Ways of Projecting Earnings
- Estimate future earnings growth rate (e.g. calculate CAGR of historical earnings growth, or calculate CAGR of historical equity growth rate), project ahead. [Note: see previous post here to know why you can use equity growth rate.]
- Using ROIC / ROC / ROE on the base, adding back retained earnings back to the base, repeat.
- Project sales growth (using historical long-period CAGR), assume a FCF margin (using historical average), project.
- Project sales growth (using historical long-period CAGR), project gross margin, project working capital schedule, project depreciation schedule, project capex schedule, project tax rate, etc. etc. You get the idea =)
It seems that methods 3 & 4 are commonly used on Wall Street. Bruce Wasserstein in his book Big Deal highlighted that for projecting unlevered free cash flow for DCF, “Margins and growth rates should be scrutinized”. I have also seen JPMorgan financial models projecting working capital schedules, etc.
But what really governs earnings? that will tell us how earnings should be projected.
- What happens in reality is indeed the growth of sales (due to both inflation and GDP growth), incurring of expenses, and ending up with some free cash flow.
- ROIC / ROC / ROE are simply different ways of displaying the results, similar to P/E and bond yields, they are the results at a point in time, not the determinants.
- Why then, does Buffettology use ROE like a determinant to project earnings moving forward?
- You can only use ROIC / ROC / ROE as determinants if earnings are added to the base (i.e. what you use for the numerator is added to the denominator), and that the process has been happening for an extended period of time (e.g. the company has been running for 20 years), then the metrics (ROIC/ROC/ROE) would have gradually adjusted to the true earnings growth rate (you can see that by doing up a quick spreadsheet). To know whether you can project earnings using ROE, you should also check that the future projection curve “fits” with the historical earnings to know if the model is appropriate.
In that case, does it still make sense for people to project future earnings growth using historical earnings (or equity) growth rates?
- Yes, but if you do that, you make the implicit assumption that the payout ratio and the ROX will remain the same moving forward.
- For companies just starting out, that cannot be the case, coz the payout ratio and ROX will definitely change (e.g. payout ratio might increase as the company matures, ROX will also go down with slowing “growth”).
- Can only do with stable companies, and also after you have checked the payout ratio and ROX.