Patrícia Ramos

Abstract
This study investigates the impact of Research and Development (R&D) investment on the performance of Portuguese companies, specifically addressing the gap in understanding how R&D influences a company's value and performance. We employ a dynamic panel data model estimated using the Generalized Method of Moments (GMM) to account for potential endogeneity issues. This approach allows us to analyze the influence of R&D investment on the Return on Operating Assets (ROA) for Portuguese companies with significant R&D investments between 2012 and 2019. The analysis reveals that while R&D investment itself may not have a statistically significant short-term impact on ROA, lagged financial performance, leverage, asset turnover ratio, and accounts payable turnover all demonstrate a statistically significant relationship with the dependent variable.

Abstract
Socially responsible investments, also referred to as ethical or sustainable investments, have experienced rapid global growth in recent years. They represent an investment approach that incorporates social, environmental, and ethical considerations into decision-making processes. Consequently, the significance of socially responsible investments has captured the attention of academics, prompting inquiries into the impact of integrating social criteria on portfolio performance. The primary objective of this work was to conduct a comparative study of the performance between socially responsible and non-socially responsible investment funds, using funds domiciled in Portugal and Spain. Various multi-factor models, including the three-factor model of Fama and French, the four-factor model of Carhart, and the five-factor model of Fama and French, were employed to assess performance. The sample comprised 125 investment funds, with 43 identified as socially responsible and 82 as non-socially responsible. The study's findings indicate that there are no significant differences between socially responsible funds and their conventional counterparts. The majority of funds experience performance alterations during periods of crisis compared to crisis-free periods. Additionally, when comparing non-conditional models with conditional models, an improvement in the explanatory power of the latter is observed. This suggests that the inclusion of the dummy variable enhances the quality of fit for the models.

Abstract
This study investigates the effectiveness of Transformer-based models for retail demand forecasting. We evaluated vanilla Transformer, Informer, Autoformer, PatchTST, and temporal fusion Transformer (TFT) against traditional baselines like AutoARIMA and AutoETS. Model performance was assessed using mean absolute scaled error (MASE) and weighted quantile loss (WQL). The M5 competition dataset, comprising 30,490 time series from 10 stores, served as the evaluation benchmark. The results demonstrate that Transformer-based models significantly outperform traditional baselines, with Transformer, Informer, and TFT leading the performance metrics. These models achieved MASE improvements of 26% to 29% and WQL reductions of up to 34% compared to the seasonal Na & iuml;ve method, particularly excelling in short-term forecasts. While Autoformer and PatchTST also surpassed traditional methods, their performance was slightly lower, indicating the potential for further tuning. Additionally, this study highlights a trade-off between model complexity and computational efficiency, with Transformer models, though computationally intensive, offering superior forecasting accuracy compared to the significantly slower traditional models like AutoARIMA. These findings underscore the potential of Transformer-based approaches for enhancing retail demand forecasting, provided the computational demands are managed effectively.

Abstract
This research aims to analyze the impact of social responsibility (SR) on the performance of 216 European companies from 2017 to 2021. The objective of this research is to determine how the operational, financial, and market performance of companies is influenced by social responsibility practices. The methodology adopted is quantitative in nature, using the estimation of models for panel data. To quantify corporate performance, this study uses the return on assets (ROA), the return on equity (ROE), and finally Tobin's Q ratio. Additionally, environment, social, and governance (ESG) and United Nations Global Compact (GC) scores are used to quantify SR. Our findings indicate a complex relationship between SR and corporate performance. While SR positively impacts market performance, it negatively affects operational and financial performance. This disparity becomes more pronounced when comparing companies with the highest and lowest SR scores. Further analysis reveals that the environment, social, and governance dimensions of ESG negatively correlate with ROA and ROE, but positively correlate with Tobin's Q. The GC's anti-corruption and environment scores exhibit a negative relationship with Tobin's Q, the human rights dimension negatively correlates with ROE and ROA, and the labor law dimension positively influences ROE. Notably, firm size amplifies these relationships, whereas firm age has a dampening effect. This research offers significant contributions to the literature by providing a comprehensive analysis of the impact of social responsibility on corporate performance based on ESG and GC scores.

Abstract
Retailers depend on accurate sales forecasts to effectively plan operations and manage supply chains. These forecasts are needed across various levels of aggregation, making hierarchical forecasting methods essential for the retail industry. As competition intensifies, the use of promotions has become a widespread strategy, significantly impacting consumer purchasing behavior. This study seeks to improve forecast accuracy by incorporating promotional data into hierarchical forecasting models. Using a sales dataset from a major Portuguese retailer, base forecasts are generated for different hierarchical levels using ARIMA models and Multi-Layer Perceptron (MLP) neural networks. Reconciliation methods including bottom-up, top-down, and optimal reconciliation with OLS and WLS (struct) estimators are employed. The results show that MLPs outperform ARIMA models for forecast horizons longer than one day. While the addition of regressors enhances ARIMA’s accuracy, it does not yield similar improvements for MLP. MLPs present a compelling balance of simplicity and efficiency, outperforming ARIMA in flexibility while offering faster training times and lower computational demands compared to more complex deep learning models, making them highly suitable for practical retail forecasting applications.