A0856
Title: Value of non-traditional data sources for official statistics perspective from developing economy
Authors: Syeda Rabab Mudakkar - Lahore School of Economics (Pakistan) [presenting]
Abstract: Web scraped data has recently been incorporated into the inflation estimates by organizations such as the Statistics Bureau of Japan, Australian Bureau of Statistics and Statistics New Zealand, among others, using the average price of a commodity over a specific period of time. However, still, the cases are few. Due to a lack of technology infrastructure, a restrictive legislative environment, data access and quality issues, and other factors, the majority of national statistical offices are wary of ``big data'' exploration. This is a first attempt to examine the challenges and opportunities of web scraped price data for a developing economy. The daily information of one hundred thousand durable and non-durable products for a time period of one year is collected. The results reveal that due to the low penetration of online retail, the market-clearing phenomenon works less efficient compared to brick and mortar retail. Further, the data exhibits the presence of features such as a higher increase in price size relative to reduction, skewness, kurtosis etc. Overall sticky-price behavior is observed but few ``random-signals'' light up in situations when the size of price increases is far from zero. This indicates the presence of Calvo price-setting behavior in internet retail in developing economies.
