The first quarter of the 21st century has been marked by a radical change in the digital banking paradigm. The banking sector is undergoing a profound transformation, shifting from analog systems and direct relationships between banks and clients to fully digitalized processes managed remotely. This evolution has led to the massive and continuous use of personal data to determine the identity and creditworthiness of loan applicants, giving rise to what we now call the Consumer 2.0. The Consumer 2.0 is constantly scanned and digitized: every action, both online and offline, is digitally recorded in order to study and predict consumption habits and, of course, credit history. Almost all financial transactions that users once carried out in person are now available online. Consumer credit, in particular, has received a significant boost from this new organization of financial intermediaries and banks. It has become faster, more flexible, and more personalized. Behind this epochal phenomenon, however, lie significant social challenges for Consumer 2.0. The goal of any lender is, in fact, to make a profit; therefore, the ability to reliably predict the probability of borrower default— in other words, to precisely assess their Creditworthiness— is crucial. The advent of artificial intelligence in all its forms seems to be the most widely adopted solution for this task. We are moving from statistical models analyzing data such as income, existing debt, payment history, and other financial indicators, toward the adoption of AI- and machine learning–based algorithms for rapid and efficient analysis of large volumes of data, including non-traditional factors such as spending behaviour and social media activity. This shift enables more accurate and personalized predictive assessments. This new category of borrowers is analysed in all their digital actions by machine learning algorithms that, among other things, determine each individual’s creditworthiness— with possible ethical and legal implications. The purpose of this work is to study the transformation that the consumer credit market is undergoing from a technological perspective, as well as its ethical and sociological implications.
Il primo quarto del 21^ secolo è stato di radicale cambiamento del paradigma digital banking. Il settore bancario sta vivendo una trasformazione epocale, passando da sistemi analogici e relazioni dirette tra banca e cliente, a processi interamente digitalizzati, gestiti a distanza. Questa evoluzione ha portato ad un massiccio e continuativo utilizzo dei dati personali per determinare l'identità è la solvibilità dei richiedenti credito e alla nascita di quello che chiamiamo il consumatore 2.0. Il consumatore 2.0 viene continuamente scansito e digitalizzato, ogni sua azione sulla rete e al di fuori da essa viene memorizzata digitalmente per poter studiare e prevedere le sue abitudini di consumo e ovviamente la sua storia creditizia. Quasi la totalità delle operazioni finanziarie che gli utenti facevano in presenza è ora possibile online. Il credito al consumo in particolare ha ricevuto un impulso potente da questa nuova organizzazione degli intermediari finanziari e delle banche. E' diventato più veloce, flessibile e personalizzato. Dietro questo fenomeno epocale però si nascondono insidie sociali di rilevante impatto per i consumatori 2.0. L'obiettivo di chiunque presti denaro infatti è quello di fare profitto, quindi di riuscire a prevedere nel modo più affidabile possibile la probabilità di insolvenza del creditore, in poche parole essere in grado di valutare con elevata precisione il suo “Merito Creditizio”. L'avvento dell'intelligenza artificiale in tutte le sue declinazioni sembra essere la soluzione maggiormente adottata per assolvere a questo compito. Stiamo muovendoci da analisi basate su modelli statistici per l'analisi di dati come reddito, debiti esistenti, storico dei pagamenti e altri indicatori finanziari, all'adozione di algoritmi basati su AI e machine learning per analisi rapida ed efficiente di grandi volumi di dati, tra cui rientrano fattori non tradizionali come comportamenti di spesa e attività sui social media. Tutto ciò porta a valutazioni predittive più accurate e personalizzate. Questa nuova categoria di debitori viene analizzata in tutte le sue azioni digitali da algoritmi di machine learning che determinano fra le altre cose il merito creditizio di ciascuno con possibili implicazioni di carattere etico e giuridico. Questo lavoro ha come obiettivo lo studio della trasformazione che il mercato del credito al consumo sta subendo dal punto di vista tecnologico e delle sue implicazioni etiche e sociologiche.
Credit scoring in consumer credit: from traditional techniques to Artificial Intelligence
Gavioli, Tommaso
2024/2025
Abstract
The first quarter of the 21st century has been marked by a radical change in the digital banking paradigm. The banking sector is undergoing a profound transformation, shifting from analog systems and direct relationships between banks and clients to fully digitalized processes managed remotely. This evolution has led to the massive and continuous use of personal data to determine the identity and creditworthiness of loan applicants, giving rise to what we now call the Consumer 2.0. The Consumer 2.0 is constantly scanned and digitized: every action, both online and offline, is digitally recorded in order to study and predict consumption habits and, of course, credit history. Almost all financial transactions that users once carried out in person are now available online. Consumer credit, in particular, has received a significant boost from this new organization of financial intermediaries and banks. It has become faster, more flexible, and more personalized. Behind this epochal phenomenon, however, lie significant social challenges for Consumer 2.0. The goal of any lender is, in fact, to make a profit; therefore, the ability to reliably predict the probability of borrower default— in other words, to precisely assess their Creditworthiness— is crucial. The advent of artificial intelligence in all its forms seems to be the most widely adopted solution for this task. We are moving from statistical models analyzing data such as income, existing debt, payment history, and other financial indicators, toward the adoption of AI- and machine learning–based algorithms for rapid and efficient analysis of large volumes of data, including non-traditional factors such as spending behaviour and social media activity. This shift enables more accurate and personalized predictive assessments. This new category of borrowers is analysed in all their digital actions by machine learning algorithms that, among other things, determine each individual’s creditworthiness— with possible ethical and legal implications. The purpose of this work is to study the transformation that the consumer credit market is undergoing from a technological perspective, as well as its ethical and sociological implications.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_10_Gavioli.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Tesi di Laurea Magistrale, Author Tommaso Gavioli
Dimensione
2.94 MB
Formato
Adobe PDF
|
2.94 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/242199