees, as seen earlier, but they were also more likely to make use of the public moratorium. Table 9 Moratorium (firm-level data) – dummy. This regression table presents estimation results obtained by using a Linear Probability Model. The dependent variable is a dummy that is equal to one for eligible firms that went under the moratorium during the Covid-19 pandemic, and zero for eligible firms that did not. Variables have been normalised to have a mean of zero and a standard deviation of one. Standard errors are clustered at the sector level. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively Dependent variable Moratorium dummy (1) (2) (3) (4) Most affected sector 0.106*** (0.023) Previous state guarantee 0.188*** (0.005) 0.186*** (0.005) 0.186*** (0.005) 0.186*** (0.005) Assets 0.067*** (0.013) 0.104*** (0.006) 0.106*** (0.007) 0.108*** (0.007) Cash/assets −0.111*** (0.004) −0.112*** (0.004) −0.111*** (0.004) −0.116*** (0.004) EBITDA/assets 0.005** (0.002) 0.005** (0.002) 0.006*** (0.002) 0.005** (0.002) Leverage 0.002 (0.002) 0.005** (0.002) 0.005** (0.002) 0.005** (0.002) Probability of default 0.052*** (0.009) 0.047*** (0.005) 0.046*** (0.005) 0.064*** (0.005) Zombie 0.032*** (0.007) 0.036*** (0.004) 0.036*** (0.004) Negative GVA −0.027*** (0.005) Fixed effects Region YES YES NO NO Sector NO YES NO NO Region-Sector NO NO YES YES Number of observations 138,997 138,997 138,997 138,997 Adjusted R20.081 0.141 0.146 0.146 368 M.Mateus, K.Neugebauer Looking at the firm-level variables, we observe that larger firms were more likely to make use of the public moratorium, while firms with more cash at hand were less likely to do so. Finally, when we look at the risk of the firms with loans under the moratorium, we observe a significant difference relative to firms applying for stateguaranteed loans. Results in Table9 show that both quality-related coefficients are now positive, indicating that higher-risk firms, i.e., firms with a higher probability of default and firms with a higher level of zombieness, were more likely to apply for the moratorium. These results show that the moratorium has benefitted relatively riskier firms, in contrast with what was observed for state guarantees. The statistical significance and magnitude of the coefficients remain by and large unchanged when we consider the share of each firm’s bank loans under the moratorium as our dependent variable instead of the simple moratorium dummy variable (Table10). Table 10 Moratorium (firm-level data) – share. This regression table presents estimation results obtained from regressions where the dependent variable is the share of the loan amount of eligible firms that went under the moratorium during the Covid-19 pandemic. Variables have been normalised to have a mean of zero and a standard deviation of one. Standard errors are clustered at the sector level. *, ** and *** denote significance at the 10%, 5% and 1% level, respectively Dependent variable Moratorium dummy (1) (2) (3) (4) Most affected sector 0.112*** (0.025) Previous state guarantee 0.138*** (0.006) 0.141*** (0.005) 0.142*** (0.005) 0.142*** (0.005) Assets 0.025 (0.015) 0.062*** (0.006) 0.064*** (0.007) 0.065*** (0.006) Cash/assets −0.100*** (0.005) −0.100*** (0.004) −0.099*** (0.004) −0.103*** (0.004) EBITDA/assets 0.006*** (0.002) 0.006** (0.002) 0.007*** (0.002) 0.006*** (0.002) Leverage 0.001 (0.003) 0.003 (0.002) 0.003 (0.002) 0.004 (0.002) Probability of default 0.052*** (0.01) 0.047*** (0.005) 0.046*** (0.005) 0.058*** (0.005) Zombie 0.022** (0.007) 0.027*** (0.004) 0.028*** (0.004) Negative GVA −0.016** (0.005) Fixed effects Region YES YES NO NO Sector NO YES NO NO Region-Sector NO NO YES YES Number of observations 138,997 138,997 138,997 138,997 Adjusted R20.057 0.126 0.13 0.13 369 Stayin’ alive? Government support measures inPortugal during… 5 Conclusion Credit guarantees and credit moratoria were widely used worldwide to support businesses affected by the Covid-19 pandemic. While these measures have been key to stabilising the economy in the short-run, supporting struggling firms in a context of high levels of uncertainty and a sharp deterioration of economic agents’ confidence, the mediumto long-term impact of these measures remains an open question. Based on the Portuguese experience, we assess the risk profile of firms accessing the state guarantees and the moratorium. Our paper has three main findings. First, we find that guaranteed loans went mostly to firms operating in the sectors most severely hit by the pandemic and to firms that had a credit relation and/or benefitted from a state guarantee before the onset of the Covid-19 pandemic. Second, the Portuguese public guarantee scheme