Impact of an exogenous shock (COVID-19) on the performance of Portuguese wine firms: a structural break analysis

ABSTRACT The COVID-19 pandemic crisis was responsible for disruptions in most industries worldwide, being the wine industry a good example of how changes in the economic and social environment impacted firm performance. This research investigates whether such impacts represent a structural change to firms’ profitability and explore the internal sources of such change. We consider firm-level data from Portuguese wineries, from 2014 to 2020 and apply a series of structural break tests to the period. The results allow us to conclude that the entire profitability function suffered significant changes, being COVID-19 indeed a game-breaker for wineries’ profitability.


I. Introduction
The unexpected changes in the firm´s economic environment provoked by exogenous shocks have become a subject of examination in the firm´s performance literature (Miklian and Hoelscher 2022).Typically, the firm's response and adaptation during these cycles are related to uncertainty management capabilities, given the magnitude of hostile factors such as lower access to capital, demand and supply disruption, increased costs of production, economic losses, or mobility restrictions (Miklian and Hoelscher 2022).
The recent case of the COVID-19 outbreak in late 2019 is an illustration of a major exogenous shock which has intensified deep fault lines in trade performance within the context of global value chains with implications on both demand and supply sides (OECD 2021).Such circumstances might have caused a structural change with a deep modification of trends and patterns of associations among firm variables (e.g.exports).
One globalized competitive industry that was severely impacted by the effects of COVID-19 was the wine industry, with producers in different countries affected differently (Niklas et al. 2022).This research focuses on Portugal´s case, which is amongst the leading wine economies and considered an 'emerging' Old World country whose production is lower than the historical Old World countries (Italy, France and Spain) but whose exports are increasing (Niklas et al. 2022).This study examines the financial performance of Portuguese wineries, which suffered a severe decline in sales of more than 236 million euros (−22.5%)through the ontrade-sector (OTS) and direct-to-consumer (DTC) wine sales channels (mostly reliant on tourism), and conversely experienced an increase in exports by more than 3% in volume and 5.3% in value.
Given the aforementioned ambiguity concerning the effect of COVID-19 on the economic structure of Portuguese wine firms, this research sets out to investigate whether wineries´ performance had a break date at the beginning of 2020.This will allow us to determine if 2020 was a structurally different year from the previous ones.To investigate the presence of a structural break we apply a new approach based on Karavias, Narayan, and Westerlund (2022) method, used on structural break testing in panel data frameworks, in which the authors propose a toolbox that follows the Bai and Perron (1998) contributions.This method allows testing unknown and multiple structural breaks as well as recognizing the location of the breakpoint.Furthermore, this research assesses the origin of the structural changes, by performing a dominance analysis to determine the importance of the econometric function´s independent variables (Luchman 2021) that explain the performance of Portuguese wine firms.Based on the preceding goal and methods, this study answers the following research questions (RQ): RQ1: Did Portuguese wine firms' performance suffer a structural change in 2020?RQ2: If confirmed that firms experienced a structural change, which determinants had a significant influence?
Given that empirical studies on structural changes have largely focused on macro and industry-level data, there is a dearth of studies exploring the microeconomics of structural change (Jitaree and Lee 2022).This is a gap this study tries to fill by using firm-level data and looking into an industry that is comprised of an array of companies with a heterogeneous economic structure, adopting dissimilar business strategies, namely concerning their exports intensity (exposure to foreign markets) or complementary development of tourism as a value-added activity.This research makes a theoretical contribution to the extant literature on structural breaks, by assessing whether firms' performance was subject to a structural break due to COVID-19 and whether such change might have been contingent on different constraints stemming from the firm's resources as well as the industry structure.The remainder of the paper includes Section 2, which provides an overview of the data used and the econometric model; Section 3, which presents and discusses the econometric results; and Section 4, which provides the concluding remarks.

Data and econometric model
The starting point is to define an econometric function that explains the performance of Portuguese wine firms.Considering operation financial performance, traditional proxies such as the return on assets (ROA) or earnings before taxes, depreciations, and amortizations (EBITDA) are typically considered as performance indicators within the resource-based view (RBV) literature (Lonial and Carter 2015;Sharma et al. 2022).In this conceptual framework, the EBITDA is frequently used by companies as a financial operating net income measure (Bouwens, De Kok, and Verriest 2019;Xu and Liu 2021).
The determinants of the performance of Portuguese wine firms are considered through the RBV approach, in which the internal characteristics of a firm are the key factors that explain its ability to perform.Based on the literature (Faria et al. 2020), as well as the available dataset, the generic form of the performance function is as follows The explanatory variables are firm size, i.e. the number of employees, the capital structure, measured as the value for capital depreciations and amortizations; age, as the number of years a firm is active in the market; the debt structure, as the ration of indebtedness of a firm; marketing, i.e. the promotion costs employed; export intensity, as the share of exports in total turnover; a dummy for the presence of wine tourism services (such as tasting rooms, restaurants, and visiting programs); and government-supported subsidies represents the institutional support to a firm.
The data sources are the official fiscal reports.To ensure technological homogeneity and comparability between firms, all firms included correspond to those that produce and sell still or liquor wines.The sample covers all the Portuguese Demarcated Regions located in mainland Portugal.
The period ranges from 2014 to 2020, which is the longest with available data.The final dataset includes 241 wine firms, which represent nearly 25% of the active companies operating in the market, accounting for approximately 65% of the total sales value of the industry (Instituto da Vinha e do Vinho 2021), which adds up to 1687 observations over 7 years.The descriptive statistics are presented in Table 1.
The sample is characterized by a high level of heterogeneity, exemplified by EBITDA's dispersion, ranging from €-2 million to €20 million.It shows a variation coefficient of nearly 393%.This highlights the discrepancies in firm size and managerial approaches.A closer examination shows the tendency for high variation coefficients in the explanatory variables, which confirms different realities within the panel of wineries included.Additionally, exports account for a mean of 20% of the total turnover of the wineries, and 28% of the firms included in the dataset offer a type of wine tourism activity.

