via Marc Joffe:
Think Tank Identifies Potential Debt Crisis With Open Source App
— Macdonald-Laurier Institute publishes Canadian government bond study based on predictive analytics modeling software —
SAN FRANCISCO -- Oct. 18 – Canadian provinces face substantial long term credit risk, according to a study published today by the Macdonald-Laurier Institute, a non-partisan, independent think tank based in Ottawa, Ontario.
The study describes the outputs of a budget simulation software tool designed to estimate annual credit default probabilities in order to accurately rate government bonds. Findings were based on analysis of each Canadian province using the open source Public Sector Credit Framework (PSCF) software, according to the study’s author, former Moody’s Analytics Senior Director, Marc Joffe.
The PSCF modeling tool employs a multi-year budget simulation. Random number draws create an array of interest rate, inflation and growth scenarios, which then drive simulated revenue and expenditure estimates. For this study, fiscal simulations for each province consisted of one million runs over a 30-year time horizon.
While attention has been focused on immediate crises, such as the U.S. fiscal cliff, Eurozone debt and California municipalities, findings of the study indicate that Canadian provinces have virtually no near-term credit risk. However, analysis found that the affects of population aging and persistent deficit spending create substantial risk in the long term, with several Canadian provinces facing long-term default probabilities in excess of 50 percent.
Open Source Rating Agency Alternative
“This is the right way to rate government bonds. The analysis is based on a quantitative model and all the assumptions are open to review,” said Joffe. “The framework produces a default probability in numerical terms and then translates it to a letter grade like AAA or AA. In advanced economies, government financial statements and projections are public information. The way these inputs are used to rate governments does not require any secret sauce.”
The PSCF model uses a presentation layer in Microsoft Excel, with back-end processing written in the C programming language. The software is free and open source, and Excel-based PSCF models are fully transparent. Other researchers can install the Windows-based software, download the models and re-run the analysis themselves. Source code for PSCF is available at http://www.github.com/joffemd/pscf.
PSCF is one of a number of ongoing efforts to disrupt the rating agency model. In 2011, two mathematicians in Switzerland introduced Wikirating, a not-for-profit website designed to crowd source bond ratings. Two German-based organizations, Bertelsmann Foundation and Roland Berger, have announced not-for-profit credit rating agency initiatives as well.
The Macdonald-Laurier Institute study is the first time that alternative methods of this nature have produced peer-reviewed, academic quality analysis.
As Joffe noted in a recent blog post: “Academics can deliver the intellectual rigor missing from both status quo rating analysis and some of the alternatives we have been seeing. The thorough data collection procedures, advanced modeling techniques, and peer review practices employed by social scientists can raise the level of sovereign [and other government] risk analysis.”
The full report can be viewed on the Macdonald-Laurier Institute’s website:
Add a Comment