Evaluating EPS Estimates

For several years, I’ve been tracking long-term EPS estimates by 7 different analysts, from:

• CNNMoney via FactSet CallStreet (next 5 years),
• YahooFinance via Thomson Reuters (next 5 years),
• Zacks.com (next 5 years),
• Reuters.com (unspecified long term),
• Morningstar via FactSet (next 5 years),
• Value Line (next 3-5 years), and
• S&P via BI data subscription (next 5 years)

 Here are examples of these estimates for six companies: Qualcomm (QCOM), Johnson & Johnson (JNJ), Costco (COST), Microsoft (MSFT), Walgreen (WAG), and Hewlett-Packard (HPQ).  The estimates go back 7 or so quarters and allow us to see which is most often high, or low, or out-of-whack, and how much they differ from each other on the same date.

Two trends are immediately obvious in the table below:

• The many negative values (in red) by Take Stock, an on-line program at the StockCentral website, and how they are radically different from the seven mainstream EPS estimates;
• Value Line’s 6-7 estimates for QCOM, COST and WAG are all consistently low.

LONG TERM EPS ESTIMATES in %    (and low & high estimates by date)
	CNN Money	Yahoo Finance	Zacks	Reuters	Mstar	Value Line	S&P	AVER-AGE	Take Stock
QCOM
10-10-10	15.00	16.94	17.63	15.57	17.90 high	6.00 low	15.00	14.86	-16.20
06-22-10	15.00	19.22 high	17.60	15.80	16.90	5.50 low	16.50	15.22	-16.20
03-23-10	16.00	19.86 high	17.63	16.27	17.50	5.50 low	N/A	15.46	-16.20
12-06-09	17.00	17.30	18.81 high	17.04	17.90	6.00 low	N/A	15.68	-16.20
09-22-09	15.00	14.46	16.23 high	15.61	XXXXX	6.00 low	14.00	13.55	-02.00
06-22-09	15.00	13.14	15.95 high	15.31	XXXXX	6.00 low	15.00	13.40	-02.00
03-29-09	16.00 high	11.75	14.70	15.45	XXXXX	6.00 low	15.00	13.15	-02.00
JNJ
08-23-10	6.00     low	6.43	6.26	6.69	6.80	7.00  high	6.20	6.48	1.40 start     -2.74 end
05-31-10	7.00	6.98	7.47	7.34	7.20	8.00 high	6.80 low	7.26	-2.70 strt        -2.74 end
02-20-10	7.00	6.97	7.63	7.26	6.70  low	8.00 high	N/A	7.26	4.20
10-16-09	7.00    lo-tie	7.68	7.98         high	7.20	XXXXX	7.00  lo-tie	7.50	7.39	4.20
08-24-09	6.00  low	8.13	8.26	8.42  high	XXXXX	7.00	N/A	7.56	4.20
02-24-09	6.00  low	8.38	9.73           high	9.50	XXXXX	7.50	7.30	8.07	4.80
COST
08-21-10	13.00	12.95	13.22 high	12.86	12.80	7.50 low	12.80	12.17	-11.30
05-16-10	12.00	13.12	13.50 high	12.90	12.00	7.50 low	13.00	12.00	-11.30
02-01-10	13.00	13.46	13.52 high	13.02	12.10	7.50 low	13.00	12.23	-11.30
08-09-09	13.00 high	11.55	11.61	11.50	XXXXX	5.50 low	N/A	10.63	9.10
05-04-09	13.00 high	12.25	12.39	12.11	XXXXX	5.50 low	12.90	11.36	9.10
02-05-09	13.00 hi-tie	12.89	12.14	12.62	XXXXX	9.00 low	13.00 hi-tie	12.11	9.10
11-04-08	13.00	13.08	12.60	12.80	XXXXX	11.00 low	14.00 high	12.75	N/A
	CNN Money	Yahoo Finance	Zacks	Reuters	Mstar	Value Line	S&P	AVER-AGE	Take Stock
MSFT
08-16-10	12.40	10.37   low	11.69	11.76	12.00	12.50 high	12.00	11.82	5.40 start  0.96 end
05-17-10	12.00 high	08.55  low	11.40	11.23	10.90	10.50	10.00	10.65	-11.00
02-16-10	12.00 high	11.40	11.48	11.42	11.00	10.00 low	11.00	11.19	-11.00
11-18-09	11.00	11.14     hi-tie	11.09	11.14      hi-tie	10.90	10.00 low	N/A	10.88	-11.00
08-18-09	11.00	10.17	10.62	10.61	XXXXX	10.00 low	12.00 high	10.73	-11.00
06-04-09	11.00	10.32	11.44	10.61	XXXXX	10.00 low	12.00 high	10.90	9.30
02-17-09	11.00	10.22   low	11.14	10.70	XXXXX	12.00 high	11.00	11.81	9.30
WAG
10-03-10	15.00 high	12.95	13.14	13.16	13.00	08.50 low	13.00	12.68	-06.00
08-02-10	13.50	13.29	13.14	13.42	13.60 high	09.00 low	15.00	12.78	-06.00
07-06-10	14.00 high	13.87	13.75	13.90	13.80	10.00 low	13.50	13.12	-06.00
12-20-09	14.00 high	14.22	14.25	13.92	13.30	10.50 low	13.50	13.52	06.80
11-20-09	14.00 high	12.55	13.63	13.10	XXXXX	09.50 low	13.50	13.05	06.80
06-25-09	15.00 high	12.00	13.03	12.89	XXXXX	08.50 low	13.50	12.49	06.80
03-23-09	14.00 high	11.55	12.86	12.61	XXXXX	07.50 low	13.50	12.00	06.80
HPQ
10-05-10	10.80 high	10.00	10.00	10.40	7.40       low	10.50	10.00	09.96	0.10 start      -4.41 end
07-07-10	10.00	12.00	12.33 high	11.50	08.80 low	10.50	11.50	10.95	0.10 start       -4.41 end
06-25-10	10.00	12.00	12.33	11.50	08.80 low	09.50	13.00 high	11.02	0.10 start      -4.41 end
01-10-10	10.00	10.29	15.53 high	11.55	11.10	09.50 low	10.00	11.14	-3.10 strt       -4.41 end
10-14-09	14.00	10.06	14.03 high	12.20	XXXXX	09.00 low	N/A	11.86	9.40 start  5.08 end
07-14-09	12.00	11.06	10.00	11.00	XXXXX	09.00 low	12.50 high	11.11	9.40 start  5.08 end
04-07-09	12.80 high	11.06	10.81	11.35	XXXXX	09.00 low	N/A	11.00	9.40 start  5.08 end
01-06-09	14.00 high	13.50	12.08 low	12.90	XXXXX	13.00	12.50	13.00	9.40 start  5.08 end

