Can we find out more about the trajectory of a country’s economy by looking past the headline unemployment numbers and delving deeper into the job prospects of various demographic groups? For example, does job loss amongst higher income or higher education individuals tell us more than job loss amongst the lower educated or lower Income? Or, is a lack of job security amongst a family’s primary earner a more troubling sign than if less people depended on that income?
To answer these questions, we undertook a study using data from The Thomson Reuters/IPSOS Primary Consumer Sentiment Index. The PCSI contains data on 24 different countries and is composed of 11 different survey questions over a range of topics related to economic sentiment. In this study we looked at the demographic breakdowns of the 3 employment related questions shown in the table below. We wanted to see how informative the responses of the various demographic groups were at picking future over or underperforming equity markets.
Adopting the same methodology used in our October research note, we constructed signals for each country from the demographic breakdowns of each PCSI component. The signals we tested were the one month change in net sentiment for each demographic group. Here “net sentiment” is calculated as the percent of respondents answering positively minus the percent of respondents answering negatively to a particular survey question. We then tested these signals in a monthly country rotation across 24 countries and measured the long-short portfolio spreads and information coefficients (defined as the rank correlation between the signals and forward one month returns).
The figures below show the Information Coefficient and the tertile spreads (8 of the 24 countries per portfolio) of the one month change signals. In each figure, the last row demonstrates the performance of the overall signal across all demographics combined. Though this total shows positive metrics for all three PCSI components, it is easy to improve on this performance if we only consider certain demographics.
Much of the variation across demographics follows our previous intuition. For example, for all three of the components, the chief income earners outperformed the non chief income earners. In fact, the best performing long-short signal was the ‘Job Security’ component amongst chief income earners which demonstrated 9.6% annualized spread returns and an IC of .07. Also, the low income group underperforms the medium and high Income group and the married demographic outperforms the non married demographic. These results suggest that job loss amongst the demographics with higher income and/or more people dependent on that income have a greater impact on returns of equity markets.
Another one of the more interesting things to notice is the difference that occurs when we consider the temporal variation of the three PCSI components. One of the components asks about past job loss while another asks about likelihood of future job loss. Performance of the education level demographics shifts starkly between the past and future job loss questions. When considering past job loss, the lower education demographic outperforms the higher education demographic. When considering the prospects of future job loss, however, the higher education demographic outperforms indicating that lower education groups having lost their jobs and higher education groups worrying about future job loss are two of the more telling indicators of economic health.
Lastly, the figure below shows the profit and loss curve of one of our best performing long-only signals– the employed demographic and job security component. This is the performance of a 3 country long only portfolio so we can compare it to the long-only equal weighted average of all countries. Over our backtest period this signal nearly doubles the return of the all country portfolio.
Once again, we have demonstrated value in using the PCSI data in top down investment approaches. At the same time we still see much value left to be uncovered as we have only explored three of the components here. We encourage our readers to test the PCSI data for themselves to see if they can incorporate it in their investment process. To begin to trial the data please contact your Refinitiv representative.