1 of 5

Xiamen University Finance Workshop�Shanghai (online), 2022/07

Labor flow shocks matter for asset pricing

Jian Chen Chunmian Ge Jiaquan Yao Guofu Zhou

Discussant：Zhuo Chen

2 of 5

Motivation and Overview

This paper constructs a labor flow measure from Linkedin resumes for the U.S. public firms and use this measure to predict aggregate market returns in the time series.
Findings:

Unexpected shocks to net labor flow positively predict market excess returns.
The return predictability of labor flow shocks remains out-of-sample and cannot be explained by existing labor-related or macroeconomic return predictors.
Expected labor flow level does not have predicative power for aggregate market.
The return predictability is likely to be driven by investors’ economic condition concerns as measured by Chicago Fed National Activity Index, Small Business Conditions Index, and optin-implied tail risk proxies.

Labor market flow data at monthly frequency using Linkedin resumes while other similar data are at annual frequency.
Provide two possible explanatory channels for the return predictability: expected cash flows and economic condition concerns, while the latter is supported by data.
Potential more application on the cross sectional asset prices.

Comment 1: What drives the unexpected labor flow shocks?

But the authors write in the draft: “We then determine whether expected cash flows drive labor flow shocks…” and “we examine whether the concern for future economic conditions drive firms’ hiring decisions…”
Shall we have next-period labor flow shocks as the dependent variable but the expected cash flows/economic conditions as the explanatory variables?
Besides the lead-lag analysis, contemporaneous regression may also help us understand the underlying driving force of unexpected labor flow shocks:

Fundamental-based macroeconomic variables, such as CPI, employment, GDP growth, consumption, investment, economic policy uncertainty, etc.
Behavioral variables such as sentiment index or change in investor sentiment.
Market-based return spread such as various characteristics-based long-short portfolios (anomalies)

After orthogonalizing w.r.t those variables, does the residual still have predictive power?�

Comment 2: Relationship with the aggregate expected investment growth?

Li et al. (2021 JME) construct a bottom-up measure of aggregate expected investment growth (AEIG) from individual firms and find that this measure negatively predict future market returns, the predictability of which is partially driven by time-varying risk premium.

What is the relationship of the unexpected/expected labor flow shock and AEIG?
Labor hiring seems to be part of firm investment (the human capital part, but also comes with physical investment), than what is the difference and connection of these two measures and why do they yield opposite predictive power? Especially given that one measure is expected and the other is unexpected?

Comment 3: What is the cross-sectional pricing implications?

As the authors have firm level expected/unexpected labor flow measures, they can form labor-flow sorted portfolios and examine whether higher unexpected labor inflow predict positive (or negative) future stock returns.
If they do find cross sectional pricing power for labor flow, they may also relate their findings to investment-based asset pricing literature (e.g., q-model, gross profitability, FF5, etc.) and investigate whether their labor-based proxy provides additional information compared to those investment-based characteristics.
Does the (potential) cross sectional pricing power come from stocks’ systemic risk exposure variation or mispricing differences?
Additional comments:

I would expect the time series predictive power is stronger for IT/finance industries.
Is the measure biased toward younger population and whether such bias affects the validity of the measure?
What is the predictive power at annual frequency?