Collaborative Research: Origins of Serial Sovereign Default

Abstract

Sovereign nations have consistently borrowed from international financial markets for centuries, and this foreign debt remains an important source of funding for advanced and developing countries to the modern day. What sustains the existence of this market when there is no formal legal means of enforcing repayment? Standard economic models predict that defaulting on sovereign debt is deterred by the threat of investors refusing to engage in future lending.

However, historically countries differ significantly in both their likelihood of default and ability to re-access financial markets after default. This research will examine what leads some countries to become serial defaulters – experiencing repeated cycles of borrowing and default – while others do not. This project will assemble a new database on sovereign borrowing and default consisting of both numerical and textual data.

These data will document detailed default and debt issuance history as well as the perspectives of the general public, investors, and analysts. The project will have potential policy implications for sovereign borrowing decisions around the world, by improving the understanding of the reasons for sovereign defaults and investor willingness to lend.

This research will investigate why some countries experience cycles of repeated borrowing and default. The project will build a new quantitative and textual database on sovereign borrowing, default, negotiations with lenders, and the economic and political circumstances relevant to sovereign debt. To assemble these data, the project will develop new adaptations of machine learning methods for natural language processing (NLP) to extract and analyze text.

The data will be obtained from the universe of articles from major financial newspapers, annual reports by investors, and periodical publications by financial analysts. The project will develop a new deep learning model to correct for common Optical Character Recognition (OCR) errors. The project will further construct new text libraries to aide in NLP content classification, which will allow determining how investors perceive the motive of the sovereign, such as economic or political, in issuing or defaulting debt.

The project will categorize countries as being a serial defaulter based on the frequency and duration of their defaults from 1820 to 1939. The research will then document how the political and economic circumstances of borrowing and default, including public, investor, and expert perceptions of the sovereign's motives, influence serial default decisions.

What determines sovereign default and investor willingness to lend is not only relevant for understanding the history of sovereign debt but has implications for ongoing policy debates over debt restructuring, payment moratoriums, and bailouts.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.