LEXONOMICA Vol. 15, No. 2, pp. 189–212, December 2023 https://doi.org/10.18690/lexonomica.15.2.189-212.2023 CC-BY, text © Šergo, Gržinić, Sučić Čevra, 2023 This work is licensed under the Creative Commons Attribution 4.0 International License. This license allows reusers to distribute, remix, adapt, and build upon the material in any medium or format, so long as attribution is given to the creator. The license allows for commercial use. https://creativecommons.org/licenses/by/4.0 CHARTING THE COURSE: TOTAL FACTOR PRODUCTIVITY TRENDS IN CROATIA POST-PRE-BANKRUPTCY ACT Accepted 6. 11. 2023 Revised 20. 11. 2023 Published 6. 12. 2023 ZDRAVKO ŠERGO, 1 JASMINA GRŽINIĆ, 2 MIRELA SUČIĆ ČEVRA 3 1 Institute of Agriculture and Tourism, Department of Tourism Poreč, Croatia zdravko@iptpo.hr 2 Juraj Dobrila University of Pula, Department of Economics and Tourism, Pula, Croatia jasmina.grzinic@unipu.hr 3 Kuoni Tumlare, Zagreb, Croatia mirela.Sucic@tumlare.com CORRESPONDING AUTHOR zdravko@iptpo.hr Keywords total factor productivity, pre-bankruptcy act, synthetic control method, placebo test, Croatia Abstract The synthetic control method (SCM) is a valuable tool for unbiased pre-bankruptcy reform analysis in economic policy evaluations. This study utilizes SCM to assess the impact of the Financial Operations and Pre-Bankruptcy Settlement Act (AFOPBS) on Croatia's total factor productivity (TFP). Control units and weights were meticulously chosen to construct a synthetic control for Croatia, creating a counterfactual scenario for the reform's absence. The policy's impact was quantified by comparing TFP growth post-policy between Croatia and its synthetic control. Placebo tests confirmed the results' significance, and further validation was achieved through panel difference-in-differences analysis (PDID). Our findings show that the pre-bankruptcy reform in late 2012 effectively reduced the gap between Croatia and its synthetic control throughout the post-treatment years. However, it had two short-term adverse impacts and a subsequent recovery-like phase. These effects were statistically significant and confirmed by cross-validation. In conclusion, Croatia's pre-bankruptcy reform significantly influenced TFP volatility, highlighting SCM's effectiveness in evaluating economic policies, especially those crucial for economic growth. 190 LEXONOMICA. 1 Introduction The AFOPBS was implemented in the fourth quartal of 2012 during a challenging period for the Croatian economy, which was grappling with high levels of illiquidity. The Act aimed to address this issue by reorganizing companies with economic potential and manageable debt, with the goal of preserving sustainable jobs in struggling firms and mitigating the decline in aggregate employment. The legislation drew inspiration from foreign models while taking into account Croatian specificities, notably its timing during a severe and prolonged recession from 2009 to 2014, which resulted in a cumulative decline in GDP of 12.5%. The reform was seen as a vital component of the economic revival policy. The experiences of other countries facing stagnation have shown that unsuccessful reform of institutional rules surrounding bankruptcy laws can perpetuate economic crises. On the other hand, successful reform can lead to better reallocation and efficiency of labor and capital, contributing to improved productivity and reducing waste. Objectives of the AFOPBS were to change the payment habits of business entities, reduce indebtedness, and increase liquidity. Additionally, the Act aimed to facilitate the restructuring of entrepreneurs (companies) to enhance their liquidity and solvency, offering more favourable conditions for creditors' settlements compared to traditional bankruptcy proceedings. By enabling the transformation of debtors into viable and competitive entities, the Act sought to benefit creditors, employees, owners, and the overall Croatian economy. The reform was also driven by political goals, as the left-wing government sought to proactively prevent the spread of high unemployment and potential social unrest during the recessionary years. Offering failed companies or debtors a second chance through pre-bankruptcy settlements was seen as a means of sustaining their social mission while rehabilitating them with involvement from economic agents and various creditors. While this paper primarily delves into the causal impact of the AFOPBS on TFP in Croatia within the economic domain, it also briefly acknowledges criticisms from legal professionals regarding the Act's influence on Croatian insolvency legislation. Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 191. Notably, Bodul and Vuković (2015, 191-193) raise concerns about the Croatian bankruptcy law, including extended processing times compared to other countries in the region, heightened costs, diminished creditor satisfaction, and a pressing need for improvements. As this paper endeavours to chart the causal impact of a legislative act on economic outcomes in the post-adoption period, it is pertinent to provide a brief overview of these outcomes. TFP plays a pivotal role in shaping the economic landscape on a macro level. It represents the efficiency and effectiveness with which a country utilizes its inputs, such as labor and capital, to produce goods and services. TFP growth is a key driver of economic progress, as it reflects advancements in technology, innovation, and overall productivity. TFP is often considered a measure of economic health, as it directly impacts a nation's long-term economic growth potential. Higher TFP growth rates are associated with increased output per worker, which can lead to rising living standards and enhanced competitiveness in global markets. Moreover, TFP can be a critical factor in addressing economic challenges, such as mitigating the impact of labor force ageing or resource scarcity. It underscores the importance of fostering an environment conducive to innovation and efficiency gains, as these factors are fundamental to sustained economic development. To study the impact of the AFOPBS on TFP, the paper will use a quasi -experimental procedure to analyze the Croatian case. The SCM will be employed to compare the TFP growth rate some years before and after the reform while controlling for certain co-predictors of aggregate TFP. The analysis will involve selecting a pool of countries with similar characteristics that did not implement the pre-bankruptcy reform during the analysis period to serve as the control group. The robustness of the SCM results shall be confirmed through a placebo test and cross-validation, employing the PDID method. The remainder of this paper is structured as follows: the next section provides a summary of the literature on our topic and introduces the methodology. Following that, we delve into the data sources and variables utilized. Subsequently, the empirical analysis, along with the final results, is presented in the ensuing section, leading to the conclusion of this study. 