Daniel Lopez-Creamer1, Yilsy M. Núñez-Guerrero1, Ángel Huerga-Gonzalez1 and Carlos Rodríguez-Monroy1

1 Dpt. of Industrial Engineering, Business Administration and Statistics. Universidad Politécnica de Madrid. C/ José Gutiérrez Abascal, 2, 28006 Madrid, Spain.

sebastian.lopez.creamer@alumnos.upm.es, ym.nunez@upm.es, angel.huerga@upm.es, carlos.rodriguez@upm.es

Keywords: Sustainability; Finance; ESG criteria; ESG Bonds

The shift towards economic growth models based on sustainability gives rise to innovations in financial products, including environmental, social, and governance (ESG) criteria, which are being applied to both governments and organizations’ investment and financing decision-making. In the present study, a systematic review of the literature has been carried out to determine the elements that influence the issue spreads of ESG bonds. This research focuses on delimiting the characteristics that affect the issue spread of ESG bonds from this referential framework. According to a study by the Bank for International Settlements (BIS) [1,2], ESG bonds have a lower issue spread (margin over the reference rate) due to their high demand, which represents a further compensation for monitoring costs which are necessary by their very nature.

This analysis will allow a better understanding of the functioning of the elements that make up the financial products issued under the ESG criteria. Besides, this information will improve risk management and decision-making by investors, demystifying that investing in sustainable projects implies giving up the profitability offered by any other investments in traditional debt.

The main elements to consider in this research are those that make up the issuance of the ESG bonds [35]. At issuance: region and country of incorporation of the issuer, issue date, the nominal value of the bond, currency, and issue spread which is the extra margin over the benchmark bond offered by the issuer, expressed in basis points. The type, green, sustainable, or social. The term of the issuance in months, and finally, the general industry group (GIG) of the issuing entity [1,6,7].

The following methodology is intended to study the characteristics that influence bond spreads. Green, sustainable, and social bonds issued between 2016 and 2019 are evaluated. The data for this analysis correspond to 584 observations of the ESG bonds issued globally between the dates mentioned above, extracted from the Dealogic database. The observed dependent variable is the issue spread over the benchmark calculated at the bond issuance date.

To understand the information provided by the variables under study, (face value, currency, type of bond, term, and industry) and the relationship between them, a modeling of neural networks and regressions was carried out using the Rstudio statistical software.

Artificial neural networks are mathematical models to approximate and interpolate nonlinear relationships between input and output data [20]. They have two variants according to their purpose, classification, and prediction.

 The main results showed that the resulting models had low R-squared, which meant that there were variables that were not contributing to the models. That is why different models (multiple and simple) were used. Out of these variables, term, region, and industry were significant. The application of the regression models and the neural network models show that the mean square errors of the different models are the following:

Method Mean square error
Simple linear regression 1399.126
Neural network with one entry 949.384
Multiple linear regression 1127.335
Neural network with multiple inputs 669.21

Table 1. Mean square error for method

The results show that the region and industry are significant variables in the multiple linear regression. This is reflected in the mean error differences R square (R2) and Square (SME) between simple linear regression and multiple linear regression (SLR: R2=0.7249, SME=1399.126; MLR: R2 = 0.8204 SME=1127.335).

Our findings show that the term, the region, and, to a lesser extent, the industry that issues the ESG debt are the influencers. In addition to analyzing the financial system’s sustainable elements, it can be extracted that opting for sustainability does not imply obtaining a worse performance. Issuers of ESG debt instruments use it as a communication tool for their corporate strategy and adapt to investors’ appetite for ESG assets. The USA, Europe, and the rest of the world differ in their strategy to develop a sustainable finance market.

While in Europe, governments have a strategy of active promotion and regulatory changes that encourage sustainable investment, in the USA, the market is allowed to act freely. In other regions, sustainable investment is not yet welcomed by investors or regulators, creating a geographical disparity in the dispersion of bonds issued in the different regions. It can also be seen that the more significant variables that are taken into account, the more the model adjusts to reality. It is an obvious result that reflects the large number of factors that influence the spread of a bond.

Finally, we can confirm that the most suitable model of those mentioned previously is, by far, the neural network with multiple inputs. The reason is that the neural network does not have to fulfill the series of hypotheses necessary in linear regression, so it has less rigidity in forming a model.


  1. Lopez-Creamer, Daniel. Estudio del mercado sostenible. Aplicación de redes neuronales y regresión a bonos verdes. Trabajo de fin de grado presentado en la Escuela Superior de Ingenieros industriales de la Universidad Politécnica de Madrid, July 2020.
  2. Ehlers, T., & Packer, F. (2017). Green bond finance and certification. Accessed on 14 May 2020, in Bank for International Settlements: https://www.bis.org/publ/qtrpdf/r_qt1709h.htm
  3. Climate Bonds Initiative. (2019). Accessed on 14 May 2020, in Explaining green bonds: https://www.climatebonds.net/market/explaining-green-bonds
  4. Climate Bonds Initiative. (2020). Accessed on 15 May 2020, in Market Blogs: https://www.climatebonds.net/market-blogs
  5. Climate Bonds Initiative. (2020). Green Bond Highlights 2019: Behind the Headline Numbers: Climate Bonds Market Analysis of a record year. Accessed on 15 May 2020, in https://www.climatebonds.net/2020/02/green-bond-highlights-2019-behind-headline-numbers-climate-bonds-market-analysis-record-year
  6. Spainsif. (2017). Labels ISR en Europa. Accessed on 11 May 2020, in https://www.spainsif.es/labels-isr-en-europa/
  7. Yolanda, R. (2019). ANN Classification with ‘nnet’ Package in R. Accessed on 2 June 2020, in Medium website: https://medium.com/@yolandawiyono98/ann-classification-with-nnet-package-in-r-3c4dc14d1f14



Icon for the Creative Commons Attribution 4.0 International License

Proceedings of the 15th International Conference on Industrial Engineering and Industrial Management and XXV Congreso de Ingeniería de Organización Copyright © by (Eds.) José Manuel Galán; Silvia Díaz-de la Fuente; Carlos Alonso de Armiño Pérez; Roberto Alcalde Delgado; Juan José Lavios Villahoz; Álvaro Herrero Cosío; Miguel Ángel Manzanedo del Campo; and Ricardo del Olmo Martínez is licensed under a Creative Commons Attribution 4.0 International License, except where otherwise noted.

Share This Book