We have compiled hundreds of thousands of news articles in Chinese to build the first ever NIPI for China. The methodology is the same as other regions and this post recaps the main features.
In a nutshell
The News Inflation Pressure Indices (NIPI) are real-time indicators tracking changes in near-term inflation trends. They rely on scanning through news sources using state-of-the-art language models.
The NIPI reads like a PMI or an ISM: 50 is neutral, above 50 means upside pressures are building up.
China NIPI time series
Frequency
Daily
SectorsHeadline, Core, Energy and Food inflation
History start1/1/2018
DatabasesNews Inflation Pressure Indices (NIPI), News Volume Indices (NVI) and Raw Entropy measures
Processed Chinese news articles
Total Chinese articles analysed: 759052
Of which, detected as relevant to China inflation: 13670
Average processed daily: 500
Average relevant daily: 10
Unique news sources in Chinese: 877
Sample period 1/1/2018-7/2/2022
Delivery
Similar to our other NIPI data:
- CSV file end point
- Online NIPI visualization app
- Data released at 10am CET for the previous day
- No revision
Use cases
- Monitor China inflation in real time
- A timely input into global goods prices
- Subdued China Inflation? Our fresh Chinese NIPI indicators provide new insights
NIPI Databases Background Info
Developed economies coverage
Australia, Canada, Euro area, France, Germany, Italy, Spain, UK, US
Emerging economies coverageArgentina, Chile*, China, India, Mexico, Peru*, South Africa
SectorsHeadline, Core, Energy, Food, Telecom services**, Wages**
LanguagesChinese, English, French, German, Italian and Spanish
First live commercial release (English news)15/11/2020
* Experimental ** Experimental for some countriesReferences
One million news project (and counting)
First NIPI release: we explain how the data is compiled.
The NIPI as a predictor of US inflation: Nov-2021 update
Our news-based indicators have correctly predicted the recent US CPI path. And they continue to point towards persistently strong price pressures.
Whose sentiment?
One or two annoying things about sentiment indicators in macro
Let the robot read the news for you
Machine learning can considerably enhance information gathering, within pretty much any investment process.
