Do accounting signals predict stock returns in Sweden? A pre-registered test on survivorship-free data, 2001–2022

I test whether ten well-known accounting signals predict next-month returns on the main list of theStockholm Stock Exchange from 2001 to 2022. Returns come from a survivorship-free price filethat includes delisted firms and their final payoffs; accounting data come from annual reports in theSerrano database, used only six months after the fiscal year end. The design, signals, thresholds andverdict rules were pre-registered and locked before any signal was related to returns. After excludingfinancial firms and microcaps, about 180 firms enter each of 251 monthly Fama-MacBeth regressionsthat control for size, book-to-market, momentum, short-term reversal and beta. No signal clears thepre-registered bar of |t| > 3. Five signals receive weak support (t between 2.31 and 2.65): operatingand gross profitability, ROE, the earnings yield and EBIT/EV. Accruals, asset growth, the F-score,book-to-market and the dividend yield show no evidence. Effects are no weaker after the originalpublication dates. Including microcaps removes the effects, while adding First North and other listsstrengthens them to t above 3. A comparison block that applies the rules of a companion pre-registeredUSstudy (S&P 500, 2010–2020, where no signal passed) passes 3 of eight signals in Sweden: ROA,cash-flow profitability, earnings yield (E/P).Keywords: anomalies, accounting signals, profitability, value, Sweden, pre-registration, survivorshipbias. JEL: G12, G14, M41