192 LEXONOMICA. 2 Literature preview The reviewed literature comprises a blend of theoretical and empirical approaches, each contributing distinct insights into the relationship between bankruptcy laws and TFP growth. While some papers delve into theoretical frameworks to establish foundational principles, others employ empirical methodologies to quantify the effects in real-world scenarios. The intersection of bankruptcy laws, legal protection, and their impact on TFP constitutes a pivotal area of the following investigation. This research domain is characterized by a spectrum of methodologies aimed at uncovering the intricate relationships between legal frameworks, economic policies, and growth outcomes. Studies have delved into these dimensions using diverse analytical approaches. To study fluctuations in TFP, Tomura (2007: 1-10) presents a dynamic general equilibrium model that considers the interactions between credit market frictions, firm entry and exit, and TFP fluctuations in response to various shocks. In Jensen's seminal work (1986: 3-5), he presents a theoretical model that highlights how debt levels can incentivize managers to enhance firm performance, thus influencing TFP growth. His paper establishes a crucial link between financial decisions and productivity. Furthermore, in Lim and Hahn's research (2004: 10-12), the wider landscape of research on bankruptcy laws offers insights into the multifaceted connections between legal protection, economic policies, and growth in East Asia. In essence, the exploration of bankruptcy laws, legal protection, and their impact on TFP is a multidisciplinary endeavour that leverages various methodologies to deepen our understanding of the mechanisms underlying economic dynamics. Additionally, other works in the field have explored the impact of bankruptcy laws and legal protection on economic dynamics. For example, in the context of bankruptcy laws and firm performance, Bodul and Vuković (2015: 181-193) explore the effects of multiple reforms in the Croatian Bankruptcy Act. Using indicative methods, the study evaluates the changes in bankruptcy and preliminary bankruptcy procedures adopted in 2012 and their relative efficiency in comparison to other countries in the region. The authors highlight the need for further improvement in the Croatian regulatory framework, considering the challenges posed by external institutional factors. The empirical study by Misra (2019: 1-10) utilizes quantitative analysis to compute TFP growth for Indian states in distinct time periods. By examining actual Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 193. TFP data, the research provides empirical evidence of the impact of post-crisis periods on TFP growth. Employing an empirical approach, Köke's study (2001, 1- 10) investigates the relationship between financial pressure and productivity growth in German manufacturing firms. The research employs econometric techniques to analyze real-world data, offering insights into the impact of financial factors on productivity. The authors Gonçalves and Martins (2016: 1-10) used empirical analysis in their study, exploring determinants of TFP growth in the Portuguese manufacturing sector. The authors employ quantitative methods to identify factors affecting TFP and suggest policy implications. The paper by Yamaguchi (2022: 11- 20) adopted a Regression Discontinuity (RD) design; this study evaluates the impact of government interventions on the productivity of airlines. The empirical approach helps quantify the effects of interventions in a real-world context. The paper by Dvouletý, Srhoj, and Pantea (2021: 1-10) employs a systematic review of empirical evidence to assess the impact of public grants in the intervention context on SME performance. The authors critically evaluate various empirical studies, identifying patterns and heterogeneity in the effects of grants on different outcomes, including TFP. The research by Miao et al. (2022: 11-20) combines firm- level accounting data with data on state aid for restructuring. It uses treatment effects estimators to analyze the impact of aid on static and dynamic efficiency, contributing empirical insights into the consequences of state aid. In addition to the mentioned studies, the work by Sadeghi and Kibler (2022: 1-10) provides an empirical investigation into the relationship between bankruptcy legislation and entrepreneurship. Employing the Synthetic Control Method (SCM), the authors explore the effect of entrepreneur-friendly bankruptcy reform on entrepreneurial activity in Finland. The study contributes to the empirical understanding of how legal reforms impact entrepreneurial decisions and growth ambitions. The author, Neira (2017: 1-47), proposes that variations in bankruptcy procedures explain a positive relationship between aggregate productivity and the proportion of large firms; he uses a model to demonstrate that worsening bankruptcy procedures lead to lending shifts towards smaller firms, impacting TFP. The study by Aleksanyan and Huiban (2016: 89-108) shows that firm productivity predicts bankruptcy risk, with credit cost and productivity playing significant roles in determining the probability of bankruptcy. The study by Acemoglu and Guerrieri 194 LEXONOMICA. (2008: 467-498) analyzes a model with formal and informal sectors, indicating that countries with low debt enforcement and high formality costs have low allocative efficiency and reduced TFP. Other authors, Hiroki, Iwatsubo, and Watkins (2022: 1-36), find that Japanese manufacturers' firm-level TFP positively predicts their future stock returns, with intangible expenditure risks explaining much of the predictive power. The study by Tamayo (2017: 225-242) highlights that poorly designed bankruptcy arrangements can substantially increase financial constraints, negatively impacting aggregate output and TFP. On the other hand, Cosci and Meliciani (2002: 37-54) in their work investigate bankruptcy determinants, finding that firm inefficiency and qualitative factors such as customer concentration and competitors' strength predict bankruptcy, reflecting the importance of monitoring beyond balance sheet variables. The research by Albalate, Bel, and Mazaira-Font (2022: 549–584) identifies and examines the inhibitory impact of zombie enterprises on TFP growth of non-zombie enterprises, using comprehensive identification methods. The study by Xia and Wu (2021: 592–600) models the impact of credit market shocks, explaining the unusual increase in aggregate TFP during the 2008 financial crisis. The paper by Cespedes, Thakor, and Yang (2022: 1-30) explores the link between competition, non-financial enterprise financialization, and TFP, finding that financialization decreases TFP but is mitigated by strong competition. By exploiting a chapter of Bankruptcy Law eligibility using a regression discontinuity design, the research by authors reveals that enhanced bankruptcy protection improves ex ante outcomes, including investments and productivity. The seminal study by Di Martino, Latham, and Vasta (2020: 936-990) employs a comprehensive data set to investigate the evolution of bankruptcy law features in European economies from 1850 to 2015, revealing that while macroeconomic changes influenced the introduction of alternative procedures, national-level differences in change and bankruptcy features challenge the idea of legal system affiliation or economic development as sole determinants, indicating a complex interplay with state formation processes. Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 195. These studies, in summary collectively demonstrate the diverse range of methods employed to dissect the intricate connection between bankruptcy laws, economic policies, and their subsequent impacts on TFP growth. 3 Methodology and Data 3.1 Synthetic Control Method The SCM has been widely employed in academia for policy evaluation in work by Abadie and Gardeazabal (2003: 113-132), Abadie, Diamond, and Hainmueller (2010: 493-505), Abadie, Diamond, and Hainmueller (2015: 495-510), Sampaio (2014: 1- 14), Lin and Chen (2018: 734-750), Albalate, Bel, and Mazaira-Font (2021: 549-584), Xia and Wu (2021: 592-600), and Qi and Han (2021: 52431-52458), offering advantages over the Difference-in-Differences (DID) method by utilizing data- driven selection to construct a virtual control group most similar to the treated group. This method reduces subjective bias, addresses endogeneity concerns to some extent, and ensures evaluation reliability (Abadie, 2021: 391-425). In this study, S C M w a s a p p l i e d t o a s s e s s t h e i m p a c t o f t h e A F O P B S intervention on TFP trajectory in Croatia, utilizing a single treatment group and timing (AFOPBS implementation at the end of 2012) and constructing a control group using economic counterparts from other countries. This research centres on a single treatment nation, specifically the Croatian economy, during a particular treatment period, which corresponds to the year 2013. This period is chosen as it aligns with the initiation of the first financial settlements, occurring with a relatively short time lag among debtors and creditors in our analysis. The SCM enables the estimation of the impact of a specific event, referred to as the "treatment" even when a counterfactual scenario for the treated country is unobservable. The focal point of our study is the enactment of the pre-bankruptcy legislation, which we aim to analyze. To achieve this, counterfactual TFP values are simulated for Croatia (central to our research) using the SCM, illustrating potential deviations in the TFP trajectory if the country had not implemented the pre- bankruptcy act. By comparing the actual and counterfactual series, we assess the measurable causal effect of implementing the AFOPBS on TFP. 196 LEXONOMICA. In this study, the treatments are represented by pre-bankruptcy settlements that trigger a distinct transition toward business restructuring. By leveraging a weighted average of TFP data from diverse economies across Europe, characterized by transparent bankruptcy regulations that underwent gradual reform, this approach facilitates comparison of TFP before and after bankruptcy reform in the treated country (Sadeghi and Kibler, 2022: 5). We have categorized the years from 2008 to 2012 as the pre-policy period, while the years spanning from 2013 to 2019 constitute the post-policy period. This distinction is rooted in the fact that the AFOPBS reform was formally announced in Croatia on October 10, 2012, and subsequently implemented across the entire nation after a short time lag. The weights assigned to the control countries (referred to as "donors") are selected to ensure that the synthetic control exhibits characteristics that closely resemble the TFP patterns of the treated country before the modification in bankruptcy laws. Through the SCM, the disparity between the feature vectors related to outcome predictors of the treated nation and those of the synthetic control is minimized prior to the implementation of the treatment. Indeed, the weight calculations were derived by minimizing a distance matrix to ensure that the synthetic control closely mirrored Croatia's attributes in the pre-intervention timeframe. Subsequently, the impact of the policy for each time point in the post-intervention period was estimated by contrasting Croatia's observed outcomes with those generated by the constructed "synthetic control". The timeline of SCM implementation, along with clarification steps, proceeded as follows: The assumption was made that pertinent TFP data for 𝐾𝐾 + 1 countries were gathered statistically during the period 𝑡𝑡 ∈ [1, 𝑇𝑇 ], with country 𝑖𝑖 (Croatia) being the subject of the AFOPBS intervention at 𝑇𝑇 0 � 1 ≤ 𝑇𝑇 0 ≤ 𝑇𝑇 � , forming the experimental group. The remaining 𝐾𝐾 countries constituted the control pool, encompassing nations that did not implement the pre-bankruptcy settlement. At time 𝑡𝑡 , 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑌𝑌 signifies the TFP values influenced by the intervention for country 𝑖𝑖 , whereas 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 represents the TFP values for country 𝑖𝑖 that remain unaffected by the pertinent mechanism. Let 𝑎𝑎 𝑖𝑖𝑖𝑖 = 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑌𝑌 − 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 denote the alteration in TFP values for country 𝑖𝑖 , which is subject to the intervention policy, at time 𝑡𝑡 . 𝐷𝐷 𝑖𝑖𝑖𝑖 is a binary variable indicating whether the country is implementing the "new legislation that addresses bankruptcy issues" at time 𝑡𝑡 . This variable takes the value of 1 if country Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 197. 