II. Results
Given that this research is based on longitudinal data, panel data techniques ought to be tested and applied.Thus, following Equation (1), we confirm the presence of panel data effects through the Breusch-Pagan LM test for random effects [Chi 2 (1) = 398.49,p-value = 0.0000].Secondly, the presence of groupwise heteroskedasticity was assessed, through the modified Wald test, with the results [(Chi 2 (241) = 1.8e + 31, p-value = 0.0000] rejecting the null hypothesis and therefore pointing to the presence of heteroskedastic errors.Consequently, panel data random effect regression with robust standard errors was performed accordingly, with the results presented in Table 2.
Generically, the RE estimation reveals that most of the explanatory variables are significant, which signals the robustness of the adopted econometric function.Particularly, size, capital, and marketing are considered positive determinants of performance, whereas debt and age are deemed with a negative impact on performance.
Departing from linear regression, the Chow-test splits the data into groups based on the breaking point.Then, an F-test is used to compare the coefficients of the groups before and after the breaking point.A rejection of the null hypothesis of equality between the parameters means that a structural break occurred.In this study, the breaking point will be the year 2020.
Table 3 shows the results of the structural break tests, namely the Chow test and the Karavias, Narayan, and Westerlund (2022) procedures.
From Table 3, the Chow-type test for structural breaks (Table 3) confirms that 2020 was indeed significantly different in terms of profitability function.This means that the economic crisis generated by COVID-19 altered firms' dynamics towards profitability, in line with previous research on structural breaks resulting from the pandemic's impacts on the Portuguese economy (Santos 2021).
The Karavias, Narayan, and Westerlund (2022) procedure was performed using the Stata toolbox developed by Ditzen, Karavias, and Westerlund (2021).To perform these tests, heteroskedastic errors were considered, due to the presence of   such phenomena in the linear regression.Moreover, as no-break variables, we considered age and the wine tourism dummy, due to their constant nature.Generically, the strategy was to first blindly test where the breakpoints occurred.
The SSR (1.56e +14 ) comparison scheme detected the existence of one breaking point, occurring in 2019.This confirms that the operation scenario changed for firms after the pandemic crisis.
The test for the structural break (5.42**) rejects the null hypothesis of no structural break, against the alternative hypothesis in favour of the existence of a break.This allows us to confirm that the econometric function that explains Portuguese wine firms' performance suffered structural changes caused during the COVID-19 pandemic.
To assess the origin of the structural changes, we performed a dominance analysis (Luchman 2021).This technique allows us to determine the relative importance of the independent variables in a model, and rank them.Dominance analysis is based on a computation of a series of nested models and a comparison of the fit statistic (R 2 ) of each possible model.Table 4 presents the results of the procedure.
The most important determinants of firm profitability are 'depreciations' and 'employees', as proxies for capital and labour.Despite the slight reinforcement of the importance of the 'depreciations' variable in 2020, the structure of the relative importance towards profitability remained the same as in the pre-pandemic period, except for the variable 'export intensity', which shows a structural change towards a greater effort to export given the decline of firms´ operating profit through the OTS and DTC wine sales channels.Therefore, this study´s outcomes highlight the less stringent approach of the Portuguese Government policies towards companies during COVID-19, allowing exports to carry on, which was different from other countries such as South Africa where wine exports were barred in the early stages of the pandemic (Davids, Vink, and Cloete 2022).It thus demonstrates the importance of maintaining stability in policy intervention and avoiding discretionary measures (International Monetary Fund 2021).On the other hand, the decline of sales on the OTS and DTC wine sales channels was similar to other countries due to distancing measures and strong restrictions on travel (Wittwer and Anderson 2021), which underlines the inherent complexity associated with a strategic extension of wineries into tourism.

III. Conclusions
In the aftermath of COVID-19, this paper analysed how Portuguese wineries' financial performance was affected, through the assessment of structural changes in profitability.Our research shows that the application of the Karavias method allowed recognition of the location of the breakpoint, proving that there was a structural change in the profitability function in 2020.Additionally, we observed that the overall function determinants did not exhibit a fundamental change.Therefore 2020 structural break in wine firms´ profitability was not affected by one particular variable, but rather the result of the transformation of the profitability function as a whole.This means that indeed COVID-19 was a game-changer for Portuguese wine firms' performance.
As a result, this research delivers practical contributions by demonstrating that a given heterogeneous industry, comprised of companies with dissimilar business structures, will eventually lead to different responses towards an all-globally disruptive effect and exogenous shock which calls for strong and stable government support.As the Portuguese case shows, the positive performance of wineries on exports, different from other countries, suggests the importance of policy stability rather than mercurial behaviour, to avoid unnecessary restrictions that would lead, namely to a high level of stock build-up and future devaluation of Portuguese wine.Also, the fact that OTS and DTC wine sales channels were disrupted in Portugal, similarly to other countries, warrants the need for policy support at an international level.Nonetheless, it also calls for business innovation to improve DTC solutions for delivery and online sales.Furthermore, this study adds new knowledge and extends the previously known implications of a health-related security threat, i.e. the COVID-19 pandemic, within the winery's business model by assessing its influence on the financial performance of firms.

Table 3 .
Tests for structural breaks.