Nowcasting Unemployment Rates with Smartphone GPS Data

Abstract

Unemployment rate is one of the most important macroeconomic indicators. Central governments and market participants heavily rely on the index to assess the economies. However, official statistics of unemployment rate are released infrequently with substantial delay. Prediction of official statistics of labor market will be helpful for these authorities as well as private companies and even workers. In this paper, we combine massive location data coming from smartphones and mixed data sampling (MIDAS) techniques to predict current unemployment rate in Japan. We found GPS data is very useful to predict the status of labor markets.

Aggregate Learning for Mixed Frequency Data

“…They suggested the potential predictive power of search query data. While most of the studies utilize web data represented by Google trends, Moriwaki (2020) [25] uses smartphone GPS data to predict unemployment rates. Unprecedented COVID-19 pandemic reminds the importance of economic nowcasting [21].…”

“…Nowcasting often takes advantage of alternative data, non-standard data such as search queries, location data, SNS data, and satellite images [3,6,8,14,25,37]. These data are suitable for economic nowcasting because of their high frequency.…”

“…However, it is hard to understand the whole picture of the economy itself because the relationship between these alternative data-based statistics and familiar economic indices such as GDP and unemployment rates is unclear. To fill this gap, many nowcasting/forecasting models that use high-frequency data such as Google search query to predict economic variables are proposed [3,5,6,25,32,35,37]. Combining high-frequency data like Google search query and low-frequency data such as monthly unemployment rates and quarterly GDP has been actively studied by economists [9,10].…”

A Machine Learning Approach for Detecting Unemployment Using the Smart Metering Infrastructure

“…Methods for detecting unemployment levels commonly rely on longitudinal data sets created using surveys from official authorities along with statistical techniques, such as Markov models [21]. Additionally, unemployment rates and their consequences at multiple scales have been observed using data from smart phones, including call detail records (CDRs) [22], Global Positioning System (GPS) log data [23] or Google searches [24]. These types of logs are comprised of time series data (e.g.…”

“…These types of logs are comprised of time series data (e.g. weekly, daily or, hourly) that has become increasingly prevalent for supporting economic statistics-based research, with the aim of modelling unemployment rates [23], [24]. However, little or non-existing literature is available on statistical analysis using non-traditional data such as smart meter data to identify unemployment levels.…”