𝑖𝑖 adopts the legislation at time 𝑡𝑡 , and 0 otherwise. The TFP growth rate of country 𝑖𝑖 at time 𝑡𝑡 is represented as 𝑃𝑃 𝑖𝑖𝑖𝑖 = 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 + 𝐷𝐷 𝑖𝑖𝑖𝑖 𝑎𝑎 𝑖𝑖𝑖𝑖 . For the control group throughout this period, 𝑃𝑃 𝑖𝑖𝑖𝑖 remains as 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 . However, for the experimental group, 𝑎𝑎 𝑖𝑖𝑖𝑖 = 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑌𝑌 − 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 = 𝑃𝑃 𝑖𝑖𝑖𝑖 − 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 . In this study, the change in TFP values influenced by the policy intervention is denoted as 𝑎𝑎 𝑖𝑖𝑖𝑖 . Here, 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑌𝑌 represents the observable TFP values influenced by the "pre-bankruptcy law" (AFOPBS). While 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 is not directly available, it can be estimated using the factor model proposed by (Xia and Wu, 2021: 1-10). The specific calculation formula is as follows: 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 = 𝛿𝛿 𝑖𝑖 + 𝜃𝜃 𝑖𝑖 𝑍𝑍 𝑖𝑖 + 𝜆𝜆 𝑖𝑖 𝜇𝜇 𝑖𝑖 + 𝜀𝜀 𝑖𝑖𝑖𝑖 (1) where 𝛿𝛿 𝑖𝑖 is the time fixed effect and 𝑍𝑍 𝑖𝑖 is the observed r × 1 dimension control variable, which is not influenced by the implemented policy. In this study, the various data are selected as control variables. The rationale for selecting those variables will be clarified later on. 𝜃𝜃 𝑖𝑖 is the 1 × r dimension unknown parameter vector, 𝜆𝜆 𝑖𝑖 is the common factor vector of the 1 × F dimension that cannot be observed, 𝜇𝜇 𝑖𝑖 is the country fixed effect of the F × 1 dimension, and 𝜀𝜀 𝑖𝑖𝑖𝑖 is the short- term shock that cannot be predicted and has an average value of 0. Under general conditions, ∑ 𝑤𝑤 𝑘𝑘 𝑃𝑃 𝑘𝑘 𝑖𝑖 𝐾𝐾 + 1 𝑘𝑘 = 2 can be used as an unbiased estimate of 𝑃𝑃 𝑖𝑖𝑖𝑖 𝑁𝑁 if the period before the policy intervention is longer than the period after policy implementation. Here, 𝑤𝑤 𝑘𝑘 represents the country specific weight in the SCM. Finally, the estimated 𝑎𝑎 1 𝑖𝑖 of the policy influence effect is calculated as follows: 𝑎𝑎 𝑖𝑖𝑖𝑖 ∧ = 𝑃𝑃 1 𝑖𝑖 − ∑ 𝑤𝑤 𝑘𝑘 𝑃𝑃 𝑘𝑘 𝑖𝑖 𝐾𝐾 + 1 𝑘𝑘 = 2 , 𝑡𝑡 ∈ � 𝑇𝑇 0 + 1, … 𝑇𝑇 � (2) Let 𝑍𝑍 1 represent a k × 1 vector containing the pretreatment characteristics' values of the treated unit, while 𝑍𝑍 0 is a k × J matrix comprising the values of the same characteristics for the donor units. The optimal weights 𝑊𝑊 ∗ minimize the following expression: ‖ 𝑍𝑍 1 − 𝑍𝑍 0 𝑊𝑊 ‖ = �( 𝑍𝑍 1 − 𝑍𝑍 0 𝑊𝑊 ) ′ 𝑉𝑉 ( 𝑍𝑍 1 − 𝑍𝑍 0 𝑊𝑊 ) (3) 198 LEXONOMICA. where 𝑉𝑉 is a k × k symmetric and positive semi -definite matrix, and it reflects the relative significance of each predictor. In our analysis, we employed the Root Mean Square Prediction Error (RMSPE) as a measure to quantify the disparity between the actual outcome of a country and the outcome predicted by its synthetic control counterpart. The RMSPE calculation is as follows: 𝑅𝑅 𝑅𝑅𝑅𝑅 𝑃𝑃 𝑅𝑅 = � 1 𝑇𝑇 ∑ � 𝑃𝑃 1 𝑖𝑖 − ∑ 𝑤𝑤 𝑘𝑘 𝐾𝐾 + 1 𝑘𝑘 = 2 𝑃𝑃 𝑘𝑘𝑖𝑖 � 𝑇𝑇 𝑖𝑖 = 1 2 (4) Where 𝑃𝑃 1 𝑖𝑖 represents the actual value of the experimental group, 𝑤𝑤 𝑘𝑘 stands for the weight, 𝑃𝑃 𝑘𝑘𝑖𝑖 signifies the value of the control group, and T denotes the number of time periods within a specific year. The early fit of the SCM, we emphasize, is crucial. An effective early fit (if occurs) indicates that the constructed synthetic control closely mirrors the pre-policy outcomes of the treated unit (or country). When the early fit is g ood, t he diffe re nce in R MSPE values meas ured after an d befo re the implementation of the policy serves as a reliable measure of the extent of influence exerted by the AFOPBS intervention. This difference in RMSPE provides valuable insights into the accuracy of the synthetic control model in replicating the actual post-policy outcomes of the treated unit, thereby shedding light on the effectiveness of the AFOPBS policy intervention. Conversely, if the RMSPE indicates suboptimal performance, we will initiate a restructuring process by forming a new control pool consisting of countries that exhibit relatively high weights assigned to the previously set of donors. As of our last knowledge update in September 2021, there isn't a specific R package that directly provides an automated method for composing optimal donor weights to minimize the MSPE in the context of synthetic control methods. The process of finding optimal donor weights involves a try and error procedure. The "Synth" package in R provides a comprehensive set of tools for implementing synthetic control methods, but it might not directly offer a pre-built function for automatically composing optimal donor weights. The optimization process is often part of the synthesis step itself. Hence, we create the code that will loop through the different permutations of control sets, create dataprep objects for each scenario, perform Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 199. synthesis, and store the MSPE values for each scenario. Finally, we will get MPSE values for all scenarios and within it one with optimal performance. 3.2 Placebo Tests To assess the statistical significance of the AFOPBS intervention effect on TFP, we conducted placebo experiments through an iterative process of reassigning the treatment status to individual control units. For each of these "placebo-treated" units, we recalculated the treatment effect using the synthetic control method and then compared the resulting estimated treatment effect to a distribution of placebo effects (Kreif et al., 2016: 1519). Specifically, we analyzed the proportion of placebo effects that exhibited at least as extreme absolute values as the estimated treatment effect for the country under investigation, such as Croatia. This examination allowed us to determine whether the impact of financial reform in Croatia exceeded what would be expected if the reform had been "randomly assigned" to another country. 3.3 Data Source The data utilized in this study were sourced from publicly available country-level datasets, specifically the latest version 10.0 of the Penn World Table (GGDC, 2021) and the World Bank (World Bank, 2023). As a result, data were retrieved from two distinct online platforms. Pertaining to the outcome variable growth rate (real TFP - RTFPNA), information from the first source was employed spanning the years 2000 to 2019. For the covariates linked to each country and deemed relevant to the outcomes, we relied on data extracted from the World Development Indicators dataset. 3.4 Control Variables To construct the synthetic controls, we included variables known from prior research to encapsulate the confounding factors that are likely to affect TFP. Table 1 presents these variables, which encompass a wide range of economic and demographic factors recognized for their influence on TFP. 200 LEXONOMICA. 