Sort firms on the capex they have committed to rather than the capex they have booked, and the sign of the investment effect turns over.
The investment factor is one of the most reliable results in asset pricing and it runs the other way: firms that invest heavily earn lower subsequent returns. This paper forecasts next year’s capital expenditure growth from cheap accounting and price data — expected investment growth — and finds high-EIG firms beat low-EIG firms by about 17% a year.
The reconciliation is the lag. A plan is not a spend: between the decision and the disbursement, the firm carries a committed future outflow that behaves like debt, and that embedded leverage raises the required return on equity. The forecast itself comes from a LASSO that shrinks eleven candidate predictors to six — Tobin’s q does not survive. A companion paper shows the same measure, aggregated, predicts the market with the opposite sign.
The transferable move has nothing to do with capex: any accounting quantity decided long before it is reported has a window in which the plan is visible and the number is not. Note also the direct challenge — Chen and Liu (2023) decompose the premium and ask whether the leverage is embedded or simply literal.
Source papers
Li, J., Wang, H., & Yu, J. (2021). The expected investment growth premium. Financial Management, 50(4), 905–933.
DOI 10.1111/fima.12340 · SSRN 3195406 — posted 27 June 2018, last revised 30 July 2020
Personal reading note. Written while studying the paper — not peer reviewed, not a replication, and not a substitute for the source. Every claim above belongs to the authors; any error in restating it is mine. Follow the link. The 17% is reported without risk adjustment, weighting scheme or size screen — check it against the full text.