Big data project may provide better inflation forecasts
News In a pilot project, three economists at the Riksbank have collected prices for fruit and vegetables from the Internet using so-called web scraping. The results indicate that the method may provide more accurate inflation forecasts over the short term and form an example of the possibilities entailed by the increased supply of data.
"New technology is making it possible to better understand the development of prices and to make better inflation forecasts over the short term. There is still a lot of work to be done, but I think this method has potential," says Markus Tibblin, senior economist at the Riksbank and one of the participants in the project.
Project members Isaiah Hull, Markus Tibblin and Mårten Löf are all economists at the Riksbank.
Forecasts of the development of inflation are important for the Riksbank as the target for monetary policy is to hold inflation at around 2 per cent per year. While it is true that monetary policy decisions are based on the development of inflation over a longer period, making reliable forecasts of future inflation requires having a good understanding of the current rate of inflation and high accuracy regarding the development of inflation over the coming months.
Some sub-components of the inflation figures fluctuate more than others and are therefore difficult to predict, for example prices of fruit and vegetables, energy and air travel. The project group therefore developed an automatic process called web scraping which collected data from the internet, every day for almost two years, on prices for oranges, bananas, peppers, apples, cucumbers, white cabbage, grapes, cauliflower, pears, leeks and tomatoes.
The project has demonstrated that it is possible to collect prices from the Internet in an efficient manner and that the prices collected can provide information to assist in making price forecasts for fruit and vegetables.