The thesis examines the phenomenon of business cycle synchronization, which is the tendency of different economies to move in a correlated way over time, and the difficulty of traditional macroeconomic models in reproducing this pattern. This discrepancy arises from the fact that standard open economy business cycle models (e.g. Backus et al. (1992)) predict negative correlations between countries’ outputs, whereas empirical data reveal positive ones. To address this mismatch, the literature has proposed different solutions over the years: among them, the anticipated-shock channel constitutes our focus. According to the school of thought started with Beaudry and Portier (2006), news about future productivity (the so-called news shocks) can alter the current behavior of economic agents, triggering an early expansion domestically. Subsequent works, such as Beaudry et al. (2007), have shown that this mechanism also operates at the international level: news shocks can be transmitted across borders and shape foreign business cycles, contributing to cross-country co-movements. This thesis contributes to this debate from an empirical side, proposing an alternative approach to identifying news shocks, based not on economic data but on the actual information sources available to agents: newspaper articles. Using Natural Language Processing (NLP), three methods are developed to extract from these articles a sentiment index related to expectations about future productivity. The resulting series are then incorporated into a VAR model estimated on U.S. data, with the shock identified through an instrumental variable approach (i.e. Proxy SVAR), in order to assess how news shocks identified in this way affect key domestic macroeconomic variables. Finally, the identified shock is tested within a VARX model applied to Canada, with the aim of evaluating the international transmission of news shocks and their potential role as a channel for business cycle synchronization.
La tesi esamina il fenomeno della sincronizzazione dei cicli economici, ossia la tendenza delle diverse economie a muoversi nel tempo in modo correlato, e la difficoltà dei modelli macroeconomici tradizionali nel riprodurre questo andamento. Tale discrepanza nasce dal fatto che i modelli standard dei cicli economici in economia aperta (ad esempio Backus et al. (1992)) prevedono correlazioni negative tra i prodotti nazionali dei vari Paesi, mentre i dati empirici mostrano correlazioni positive. Per affrontare questo disallineamento, la letteratura ha proposto nel tempo diverse soluzioni: tra queste, il canale degli shock anticipati rappresenta il fulcro di questo lavoro. Secondo la scuola di pensiero inaugurata da Beaudry e Portier (2006), le notizie riguardanti la produttività futura (i cosiddetti shock di notizia, o news shocks) possono modificare il comportamento attuale degli agenti economici, innescando un’espansione anticipata a livello nazionale. Studi successivi, come Beaudry et al. (2007), hanno mostrato che questo meccanismo opera anche a livello internazionale: gli shock di notizie possono trasmettersi oltre confine e influenzare i cicli economici esteri, contribuendo alla co-movimentazione tra Paesi. La tesi contribuisce a questo dibattito sul piano empirico, proponendo un approccio alternativo all’identificazione dei news shocks, basato non sui dati economici ma sulle effettive fonti informative disponibili agli agenti, ossia gli articoli di giornale. Utilizzando tecniche di Natural Language Processing (NLP), vengono sviluppati tre metodi per estrarre da tali articoli un indice di sentiment legato alle aspettative sulla produttività futura. Le serie risultanti vengono poi incorporate in un modello VAR stimato su dati statunitensi, dove lo shock è identificato tramite un approccio a variabili strumentali (Proxy SVAR), al fine di valutare come i news shocks così identificati influenzino le principali variabili macroeconomiche domestiche. Infine, lo shock identificato viene testato all’interno di un modello VARX applicato al Canada, con l’obiettivo di valutare la trasmissione internazionale dei news shocks e il loro potenziale ruolo come canale di sincronizzazione dei cicli economici.
An NLP-based approach to international business cycle synchronization
ZANETTA, EDOARDO;Mancini, Francesco
2024/2025
Abstract
The thesis examines the phenomenon of business cycle synchronization, which is the tendency of different economies to move in a correlated way over time, and the difficulty of traditional macroeconomic models in reproducing this pattern. This discrepancy arises from the fact that standard open economy business cycle models (e.g. Backus et al. (1992)) predict negative correlations between countries’ outputs, whereas empirical data reveal positive ones. To address this mismatch, the literature has proposed different solutions over the years: among them, the anticipated-shock channel constitutes our focus. According to the school of thought started with Beaudry and Portier (2006), news about future productivity (the so-called news shocks) can alter the current behavior of economic agents, triggering an early expansion domestically. Subsequent works, such as Beaudry et al. (2007), have shown that this mechanism also operates at the international level: news shocks can be transmitted across borders and shape foreign business cycles, contributing to cross-country co-movements. This thesis contributes to this debate from an empirical side, proposing an alternative approach to identifying news shocks, based not on economic data but on the actual information sources available to agents: newspaper articles. Using Natural Language Processing (NLP), three methods are developed to extract from these articles a sentiment index related to expectations about future productivity. The resulting series are then incorporated into a VAR model estimated on U.S. data, with the shock identified through an instrumental variable approach (i.e. Proxy SVAR), in order to assess how news shocks identified in this way affect key domestic macroeconomic variables. Finally, the identified shock is tested within a VARX model applied to Canada, with the aim of evaluating the international transmission of news shocks and their potential role as a channel for business cycle synchronization.| File | Dimensione | Formato | |
|---|---|---|---|
|
2025_12_Mancini_Zanetta_Executive_Summary.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Executive Summary
Dimensione
436.96 kB
Formato
Adobe PDF
|
436.96 kB | Adobe PDF | Visualizza/Apri |
|
2025_12_Mancini_Zanetta_Thesis.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Testo della tesi
Dimensione
1.59 MB
Formato
Adobe PDF
|
1.59 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/10589/247315