3.4 Control Pool In configuring the donor pool, our selection targeted countries sharing Croatia's level of development, predominantly encompassing several former socialists as well as other nations within the European Union (EU). The set of 10 potential control countries, a noteworthy highlight, represents a sampling of economies that did not adopt the Pre-Bankruptcy Settlement Act during the analyzed time window. The donor countries include Austria, Bulgaria, Estonia, Spain, Greece, Hungary, Lithuania, Latvia, Poland, and Slovenia. Additionally, the one country subjected to intervention was treated as the focal point of the investigation, resulting in a final sample of 11 countries. Our intention was to assemble a highly varied assortment of nations within the donor pool, aiming to comprehensively explore the mediated influence on TFP brought about by the AFOPBS in this analysis. This perspective is closely aligned with the "convergence theory" (Barro, 1998: 10), which posits that less developed countries are likely to experience faster growth rates during transition dynamics, contributing to a catching-up process with advanced economies. It is well recognized that long-term growth is substantially spurred by TFP advancements. The selected "control pool" is consistent with the proposed co-predictors of TFP in the SCM, particularly in light of the hypothesis suggesting that the proliferation of the service economy may have contributed to weaker productivity growth compared to growth driven by manufacturing. Within the neoclassical growth framework, the equation for growth underscores technological progress translating into TFP advancements, acting as a pivotal driver of long-term economic expansion. The Solow residual, characterized by Solow as the "manna from heaven," is far from being a mere accounting artefact; rather, it plays a fundamentally important role in explaining the trajectory of real GDP per capita growth (Vollrath, 2019: 39). Recognizing the significance as well as the challenges inherent in measuring it accurately, and acknowledging the plethora of potential contextual variables that can be drawn from endogenous growth empirics, we focus on variables (such as production, capital formation, employment distribution between service and Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 201. manufacturing sectors, human capital, openness, labor economy issues) that might contribute to the volatile nature of TFP growth. These factors collectively constitute a balanced selection of potential variables that could serve as co-predictors for TFP. Our approach is underpinned by the notion put forth by Ferranti (2012: 132-135), asserting that the demographic transition toward a services sector, along with the contemporary TFP growth observed in our dataset, were predominantly influenced by the following factors: the process of deindustrialization coupled with the expansion of the service industry, the role of physical infrastructure and employment as fundamental components in the production function, and the concept of output per capita as a catalyst for perpetuating technological change and enhancing productivity through the principle of "learning by doing." 3.6 Cross-Validation For the purpose of cross-validating the estimates of the pre-bankruptcy reform effect derived from the SCM, an additional analysis was conducted the PDID method. The specific steps of the PDID analysis are outlined as follows: 𝑌𝑌 𝑖𝑖𝑖𝑖 = 𝛽𝛽 0 + 𝛽𝛽 1 𝑇𝑇 𝑇𝑇𝑇𝑇 𝑎𝑎 𝑡𝑡 𝑖𝑖𝑖𝑖 𝑃𝑃 𝑃𝑃 𝑃𝑃 𝑡𝑡 𝑖𝑖𝑖𝑖 + ∑ 𝛾𝛾 𝑘𝑘 10 𝑘𝑘 = 1 𝑍𝑍 𝑖𝑖𝑖𝑖 𝑘𝑘 + 𝑢𝑢 𝑖𝑖 + 𝜏𝜏 𝑖𝑖 + 𝜀𝜀 𝑖𝑖𝑖𝑖 (5) Here, 𝑌𝑌 𝑖𝑖𝑖𝑖 represents the TFP outcome variable for Croatia at time t, and both 𝑇𝑇𝑇𝑇𝑇𝑇 𝑎𝑎 𝑡𝑡 𝑖𝑖𝑖𝑖 and 𝑃𝑃 𝑃𝑃 𝑃𝑃 𝑡𝑡 𝑖𝑖𝑖𝑖 are dummy variables specific to country i and time t. The coefficient 𝛽𝛽 1 of the interaction term 𝑇𝑇𝑇𝑇𝑇𝑇 𝑎𝑎 𝑡𝑡 𝑖𝑖𝑖𝑖 × 𝑃𝑃 𝑃𝑃 𝑃𝑃 𝑡𝑡 𝑖𝑖𝑖𝑖 captures the effect of the pre-bankruptcy act on TFP. Additionally, 𝑍𝑍 𝑖𝑖𝑖𝑖 𝑘𝑘 (where k=1,2,...,10) represents the control variables, 𝑢𝑢 𝑖𝑖 is the individual fixed effect term, and 𝜏𝜏 𝑖𝑖 is the time-fixed effect term. For our analysis, we continue to consider Croatia as the treatment unit, with the remaining countries serving as the reference group. 202 LEXONOMICA. 4 Results 4.1 Baseline Results In the initial phase, adhering to a "general to specific" approach, we embarked on the task of optimizing the composition of the control pool, which consists of 10 countries with features outlined in the control pool section within the methodology chapter. In order to capture the diverse array of economic, social, and policy factors that may influence the trajectory of our treated unit, we set a minimum threshold of 10 countries in our control pool. This approach was motivated by the need to provide a rich set of potential matches for our treated unit, contributing to the credibility of our results. We systematically generated permutations by iteratively excluding one donor country at a time from the sorted dataset. We then proceeded to evaluate the various simulating scenarios, each corresponding to one of the permutations we generated. For each scenario, we created a dataset using the treatment unit (Croatia) and one of the control sets from the permutations. Subsequently, we conducted the synthesis and calculated the RMSPE for each scenario. Our goal was to identify the scenario that resulted in the lowest MSPE, indicating the most appropriate set of control units for the synthesis. Our calculations, reveal that one of the simulating scenarios, with a specific set of donor countries, produced the minimum RMSPE (0.006475). The control set was meticulously curated to include the following countries: Austria, Bulgaria, Estonia, Spain, Greece, Hungary, Lithuania, Latvia, Poland, and Slovenia. Utilizing the SCM, we estimate what the TFP in Croatia would have been if the AFOPBS had not been approved. This estimation is based on a carefully designed set of donor countries that represents the best-case scenario. In this analysis, we consider TFP growth rate and incorporate the control factors discussed in the earlier data section to approximate the outcome variable. We explore all possible combinations of control countries to achieve the closest match to Croatia's actual TFP and its characteristic variables before the treatment period, specifically, pre-2013, preceding the adoption of the pre-bankruptcy reform. Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 203. The optimal match for Croatia's TFP (the outcome variable) and other characteristic variables, such as the minimum distance between synthetic and actual outcome and characteristic variables, is achieved using the corresponding weights of control countries, as detailed in Table 2. These weights for the control countries are determined through the minimization problem outlined in Equation 5. To synthesize the counterfactual results, we assign weights to each country in the control group, and these weights are displayed in Table 2. Notably, Table 2 reveals that TFP in Lithuania, Greece, and Hungary carries the highest weight, each at 99.998%, while Bulgaria and Poland contribute with minimal weights of approximately 0.001% each. In Table 3, we compare the mean values of the outcome variable and covariates in Croatia and the synthetic control created through the minimization process, along with the averages of all countries in the donor pool during the pre-policy period. As depicted in Table 3, the disparities between actual and synthetic outcomes, as well as characteristic