Supervised learning in the improvement of public investment policies for the promotion of economic development

Abstract

Investments in energy-efficient technologies are challenging for SMEs, which face which face major financial obstacles when investing in energy efficiency. One of the main drivers to promote investment in energy efficiency measures in SMEs are public aids, including subsidies to industrial SMEs for renewable energies, aimed at the installation of clean energy sources and the replacement of industrial equipment with more efficient alternatives, as well as tax incentives for these SMEs to adopt renewable energies. However, there is a gap of information about the features of which industrial SMEs are more likely to apply for them. A profile of potential beneficiary SMEs is defined applying the random forest approach for unbalanced samples to extract and identify the most relevant indicators for those companies, what will encourage the design of more focused calls, helping to boots the effectiveness of public subsidies. Apart from public aids aimed at energy transition, this thesis focuses on analysing the role of public guarantee policies, which facilitated access to credit for SMEs during the COVID 19 pandemic. In the present PhD thesis, it is analysed if there was an efficient allocation of this program of public guarantees. SMEs who suffered any distress prior to COVID 19 pandemic (structural distress) should not have been beneficiaries of that public aid because they would tend to invest this money in very risky projects, thereby increasing the shareholder-creditor agency conflict and reducing efficient allocation of this public aid. The third chapter of this PhD thesis is once again linked to the importance of mitigation strategies aimed at reducing greenhouse gas emissions. Regulated and voluntary carbon markets are key instruments to promote the reduction of greenhouse gas emissions. This PhD thesis focuses on regulated carbon markets, where carbon is priced and companies can reduce their emissions by trading carbon credits. We center on individual high- and low-emission companies, providing a more nuanced view of how carbon markets affect different types of firms. Second, it specifically investigates how recent environmental reforms in the EU influence volatility transmission, offering valuable insights into the regulatory impact on financial markets and corporate behavior. Third, this study analyzes the evolution of volatility transmission from carbon markets over the third and fourth phases of the EU-ETS, highlighting how changes in market structure and regulatory adjustments have influenced market volatility over time. Finally, it compares the patterns of increased volatility transmission between high-emission and low-emission companies, illustrating how different levels of emissions exposure lead to varying impacts from carbon market fluctuations, which is critical for understanding the differential effects of environmental policies on firms. In doing this, the dynamic connectivity index of Diebold and Yilmaz (2012, 2014), combined with the decomposition of interdependence into frequency bands (Barunik and Krehlik, 2018) are applied. Our results show that the connectivity between the EU ETS returns and returns of high- and low-emitting European companies is high and changing over the third and fourth phases of the EU ETS (from 2013 to 2024). Specifically, carbon volatility transmission increases over this period due to the implementation of EU ETS reforms to accelerate decarbonisation and achieve the EU climate objectives. Moreover, according to our results, and in line with previous studies, carbon is a volatility receiver, especially in energy or energy-intensive sectors, and most of the connectivity occurs in the short term (1 day-5 days).