(Bloomberg Economics) – – With the Covid-19 recession rendering many traditional indicators outdated before they are published, Bloomberg Economics is using a set of high-frequency, alternative data to build daily activity indicators. Here’s what the latest data show:
- As expected, economic activity in several of the world’s largest advanced economies plummeted over the Christmas holidays.
- An increase in Covid-19 infections and stricter containment measures in November and early December, led activity to drop sharply in the third week of December. A decline in mobility and other alternative data indicators is not unusual over the Christmas and New Year holidays. But in some countries, such as Germany and Italy, very strict containment measures imposed shortly before Christmas added to the weakness.
- Japan is the only bright spot among advanced economies. Activity held steady at more than 90% of its pre-virus level both because of a large measure of success in controlling the virus and the lack of the seasonal factor of Christmas.
- Our indicators provide a high-frequency guide to the pace of recovery across countries, and attempt to improve the signal-to-noise ratio in alternative data. They are not a substitute for the detailed country view.
The activity indexes are estimated using a dynamic factor model. This methodology extracts an unobservable latent common factor of the underlying high-frequency data in the spirit of Stock and Watson. The model is estimated with daily figures from Jan. 1, 2020 to Dec. 29, 2020.
The high-frequency statistics we use have some obvious advantages — providing a more timely read than traditional data series. They also come with some caveats attached:
- The high weight of travel and mobility indicators may lead to overweighting this type of activity in the index.
- The index is not fully comparable across countries as we partly use different indicators for different countries, and the relationship between mobility and output likely varies between countries. A complete set of sources is shown in the table below.
- We don’t have a long enough time series to map the relationship between the activity indicators and GDP.
- In a dynamic factor model, component weights adjust as new data become available. Future updates of the index will likely result in small backward revisions to historical readings.
- Our model nowcasts missing data points. That can mean backward revisions of past readings as new data become available.
Comments ( 0 )