variables, are notably small. Both realized and synthetic TFP growth is depicted in Figure 1. As illustrated in the graph, a close alignment exists before the treatment period (2008–2013), indicating our effective preliminary work in selecting the optimal donor pool and control variables. Following the treatment period, the series labelled 'synthetic Croatia' illustrates the estimated TFP growth Croatia would have experienced if the pre- bankruptcy act had not been adopted. The figure reveals a distinctive pattern with two sinusoidal curves intersecting three times during the post-intervention period. The impact of the AFOPBS, as observed in Figure 1, was not immediate and varied over time. In the short term, spanning from 2013 to 2015, the policy change led to instantaneous disruptions and adjustments that temporarily lowered TFP growth. This could be attributed to businesses struggling to adapt to new regulations and other transitional challenges. However, a noteworthy shift occurred after 2015, where realized TFP growth outperformed its synthetic counterpart for the following two years until 2017. 204 LEXONOMICA. Figure 1: The TFP growth in real and synthetic Croatia Source: Authors’ calculation. Despite these marked positive signs, we also acknowledge that policies like those under AFOPBS, even when well-intentioned, can sometimes yield unintended consequences. In light of this, the negative impact observed in the short term (2017- 2019) warrants further investigation to comprehend the underlying factors contributing to it. This exploration will help determine whether these effects were solely due to the treatment itself or influenced by external factors. Importantly, it is encouraging to highlight that the gap in TFP growth is gradually narrowing over time, with an average reduction of -0.055% during the post- treatment period. This suggests that the treatment may indeed be having a positive, albeit gradual, effect. This perspective is particularly relevant given our belief that the treatment's objectives were, or should have been, oriented toward long-term outcomes. Figure 2: The TFP growth gap between real Croatia and synthetic Croatia Source: Authors’ calculation. Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 205. Figure 2 above illustrates the evolution of the TFP growth gap, providing a comparative view of Croatia and the synthetic control group. Notably, this visual comparison highlights two distinct short-term negative effects resulting from the implementation of the AFOPBS on Croatia's TFP growth. The two distinct phases of convergence and divergence, as depicted by the cumulative gap values, reflect the dynamic impact of the AFOPBS policy on Croatia's TFP evolution. While the early years post-treatment witnessed a rapid convergence, the subsequent years demonstrated a concerning decline in Croatia's TFP growth, underscoring the need for a comprehensive analysis of these shifts. Our results contradict those of authors (Sadeghi and Kibler, 2022: 5), who utilized the same method as we did and found no effect whatsoever on the outputted economic indicator. Additionally, their robustness tests indicate a zero-effect impact of bankruptcy reform in 2004 on the overall level of entrepreneurial activity in Finland. 4.2 Robustness Tests In this analysis, we employ in-space placebo tests to compare the estimated treatment effect for treated Croatia with all the (simulated) treatment effects of the control countries. These simulated effects are derived from experiments where each control country is hypothetically affected by the same event (the pre-bankruptcy reform) in the same year (2013) as treated Croatia. If the estimated effect in treated Croatia is larger than the majority of the effects obtained from these simulated experiments, it allows us to conclude that the observed findings are not attributable to chance. In such a scenario, we can confidently attribute the effect to the adoption of AFOPBS. The p-value derived from the post/pre-treatment MSPE ratio for treated Croatia and the placebo tests is relatively high (0.18). This p-value is often used to assess the validity of the synthetic control approach. In general, a higher p-value suggests that the synthetic control approach is performing well because it indicates that the actual treated Croatia is not an extreme outlier in terms of prediction error. None of the placebo effects were found to be as significant as the estimated effect in Croatia regarding TFP outcomes (as shown in Fig. 3). Only one country, Spain, exhibited a placebo effect greater than that of Croatia in the TFP movement scenario, indicating 206 LEXONOMICA. a substantial policy effect stemming from the pre-bankruptcy settlement or financial reform. Figure 3. The Ratio of post-intervention MSPE and pre-intervention MSPE: Croatia and donor pool Source: Authors’ calculation. It's worth noting that, following the standard practice of Abadie, Diamond, and Hainmueller (2010), we may choose to exclude Spain due to pre-policy MSPE values of TFP that were more than five times greater than the MSPE observed in Croatia during the placebo test analysis. We interpret these seemingly substantial placebo effects in the excluded country, Spain, as a result of a lack of fit rather than the effect of the assumed reform. The placebo testing suggests that our synthetic control approach provides a good match for Croatia's post-treatment outcomes, and there is no strong evidence to suggest that Croatia's outcomes significantly deviate from what would be expected based on the control group. In other words, we can conclude that we are observing a treatment effect that exceeds what would typically occur by chance. In the context of placebo testing, we constructed Table 4 Post/Prior RMSPE ratio. The provided data offers valuable insights into the effectiveness of the intervention in Croatia, serving as a focal point for understanding its impact on neighbouring Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 207. countries. The data encompasses a diverse set of countries, each subjected to a hypothetical intervention akin to Croatia's pre-bankruptcy reform. These countries serve as a control group, enabling a comparative assessment of the treatment's outcomes. The Prior-Intervention RMSPE values quantify the predictive accuracy of the model before the intervention, with lower values reflecting a more precise fit of the model to historical data, whereas the Post-Intervention RMSPE values depict the predictive accuracy of the model after the intervention. Notably, countries like Austria, Bulgaria, Estonia, Hungary, Lithuania, Latvia, Poland, and Slovenia exhibit a substantial increase in prediction errors, suggesting a detrimental impact of Croatia's intervention. Spain stands out with an exceptionally