seems to have mainly supported lower-credit-risk firms, i.e., those with a lower probability of default and a lower level of zombieness. Our results contrast with the results obtained for Italy by Core and De Marco (2024), where riskier firms were more likely to participate in the Italian public guarantee scheme. Additionally, in the case of Portugal, riskier firms paid higher interest rates and obtained smaller guaranteed loans than more viable firms. Third, in contrast to the state guarantees, our results show that riskier firms were more likely to apply for the public moratorium. Overall, our results suggest that state-guaranteed loans were mostly granted to firms that had a lower level of risk before the onset of the Covid-19 pandemic. In this sense, the strict access requirements to the Portuguese public guarantee scheme seem to have mitigated a significant increase in riskier lending, while supporting those sectors most affected by the pandemic. On the other hand, our results also show that riskier firms have benefitted relatively more from the moratorium. Going forward, the increase in the level of zombiefication of Portuguese firms will not only depend on their pre-Covid risk level but also on their ability to recover from this shock now that the pandemic has subsided. Ultimately, an increase in the level of zombiefication of the Portuguese economy would have an impact on the banking sector as well, as these firms might default on their loans, thereby leading to an increase in the level of non-performing loans in the Portuguese banking sector. More research on that will be needed in the future. Appendix Calculating thequasi‑zombie indicator (based onMingarelli etal. (2022)) The fuzzy zombie definition can be formalised as follows. Let V(i) y be a vector such that V(i) y =(ROA (i) y ,NIR (i) y ,DSC (i) y −5 %) 370 M.Mateus, K.Neugebauer with ROA(i) y = return on assets of firm i in year y, NIR(i) y = net investment ratio of firm i in year y, DSC(i) y = debt service capacity of firm i in year y. In the binary setup by Storz etal. (2017), firm i is defined as a zombie if its ROA < 0, NIR < 0, and DSC < 5% for Y = 2 consecutive years. This can be written as a simple geometric mean: For the fuzzy version of the zombie, we define a kernel k(x) as where x ≡ median(x). The geometric mean thus becomes We adapt the Storz et. al. (2017) zombie definition in order to arrive at a continuous (fuzzy) definition, following the same reasoning of Mingarelli etal. (2022). As in Storz et. al. (2017), we exclude certain (structurally different) sectors: primary sector (NACE 01–09), financial sector (NACE 64–66), public administration (NACE 84), activities of households (NACE 97–98), extraterritorial organisations (NACE 99). Firms with negative total assets, negative liabilities or negative stock of capital12 were also excluded. Contrary to Storz et. al. (2017), we keep listed and large firms in the sample. Inactive firms were also dropped from our sample. A firm is considered inactive if: a) their assets and liabilities do not change in two consecutive years, b) firms’ EBITDA is equal to zero, and c) the firm is inactive in their last year in the sample. Finally, we interpolate values for a single missing year in between two years for which data exists (simple average) for assets, liabilities, EBITDA, stock of capital, cash holdings, equity and net income. When one of the indicators that are required to classify a firm as zombie is missing but one of the other indicators is available and fails the zombie threshold, we classify the firm as a non-zombie. If a firm is not a zombie in year t-1, we assume it is also not a zombie in year t (by definition) if it has no data available. When estimating the fuzzy zombie, the median of each variable (net investment, ROA and DSC) is being estimated based only on positive values ( x≡median(x) +). Z i,y=⎛ ⎜ ⎜ ⎝ Y−1 � w=0 � xy∈V(i) y 𝟙xy−w <0⎞ ⎟ ⎟ ⎠ 1 Y�V � k (x)=𝟙x<0+ x−x x 𝟙0≤x< x Z ∗ i,y=⎛ ⎜ ⎜ ⎝ Y−1 � w=0 � x y ∈V(i) y k(xy−w)⎞ ⎟ ⎟ ⎠ 1 Y�V � 12 Calculated as the book value of each firm’s tangible and intangible assets. 371 Stayin’ alive? Government support measures inPortugal during… This assumption differs from Mingarelli etal. (2022), who also consider the zeros to estimate the median. Our option is explained by the large number of firms in Portugal with zero net investment that would make the median of investment equal to 0 (if the zero was included in the calculation). The zeros were also excluded from the calculation of the median of other variables in order to be aligned with the treatment of investment. Acknowledgements We thank Ana Cristina Leal, Inês Drumond, Carlos Santos, Tiago Pinheiro, and two anonymous referees for helpful comments and suggestions. We are grateful to Luca Mingarelli, Jonas Wendelborn, Maciej Grodzicki, and Martina Spaggiari for sharing their expertise with us. We gratefully acknowledge the support from Banco de Portugal’s data centre BPLIM.The views expressed in this article are those of the authors and do not necessarily reflect the views of Banco de Portugal,theEurosystemor the European Systemic Risk Board. Any errors and mistakes are ours. Funding Open access funding provided by FCT|FCCN (b-on). Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/ licenses/by/4.0/. References Acharya VV, Eisert T, Eufinger C, Hirsch C (2019) Whatever It Takes: The Real Effects of Unconventional Monetary Policy. Rev Financ Stud 32(9):3366–3411 Acharya VV, Crosignani M, Eisert T, Eufinger C (2024) Zombie Credit and (Dis-)Inflation: Evidence from Europe. J Finance 79(3):1883–1929 Altavilla C, Ellul A, Pagano M, Polo A, Vlassopoulos T (2021) Loan Guarantees, Bank Lending and Credit Risk Reallocation. CSEF Working Papers 629, Centre for Studies in Economics and Finance (CSEF), University of Naples, Italy Antunes AR, Prego P, Gonçalves H (2016) Firm default probabilities revisited. Banco de Portugal Economic Studies 2(2):21–45 Azevedo N, Mateus M, Pina Á (2022) Bank credit allocation and productivity: stylised facts for Portugal. Stud Econ Financ 39(4):644–674 Banerjee RN, Hofmann B (2018) The rise of zombie firms: causes and consequences. BIS Quarterly Review September 2018:7–78 Banerjee R, Hofmann B (2020) Corporate zombies: anatomy and life cycle. BIS Working Papers 882, Bank for International Settlements Beck T, Klapper LF, Mendoza JC (2010) The typology of partial credit guarantee funds around the world. J Financ Stab 6(1):10–25 Berger AN, Udell GF (2006) A more complete conceptual framework for SME finance. J Bank Finance 30(11):2945–2966 Bighelli T, Lalinsky T, CompNet Data Providers (2021) COVID19 government support and productivity: micro-based cross-country evidence. The Competitiveness Research Network, Halle Institute for Economic Research.CompNet Policy Brief (14) Blattner L, Farinha L, Rebelo F (2023) When Losses Turn Into Loans: The Cost of Weak Banks. American Economic Review. 113(6):1600–1641 Bonfim D, Cerqueiro G, Degryse H, Ongena S (2023) On-site inspecting zombie lending. Manag Sci 69(5):2547–2567 372 M.Mateus, K.Neugebauer Caballero RJ, Hoshi T, Kashyap AK (2008) Zombie Lending and Depressed Restructuring in Japan. Am Econ Re 98(5):1943–1977 Cascarino G, Gallo R, Palazzo F, Sette E (2022) Public guarantees and credit additionality during the Covid-19 pandemic. Temi di discussione1369, Bank of Italy Core F, De Marco F (2024) Information Technology and Credit: Evidence from Public Guarantees. Manag Sci 70(9):6202–6219 Correia S (2017) reghdfe: Stata module for linear and instrumental-variable/GMM regression absorbing multiple levels of fixed effects. Statistical Software Components s457874, Boston College, Department of Economics Gale W (1990) Federal lending and the market for credit. J Public Econ 42(2):177–193 Gale WG (1991) Economic Effects of Federal Credit Programs. Am Econ Rev 81:133–152 Jiménez G, Laeven L, Martinez Miera D, Peydró JL (2024) Public guarantees, private banks’ incentives, and corporate outcomes: evidence from the COVID-19 crisis. Working Paper Series 2913, European Central Bank, Frankfurt/Main Khwaja AI, Mian A (2005) Do Lenders Favor Politically Connected Firms? Rent Provision in an Emerging Financial Market. Q J Econ 120(4):1371–1411 Mankiw NG (1986) The Allocation of Credit and Financial Collapse. Q J Econ 101(3):455–470 Mateus M, Pinheiro T (2021) Bank pricing of corporate loans. Banco de Portugal Economic Studies 7(2):21–43 McGowan MA, Andrews D, Millot V (2018) The walking dead? Zombie firms and productivity performance in OECD countries. Econ Policy 33(96):685–736 Mingarelli L, Ravanetti B, Shakir T, Wendelborn J (2022) Dawn of the (half) dead: the twisted world of zombie identification. Working Paper Series 2743, European Central Bank, Frankfurt/Main Schivardi F, Romano G (2020) A simple method to estimate firms liquidity needs during the COVID-19 crisis with an application to Italy. Covid Econ 35:51–69 Schmidt C, Schneider Y, Steffen S, Streitz D (2021) Capital misallocation and innovation. Available at: https:// doi. org/ 10. 2139/ ssrn. 34898 01 Schumpeter JA (1942) Capitalism, socialism, and democracy. Harper & Brothers, New York Storz M, Koetter M, Setzer R, Westphal A (2017) Do we want these two to tango? On zombie firms and stressed banks in Europe. ECB Working Paper 2104, European Central Bank, Frankfurt/Main Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.