ROE Term Structure: Reading the Shape of Future Profitability

The Question

An analyst predicts a company will deliver 15% ROE this year, 18% next year, and 20% in year three. Another company also trades at 15% current ROE, but forecasts show 15% → 13% → 11%. Both have the same starting point—why does the market price them differently?

Because the trajectory matters as much as the level. A rising ROE forecast curve signals improving competitive positioning, operating leverage kicking in, or margin expansion. A declining curve warns of peak earnings, intensifying competition, or unsustainable current profitability.

Traditional quant models use single-point ROE estimates (current or 1-year forward). But the entire term structure—the shape of the multi-year forecast curve—contains information about business quality and inflection points that single-point metrics miss.

Can you extract alpha from the shape and evolution of ROE forecast curves?

The Approach

We collect multi-year ROE forecasts from analyst consensus (covering years t+1, t+2, t+3) for every stock with coverage. For each stock, we construct an ROE term structure showing the trajectory of profitability expectations.

Shape classification:

• Rising curve (upward slope): ROE(t+3) > ROE(t+2) > ROE(t+1) → growth story, improving returns
• Flat curve (neutral slope): ROE stable across years → mature, steady-state business
• Declining curve (downward slope): ROE(t+1) > ROE(t+2) > ROE(t+3) → peak earnings, deteriorating returns

We also calculate second-order features:

• Curvature: Is the slope accelerating (getting steeper) or decelerating?
• Volatility: How much do forecasts differ across years? High volatility suggests uncertainty.

Dynamic signals come from tracking changes in term structure shape over time:

• Inflection from decline to rise: ROE forecasts were dropping, now they're rising → turnaround signal
• Inflection from rise to decline: ROE forecasts were climbing, now they're flattening or falling → peak signal
• Steepening vs. flattening: Is the gap between near-term and long-term ROE widening or narrowing?

These dynamic signals capture analyst expectation revisions in a more granular way than simple "forecast upgrades"—they show how the narrative is evolving across the entire forecast horizon.

The Finding

Term structure shape strongly predicted returns:

• Rising ROE curves (positive slope): Annualized alpha ~8-10% in long portfolios
• Declining ROE curves (negative slope): Annualized underperformance ~6-8%
• Flat curves: Near-market returns

The effect was robust across market cap segments and industries—rising ROE trajectories signaled quality regardless of sector context.

Dynamic signals (shape changes) were even more predictive:

• Curve inflection from declining to rising: +12-15% annualized alpha—these are turnaround stories the market under-appreciates initially
• Curve inflection from rising to declining: -10-12% annualized alpha—peak earnings warnings that investors ignore until they materialize

Why changes matter more than levels: Static term structure shape is somewhat known to the market—growth stocks already trade at premiums reflecting rising ROE expectations. But when the shape changes—analysts revise their multi-year outlook—there's information lag. It takes time for investors to adjust valuations to reflect the new trajectory.

Steepening curves (widening gap between near-term and long-term ROE) predicted outperformance: This signals that analysts see accelerating improvements, not just linear growth. The market often prices in the near-term forecast but under-appreciates the long-term trajectory.

Combining with valuation enhanced performance: Rising ROE curves at low P/B ratios (value + improving fundamentals) generated the highest risk-adjusted returns—classic GARP (growth at reasonable price) territory.

Try It Yourself

This strategy requires access to multi-year consensus ROE forecasts (t+1, t+2, t+3), which most data providers (Bloomberg, FactSet, Wind) offer for covered stocks.

Implementation steps:

1. Calculate term structure features: For each stock each month, compute slope (ROE(t+3) - ROE(t+1)), curvature, and volatility of the ROE forecast curve.

2. Track changes over time: Store historical term structure shapes and calculate month-over-month deltas to capture inflections and steepening/flattening dynamics.

3. Filter for forecast coverage: Focus on stocks with analyst coverage for all three forecast years (t+1, t+2, t+3). Thin coverage creates noisy signals.

4. Combine with quality screens: ROE term structure works best when layered with profitability quality filters (cash flow alignment, margin stability) to avoid value traps where rising ROE forecasts reflect unsustainable cost-cutting.

5. Rebalance quarterly: Term structure shapes evolve slowly (analysts revise annually or semi-annually). Monthly rebalancing adds turnover without meaningful signal improvement—quarterly is sufficient.

Practical nuances:

• Coverage bias: Small-cap stocks often lack multi-year forecasts. This strategy naturally tilts toward large/mid caps with deep analyst coverage.
• Forecast accuracy issues: Analysts are notoriously bad at long-term forecasts (t+3 estimates often miss badly). But directional accuracy (rising vs. declining trajectory) is much better than point accuracy—which is all this strategy needs.
• Sector effects: Cyclical sectors (materials, industrials) show more pronounced term structure dynamics (boom-bust ROE curves) than stable sectors (utilities, staples). Apply sector-neutral overlays to isolate stock-specific signals.

This approach is particularly valuable for fundamental long-term investors building quality-growth portfolios. It systematically identifies companies where profitability inflections are underway but not yet fully priced.

Want to integrate ROE term structure analysis into your fundamental screening process? Book a call to discuss forecast data sourcing, signal construction, and portfolio integration.

Source: 东北证券《根据ROE预测值期限形态及其变化进行选股》 (2023-05-21).