high Post-Intervention RMSPE, reflecting a substantial decline in predictive accuracy. This may be due to factors unrelated to the assumed reform. In contrast, Greece emerges as a success story, with a significantly lower Post-Intervention RMSPE, indicating a positive influence of Croatia's intervention on predictive accuracy. Croatia, as the treated country, experienced an improvement in predictive accuracy, with a lower Post-Intervention RMSPE, indicating a positive impact on its own outcomes. The analysis underscores the importance of assessing spillover effects, as the impact of Croatia's intervention varies across countries. Some neighbouring nations experience adverse effects, while others benefit from improved predictive accuracy. In summary, the placebo testing and RMSPE analysis provide a robust framework for evaluating the effectiveness of the pre-bankruptcy reform in Croatia and its varying impacts on neighbouring countries. It allows us to discern the diverse impacts on neighbouring countries, emphasizing the need for a nuanced understanding of spillover effects and the potential for reforms to influence predictive modelling accuracy. In addition to the placebo test, we employed the PDID method for cross-validation of our previous results. The results are presented in Table 5. Irrespective of whether control variables (including production, capital formation, employment distribution between service and manufacturing sectors, human capital, openness, and labor economy issues) are included or not, the results consistently indicate a significant impact of the AFOPBS (pre-bankruptcy act) on reducing TFP 208 LEXONOMICA. movements in Croatia, accounting for both time and county fixed effects (mod-3 and mod-6). This robust effect of the AFOPBS, in this paper, is further substantiated. The consistent negative effect of the AFOPBS on TFP is observed across these different specifications, varied by the inclusion of control variables, time-fixed effects, and country-fixed effects. 5 Conclusion TFP is a crucial economic indicator as it plays a pivotal role in improving the standard of living for the average individual in Croatia. In 2013, with a short time lag, the initial impacts on TFP materialized following the adoption of the AFOPBS. Croatia recently enacted this Act to enhance the payment process among economic entities, alleviate liquidity issues, and revitalize the economy after experiencing a severe recession. This measure aimed to shorten the duration of challenging economic years. Through the SCM, we evaluated the effectiveness of the AFOPBS in potentially enhancing the trajectory of TFP, a fundamental driver of long-term improvements in living standards. The findings reveal a significant decrease in TFP shortly after the adoption of the AFOPBS in 2012, compared to the pre-reform period. By 2015, Croatia's TFP declined relative to synthetic Croatia, with another phase of decreasing trends emerging after 2017. Only the intermediate period experienced a brief upswing in TFP. Importantly, all of these effects withstood robustness tests, including placebo examinations and cross-validation using the PDID method. This study demonstrates that, as a remedial policy aimed at addressing liquidity issues in Croatia, the AFOPBS has yielded mixed results regarding TFP. Both the initial and subsequent impacts of the AFOPBS on TFP reduction are statistically significant. It raises the question of whether the movement in TFP can be attributed to the introduction of the AFOPBS in 2012. While it is possible, the determination of whether these consequences are positive or negative is not straightforward. Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 209. Acknowledgement This paper is a result of the scientific project Tourism Crises - stakeholders' Roles and Recovery Strategies (2023-2026) supported by the Faculty of Economics and Tourism "Dr. Mijo Mirković", Juraj Dobrila University of Pula, Croatia. References Abadie, A. (2021). Using synthetic controls: Feasibility, data requirements, and methodological Aspects, J. Econ. Lit., 59(2), 391–425. DOI: 10.1257/jel.20191450. Abadie, A., Gardeazabal, J. (2003). The economic costs of conflict: A case study of the Basque Country, Am. Econ. Rev., 93, 113–132. DOI: 10.2139/ssrn.293120. Abadie, A., Diamond, A., Hainmueller, A. J. (2010). Synthetic control methods for comparative case studies: Estimating the effect of California’s tobacco control program, J. Am. Stat. Assoc., 105, 493–505. DOI: 10.1198/jasa.2009.ap08746. Abadie, A., Diamond, A., Hainmueller, J. (2015). Comparative Politics and the Synthetic Control Method. Am. J. Pol. Sci., 59, 495–510. DOI: 10.2139/ssrn.1950298. Acemoglu, D., Guerrieri, V. (2008). Capital Deepening and Non-Balanced Economic Growth, Journal of Political Economy, 116, 467–498. DOI: 10.1086/589523. Albalate, D., Bel, G., Mazaira-Font, F. A. (2021). Decoupling synthetic control methods to ensure stability, accuracy, and meaningfulness, SERIEs, 12(1), 549–584. DOI: 10.1007/s13209-021- 00242-8. Aleksanyan, L., Huiban, J. P. (2016). Economic and financial determinants of firm bankruptcy: evidence from the French food industry, Review of Agricultural, Food and Environmental Studies, 97(2), 89- 108. DOI: 10.1007/s41130-2016. Barro, R. J. (1999). Determinants of Economic Growth: A Cross-Country Empirical Study, American Political Science Association, 76(6), 1-55. DOI: 10.2307/2585721. Bodul, D., Vuković, A. (2015). Još jedna reforma stečajnog zakonodavstva – funkcionalizacija stečajno-pravne zaštite ili placebo efekt?, Zbornik Pravnog fakulteta Sveučilišta u Rijeci, 36(1), 181-212. Cespedes, J., Thakor, R. T., Yang, K. (2022). The Ex Ante Effect of Bankruptcy Law: Evidence from Chapter 12, 1-30. DOI: 10.2139/ssrn.4236600. Cosci, S., and Meliciani, V. (2002). Multiple banking relationships: Evidence from the Italian experience. Manchester School, 70(0), 37–54. DOI: 10.1111/1467-9957.70.s1.3. Di Martino, P., Latham, M., Vasta, M. (2020). Bankruptcy Laws Around Europe (1850–2015): Institutional Change and Institutional Features, Enterprise & Society, 21(4), 1-55. DOI: 10.1017/eso.2019.46. Dvouletý, O., Srhoj, S., Pantea, S. (2021). Public SME grants and firm performance in European Union: A systematic review of empirical evidence, Small Business Economics, 57, 243–263. DOI: 10.1007/s11187-019-00306-x. Ferranti, D., Baumol, W., Malach, M., Pablos-Mendez, A., Tabish, H. and Wu, L. (2012). The Cost Disease: Why Computers Get Cheaper but Healthcare Doesn't. New Haven, CT: Yale University Press. DOI: 10.12987/9780300188486. Gonçalves, D., Martins, A. (2016). The Determinants of TFP Growth in the Portuguese Manufacturing Sector, Boletin Mensan de Economia Portugesa, 9, 47-55. Groningen Growth and Development Centre (GGDC). (2021). Penn World Table 10.0. Retrieved from URL: https://www.rug.nl/ggdc/productivity/pwt/pwt-releases/pwt100. Hiroki, T., Iwatsubo, K., Watkins, C. (2022). Does Firm-Level Productivity Predict Stock Returns? Pacific-Basin Finance Journal, 72, 1-45. DOI: 10.1016/j.pacfin.2022.101710. Hurst, E., Pugsley, B. W. (2011). What do Small Businesses Do? Brookings Papers on Economic Activity, 43(2), 73-142. DOI: 10.1353/eca.2011.0017. Jensen, M. C. (1986). Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers, The American Economic Review, 76(2), 323-329. 210 LEXONOMICA. Köke, J. (2001). Control Transfers in Corporate Germany: Their Frequency, Causes, and Consequences, EFA 2001 Barcelona Meetings, p. 00-67. Kreif, N., Grieve, R., Hangartner, D., Turner, A. J., Nikolova, S., Sutton, M. (2016). Examination of the synthetic control method for evaluating health policies with multiple treated units, Health Econ, 25(12), 1514-1528. DOI: 10.1002/hec.3258. Lim, Y., Hahn, C. H. (2009). Bankruptcy Policy Reform and Total Factor Productivity Dynamics in Korea: Evidence from Microdata. In Ito, T., Rose, A. K., eds. Growth and Productivity in East Asia, NBER-East Asia Seminar on Economics, University of Chicago Press, 13: 10-33. Lin, B., Chen, X. (2018). Is the implementation of the increasing block electricity prices policy really effective? Evidence-based on the analysis of synthetic control method, Energy, 163(3), 734–750. DOI: 10.1016/j.energy.2018.08.178. Miao, L., Zhuo, Y., Wang, H., Lyu, B. (2022). Non-Financial Enterprise Financialization, Product Market Competition, and Total Factor Productivity of Enterprises, SAGE Open, 12(2), 1-15. DOI: 10.1177/21582440221089956. Misra, B. S. (2019). Determinants of total factor productivity in Indian states, Indian Growth and Development Review, 13(1), 259–82. DOI: org/10.1108/igdr-01-2019-0008. Neira, J. (2017). Bankruptcy and Cross-Country Differences in Productivity. Journal of Economic Behavior & Organization, 157, 1-47. DOI: 10.1016/j.jebo.2017. Qi, X., Han, Y. (2021). Energy quota trading can achieve energy savings and emission reduction: Evidence from China’s pilots, Environ. Sci. Pollut. Res., 28(1), 1-28. DOI: 10.1007/s11356- 021-14409-0. Sadeghi, A., Kibler, E. (2022). Do bankruptcy laws matter for entrepreneurship? A Synthetic Control Method analysis of bankruptcy reform in Finland, Journal of Business Venturing Insights, 18(2), 1- 9. DOI: 10.1016/j.jbvi.2022. Sampaio, B. (2014). Identifying peer states for transportation policy analysis with an application to New York’s handheld cell phone ban, Transp. A Transp. Sci., 10, 1–14. DOI: 10.1080/18128602.2012.688073. Schweiger, H. (2011). The impact of state aid for restructuring on the allocation of resources, EBRD Working Paper No. 127, 1-40. Available at SSRN: https://ssrn.com/abstract=2180457 or http://dx.doi.org/10.2139/ssrn.2180457. Shao, W., Sun, Y., Bai, X., Naeem, M. A., Taghizadeh-Hesary, F. (2022). Zombie enterprises, crowding out effect, and total factor productivity: Empirical evidence from Chinese manufacturing listed companies, International Journal of Finance & Economics, 28(4), 4512-4531. DOI: 10.1002/ijfe. Tamayo, C. (2017). Bankruptcy choice with endogenous financial constraints, Review of Economic Dynamics, 26, 225-242. DOI: 10.1016/j.red.2017.06.004. Tomura, Hajime. (2007). Firms Dynamics, Bankruptcy Laws, and Total Factor Productivity, Bank of Canada Working Paper, 17, 1-10. DOI: 10.34989/swp-2007-17. Vollrath, D. (2019). Fully Grown: Why a Stagnant Economy Is a Sign of Success. The University of Chicago Press, Chicago, and London. DOI: 10.7208/Chicago/9780226666143.001.0001. World Bank. (2023). World Development Indicators. Retrieved from URL: https://datatopics.worldbank.org/world-development-indicators/. Xia, H., Wu, S. (2021). Effect of the International Tourism Island Policy on Hainan Farmers’ Income and the Urban-rural Income gap: An analysis based on the GPCA and SCM Models. J. Adv. Comput. Intell. Intell. Inform., 25(5), 592–600. DOI: 10.20965/jaciii.2021.p0592. Yamaguchi, K. (2022). The productivity impact of the government-led bailout of Japan Airlines, Asian Transport Studies, 8, 1-9. ISSN 2185-5560. DOI: 10.1016/j.eastsj.2022. Z. Šergo, J. Gržinić, M. Sučić Čevra: Charting the Course: Total Factor Productivity Trends in Croatia Post-pre-bankruptcy Act 211. Appendix Table 1: List of Confounding Variables and Their Explanations Variable Explanation GDPPC Gross Domestic Product per capita (current US$) UEM Total unemployment (% of labor force) TLFPR Labor force participation rate (%) EMPPOP Employment to population ratio (%) GCF Gross capital formation (% of GDP) FDI Net inflow of foreign direct investment (current US$) PRMENR Primary school enrollment (%) INDEMP Employment in industry (%) SRVEMP Employment in services (%) EXPGNS Exports of goods and services (% of GDP) Source: The World Bank (World Bank, 2023.) Table 2: Country Weights for the Synthetic Controls Unit numbers Country Weight 1 Austria 0 2 Bulgaria 0,001 3 Estonia 0 4 Spain 0 6 Greece 0,283 8 Hungary 0,221 11 Lithuania 0,494 12 Latvia 0 13 Poland 0,001 16 Slovenia 0 1,000 Source: Authors’ calculation. Table 3: Predictor Balance for TFP Treated Synthetic Sample Mean GDPPC 0,986 0,985 0,966 UEM 14388,878 17469,922 21056,861 TLFPR 11,792 12,913 11,357 EMPPOP 52,406 54,341 56,831 GCF 46,245 47,301 50,378 FDI 22,894 19,693 22,945 PRMENR 4,336 4,848 4,974 INDEMP 93,497 100,315 101,271 SRVEMP 28,635 25,392 27,971 EXPGNS 57,917 65,855 64,412 Source: Authors’ calculation. 212 LEXONOMICA. Table 4: Post/prior RMSPE ratios for the TFP growth rate in Croatia and the control countries Country Prior Intervention RMSPE Post Intervention RMSPE Ratio Austria 0,009 1,034 109,883 Bulgaria 0,009 1,915 207,161 Estonia 0,012 1,561 128,742 Spain 0,002 9,137 5529,906 Greece 0,133 0,167 1,255 Hungary 0,010 1,548 148,060 Lithuania 0,016 1,099 66,743 Latvia 0,045 0,666 14,668 Poland 0,023 0,362 15,690 Slovenia 0,019 1,241 64,979 Croatia 6,977 2,973 0,426 Source: Authors’ calculation. Table 5: PDID Estimation Results of the AFOPBS Effects on TFP in Croatia Variable mod-1 mod-2 mod-3 mod-4 mod-5 mod-6 treat*post -0.083* (0.041) -0.109* (0.034) -0.157* (0.029) -0.072* (0.032) -0.116* (0.048) -0.024* (0.025) Control variables NO YES NO YES YES YES Time fixed NO NO YES NO YES YES Country fixed NO NO YES YES NO YES R 2 0.121 0.158 0.356 0.324 0.329 0.278 Num. obs. 300 300 300 300 300 300 Source: Authors’ calculation. Notes: *p < 0.05; the "*" symbol indicates statistical significance and values in parentheses represent standard errors. Digitalna knjižnica Slovenije - dLib.si
NAPAKA - STRAN NI BILA NAJDENA


Žal nam je, a strani, ki ste jo iskali, ni bilo moč najti.
Prosimo, poskusite uporabiti ISKALNIK in poiskati publikacijo, ki jo potrebujete.


Knjige
Periodika
Rokopisi
Slike
Glasba
Zemljevidi