Online price indices : methodological improvements and applications to inflation forecasting and real exchange rate estimation
Date
2019-06
Authors
Bertolotto, Manuel I.
relationships.isContributorOfPublication
Cavallo, Alberto
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de San Andrés. Departamento de Economía
Abstract
El primer trabajo presenta una nueva metodología para calcular índices de precios al consumidor (IPC) con precios online ya que las utilizadas corrientemente muestran tendencias negativas anormales provocadas por la alta rotación de productos. Mientras que el enfoque tradicional para calcular el IPC busca un reemplazo cuando se descontinua un producto, la metodología de “closest-match” busca un artículo comparable cada vez que ingresa uno nuevo al mercado. El método propuesto es escalable y produce resultados similares al IPC tradicional para todos los países en la muestra (Alemania, Países Bajos, España, Reino Unido, y Estados Unidos). El segundo trabajo introduce a los índices de precios online como variables relevantes para pronosticar el IPC. Con la aplicación de la metodología propuesta, los pronósticos de inflación a un mes son más precisos que los publicados por Bloomberg y otros modelos estadísticos para los casos de Australia, Canadá, Francia, Alemania, Grecia, Irlanda, Italia, los Países Bajos, el Reino Unido y los Estados Unidos. Así mismo, al de la Reserva Federal de Philadelphia, realizado con tres meses de anticipación. El tercer trabajo utiliza una base de datos con frecuencia diaria para estimar que la tasa de reversión al equilibrio del tipo de cambio real (TCR) es inferior a un año. Hasta ahora, las estimaciones de esta tasa oscilaban entre un mínimo de dos a cinco años, lo cual era inconsistente con la teoría de paridad de poder de compra y la ley del único precio.
The first paper introduces a new methodology to calculate the consumer price index (CPI) using online prices that avoids its abnormal downward trend. The method reflects a change in paradigm for how old and new varieties of products are linked. While the traditional approach looks for a replacement when an item is discontinued, the closest-match approach searches for a comparable item every time a new product enters the market. The method is scalable and yields remarkably similar results to the traditional CPI for every country in the sample, namely Germany, the Netherlands, Spain, the United Kingdom, and the United States. The second paper introduces the online price index as a useful source of information to forecast the CPI. The baseline one-month forecast outperforms Bloomberg surveys of forecasters. The baseline specification also outperforms statistical benchmark forecasts for Australia, Canada, France, Germany, Greece, Ireland, Italy, the Netherlands, the United Kingdom, and the United States. Similarly, the quarterly forecast for the US inflation rate substantially outperforms the Survey of Professional Forecasters. The third paper uses a novel dataset with a daily frequency of price collection to measure that the mean-reversion rate of the real exchange rate is typically less than one year. This is the first study to show estimates in harmony with the law of one price theory. Until now, estimates suggested that the rate of mean-reversion was unreasonably slow, ranging from a minimum of two to five years.
The first paper introduces a new methodology to calculate the consumer price index (CPI) using online prices that avoids its abnormal downward trend. The method reflects a change in paradigm for how old and new varieties of products are linked. While the traditional approach looks for a replacement when an item is discontinued, the closest-match approach searches for a comparable item every time a new product enters the market. The method is scalable and yields remarkably similar results to the traditional CPI for every country in the sample, namely Germany, the Netherlands, Spain, the United Kingdom, and the United States. The second paper introduces the online price index as a useful source of information to forecast the CPI. The baseline one-month forecast outperforms Bloomberg surveys of forecasters. The baseline specification also outperforms statistical benchmark forecasts for Australia, Canada, France, Germany, Greece, Ireland, Italy, the Netherlands, the United Kingdom, and the United States. Similarly, the quarterly forecast for the US inflation rate substantially outperforms the Survey of Professional Forecasters. The third paper uses a novel dataset with a daily frequency of price collection to measure that the mean-reversion rate of the real exchange rate is typically less than one year. This is the first study to show estimates in harmony with the law of one price theory. Until now, estimates suggested that the rate of mean-reversion was unreasonably slow, ranging from a minimum of two to five years.
Description
Fil: Bertolotto, Manuel I.. Universidad de San Andrés. Departamento de Economía; Argentina.
Keywords
Consumer price indexes -- Econometric models. , Price indexes -- Econometric models. , Inflation (Finance) -- Forecasting -- Econometric models. , Índices de precios al consumidor -- Modelos econométricos. , Índices de precios -- Modelos econométricos. , Inflación (Finanzas) -- Pronósticos y proyecciones -- Modelos econométricos.
Citation
Bertolotto, M. I. (2019). Online price indices : methodological improvements and applications to inflation forecasting and real exchange rate estimation. [Tesis de doctorado, Universidad de San Andrés. Departamento de Economía]. Repositorio Digital San Andrés. http://hdl.handle.net/10908/16699