Seminario 539. Are debt sustainability models useful for forecasting? In- and out-of-Sample Evidence from 31 countries

Este evento hace parte de la serie:
Otros seminarios y talleres
  • Director del Instituto de Investigaciones Económicas, Universidad de Osnabrück, Alemania

 

Entrada libre. Indispensable inscribirse en el siguiente vínculo: Inscripciones

 

Hora: 12:00 p. m. (refrigerio) y 12:30 p. m. (inicio del seminario)
Tiempo de exposición: 12:30 p. m. a 1:30 p. m.
Lugar: Banco de la República, carrera 7 # 14-78, piso 13 (Sala de prensa), Bogotá-Colombia.
Idioma de la exposición: Inglés


Resumen del documento: A large literature in empirical public finance is applying time series techniques to historical data for individual countries and draws inference about public debt sustainability of individual countries. These methods include unit root test on primary deficits, co-integration between revenue and expenditure as well as, government reaction-functions with respect to past debt levels. In this paper, we take a systematic approach evaluating the in- and out-of-sample forecasting properties of various methods with regard to predicting sovereign debt crisis. We find that there is a substantial amount of type-one and type-two errors: In sample, for countries deemed sustainable by the indicators, we see many instances of heavy IMF-borrowing, lender of last-resort activities and sprints of inflation.  Even official defaults have occurred frequently during ‘sustainable’ periods. Unsustainable periods are not followed by a larger propensity to default. Out-of-sample, in a panel probit-regression analysis, we do not find any significant benefits for forecasting. This finding is robust to alternative time-series methods, sample windows and default definitions.

 

 

Si desea inscribirse a la lista de correos del Seminario semanal de Economía de Bogotá para recibir en su cuenta la información sobre nuestra programación, debe remitir un correo electrónico a la dirección seminariossemanales@banrep.gov.co con el asunto "Inscripción a la lista de correos", cabe resaltar que por esta cuenta no se reciben inscripciones a los seminarios.