What would you consider a good ROE?

Higher is always better than lower. But for a rough target, there are many firms that statistically do well with ROE north of 30%.
For a sense of the distribution, in the VL universe in the last ~20 years, using the annual "Return on Shareholders Equity" field
about 95% of covered firms have a stated ROE of some kind
about 83% of covered firms have positive ROE
about 35% of covered firms have ROE > 15%
about 10% of covered firms have ROE > 30%


So, either >30% or negative.

You may wonder why negative. The idea is that you want a company that can make excellent money with the fewest assets. A few very good firms can make money quite reliably with no net assets at all. So if a firm has positive earnings (and they aren't just a one-time anomaly) and negative book per share, you get a negative ROE but (statistically) a good investment. Positive earnings divided by negative net assets gives a negative ROE. Not to be confused with firms with negative earnings on positive assets! Avoid those.

e.g, you'll see good average performance from firms with negative book but positive current earnings, positive trailing earnings, and positive EPS growth forecast. In effect, and infinite ROE.
You'll get a lot of tobacco firms. Great financially in the past but I screen those out.


If looking for high ROE, don't forget that there are firms with high ROE and bad situations. An easy way for a firm to get a high ROE is to borrow way too much money and pay it all out as a dividend, leaving them vulnerable. What used to be called a "recap" long ago, but mostly known as LBO victims these days. Be sure to screen for sustainable debt levels before checking for high ROE. I check long term debt < N times typical net earnings. Another way to get a superficially high ROE is to have something weird on the balance sheet, like a truly gigantic loss in the past that wiped out the bookkeeping equity while leaving the actual business unchanged. That one is hard to screen for. It's one reason people also look at return on *tangible* assets rather than just return on net assets.


An interesting side effect: since the best firms have high ROE figures, and since many firms of all types trade at pretty similar multiples of earnings, you often see the best businesses trading at the highest P/B ratios. (same earnings, same price, lower book value, means higher P/B). This is why low P/B doesn't work as a value screening step...the very best firms need very few assets to make their money. Seeking low P/B firms is mostly the same as seeking firms with poor business economics: they need a lot of assets to earn each buck of profit. Counterintuitively, seeking the few firms with the highest P/B works pretty well as a screen. (not too many picks, as there aren't too many fantastic companies). This doesn't work because it's merely picking overpriced momentum picks that are going to keep zooming just a while longer...it often picks very good underlying businesses. In the VL world, top 10 monthly by P/B beat SPY by about 10%/year in the last 20 years, no other criteria at all.

Jim