Observations/Conclusions:

(1) I omitted Take Stock from the averages because, compared to the seven mainstream estimates, it was radically different and outrageously low.   Consider:

** The latest seven estimates for COST ranged from a low of 7.50% to a high of 13.22% and averaged 12.17% whereas Take Stock was an untrustworthy -11.30% (that’s a negative or minus 11.30% for the next 5 years);

** For JNJ, the latest seven estimates were very close, ranging from 6.00% low to 7.00% high and averaging 6.48%, while Take Stock was – 2.74% (a spread of nearly 9% between the very lowest estimate and Take Stock);

** For QCOM, the average of the seven was 14.87% this October with 17.90% high and 6.00% low whereas Take Stock was a whopping -16.20%.

(2) I’ve noticed that Value Line is often at the far end (either high or low) of the many estimates I try to follow.

** For JNJ’s last three quarters, VL was the highest estimate; for QCOM, COST and WAG, it was consistently the lowest for all seven quarters; and for MSFT, its estimate was high three times and low four times.

** VL’s latest estimate for COST was so low (7.50% vs 12.17% average) that I was tempted to eliminate it from the average as an outlier.

(3) Morningstar, which also provides SSG data to StockCentral, had 2 high and 4 low estimates (3 for HPQ) out of 29 total.  Mstar does not make its own EPS estimates, but gets them from FactSet.

(4) Reuters had no high or low estimates for COST and WAG while VL was consistently low for both.  Out of 29 estimates total, Reuters was high only twice and never low.  

(5) Reuters is the only one of the seven estimates that provides additional details and useful context: it includes the number of analysts contributing to its consensus and the high as well as low estimates.   For example:

** Reuters’ latest estimate for MSFT was 11.76% with 11 analysts contributing and they ranged from 15.0% high to 8.0% low.  Eleven analysts provides some confidence that the consensus is firmly based and, with no further research, anything less than 6-8% seems too low by comparison.

** In 2009, for WAG, Reuters was estimating 13.00% rounded (12.89% actual) long-term EPS with 9 analysts contributing and they ranged from 16% high to 10% low.  Again, without further research, anything less than 8-10% seems unjustly low.  

(6) There are several reasons to check more than one analyst: we can learn which may be out-of-whack, which is the lowest, and we can compare the average to the lowest and also to the very lowest at Reuters to judge what constitutes a reasonably conservative estimate.

(7) If I had to rely on only one EPS estimate, it would be Reuters because it has more useful data than any other data provider and the fewest extremes

(8) And, as you might already know, the Research Links page at StockCentral offers long-term EPS estimates from five of the seven different analysts I always try to check (all but VL and S&P).  For info on StockCentral and other websites I use all the time, see my Favorite Links page.

(9) I’m interested in long-term EPS estimates only because each SSG requires an EPS estimate for the next 5 years.  My main concern for this research was to obtain the best EPS guidance I can find to help me complete my SSGs. For more on how I estimate long-term EPS, see: Estimating EPS.

 Armin

PS:  The table was a nightmare to prepare and proof so please let me know if you find any typos. Of course, any and all comments are greatly appreciated as are your ratings on the star scale below (just mouse-over and click).

 

Advertisement