Augmenting Household Expenditure Forecasts with Online Employee-generated Company Reviews.

Efthymia Symitsi,Antonios Karatzas,Panagiotis Stamolampros

doi:10.1093/poq/nfab017

Abstract

We assess the ability of online employee-generated content in predicting consumption expenditures. In so doing, we aggregate millions of employee expectations for the next six-month business outlook of their employer and build an employee sentiment index. We test whether forward-looking employee sentiment can contribute to baseline models when forecasting aggregate consumption in the United States and compare its performance to well-established, survey-based consumer sentiment indexes. We reveal that online employee opinions have incremental information that can be used to augment the accuracy of consumption forecasting models and inform economic policy decisions.

Highlights

Consumer spending is a key engine that drives economic growth accounting for almost 60% of the Gross Domestic Product worldwide (The World Bank 2020)
The purpose of this study is to examine whether an index that aggregates employee expectations for the near-term business outlook of their employer, named Employee Sentiment (ES), predicts changes in private consumption
We assess the informational value of data generated by an important group of stakeholders with unprecedented potential, that of employees, in forecasting private consumption

Summary

Introduction

Consumer spending is a key engine that drives economic growth accounting for almost 60% of the Gross Domestic Product worldwide (The World Bank 2020). The ES index uses the business outlook rating (enabled after May 2012), allowing employees to evaluate the 6month ahead prospects of their employer as “Better”, “Same” or “Worse” This information resembles the information in widely applied survey-based consumer sentiment indexes (MCI, CCI) (see Bram and Ludvigson 1998; Ludvigson 2004). These survey-based indicators are restricted to a limited sample of participants, while ES uses online information arriving from millions of employees from thousands of companies across all sectors Employees express their expectations about the business outlook of their employers, while participants in the MCI and CCI surveys are asked, besides their own family conditions, to predict overall business and market conditions. The findings remain consistent with ES adding value to both parsimonious and augmented consumption forecasting models

Discussion and conclusions

Findings

Robustness Checks