A Study of Financial Analyst: Practice and Theory

The exams, curriculum materials, and seminars designed for the CFA (Charactered Financial Analyst). Program are based on knowing what is im ortant to racticing financial analysts. !et, little documentation exists about what financial analysts actually belie"e in and do. The intent of this research was not necessarily to identify the normati"e a roaches but, rather, to identify the most widely used a roaches. #oreo"er, the result are not intended to suggest that future analysts be directed to the most commonly used a roaches. The intention of this article is to share knowledge about what goes on in the day$to$day ractice of financial analysts. For exam le, use of resent "alue analysis is hea"ily stressed in the CFA curriculum and is a ma%or focus of textbooks on in"estments, but how widely is resent "alue analysis actually used and by whom& Also, new techni'ues for analysis, such as economic "alue added, ha"e recei"ed relati"ely less attention than traditional measures of analysis, but little is known about how widely acce ted ()A is by ractitioners. This sur"ey addressed such issues.

The study
The artici ants in this study came from the membershi of A*#+ (the Association for *n"estment #anagement and +esearch). ,uestionnaires were mailed to a random sam le of -.. A*#+ members in the /nited 0tates in 1ctober 2--3. 4ecause of address changes and other factors, 33. mailings successfully arri"ed at their intended destinations. 1f that number, 5-6 usable res onses were recei"ed, for return ratio of 77.68 ercent. A follow$ u tele hone sur"ey of randomly selected non res ondents indicated no statistically significant differences between those who initially answered the 'uestionnaire and those who did not. The final 'uestionnaire, which is re roduced in A endix A, had been re"iously tested in three ilot grou sur"eys.

The 'uestionnaire materials made clear to

artici ant that the sur"ey was

s onsored by the author and not by any business organi9ation or A*#+ itself.

The Respondent Group

The first three tables in this article re"eal key characteristics of those who res onded to the 'uestionnaire. *n Table 2, the 5-6 res ondents are delineated by the ty e of firm for which they worked. The largest number of res onding financial analysts were em loyed by brokerage firms and ri"ate money management grou s. *n"estment management counseling firms, mutual fund, and bank trust de artments are also re resented substantially. Althought no attem t was made in this study to stratify the sam le by industy classificationin ad"ance, the com osition of res ondents does reasonably re resent the membershi rofile by industry classification as re orted by the more than 75,... A*#+ members in the 2--3 #embershi :irectory. As indicated in Table 5, ;6.6 ercent of the res ondents were CFA charterholders and 87.ercent held #.4.A. degrees. The charterholder number in this sam le is slightly smallet than for the total organi9ation (6. ercent), whereas the #.4.A. degree number is slighty larger than for the total membershi (<6 ercent). =ote that the a"erage ex erience of the res ondents is 28.7 years. Table 7 re orts the undergraduate ma%or of the res ondents. A large ercentage of the res ondents (and erha s, inferentially, a large ercentage of A*#+ members, although no industry data are a"ailable with which to com are these data) had undergraduate degrees in business and economics. The notion that the ty ical route to becoming a financial analyst is for an indi"idual to get a liberal arts degree and then use that broad$based background to concentrate on financial analysis is not su these data. orted by

The Result
This section contains discussion of the sur"ey finding regarding the "ariables (or in uts to "aluation) and tools financial analysts use in e'uity "aluation, their attitudes toward issues im ortant in ortofilio management , and their attitudes toward market efficiency "ersus market anomalies.

Valuation Input. +es ondents were asked about their use of se"eral "ariable
and tools in analy9ing securities. Among the most im ortant was resent "alue (P)) analysis> others included cor orate earnings and cash flow. Present value. The use of P) analysis is a central theme in "aluation theory. There is robably not a CFA exam re aration course being taught around the world or an in"estments course being offered at a uni"ersity that does not include P) analysis techni'ues. 4ut as Panel A of Table < indicates, only 28.5 ercent of res ondents always use P) analysis and for <8.6 A ercent, it is not art of their normal rocedures. arently, ractitioners s lit about 8.?8. in their use of P) techni'ues. 0hould thisfinding be taken as an indictment of the rofession& @ardly. Ahen faced with the reality of "aluation in the market lace, the task of ro%ecting earnings, di"idends, and a stock rice into the future and determining a ro riate discount rately on discounted cash flow (:CF) analysis in the determination of "alue. As noted financial economist 0teward #yers (2-3<) of the #assachusetts *nstitute of Technology has suggested, B:CF is sensible, and widely used, for "aluing relati"ely safe stocks aying regular di"idends, but :CF is not as hel ful in "aluing com anies with significant growth o ortunities.C ( . 25;$276) =e"ertheless, because P) analysis is art of the foundation of finance, * decided to analy9e its use by "arious categories of artici ants. 0hown in Panels 4 and C of Table < are the use and nonuse of P) analysis by CFA charterholders (hereafter, sim ly BcharterholdersC) "ersus noncharterholders and #.4.A.s. Although the charterholder grou indicated a slightly larger tendency to use P) analysis than the noncharterholder grou , the difference is not statistically significant at anyreasonable le"el of significance on the basis of a chi$s'uare inde endence of classification test (re orted in A endix 4). The same conclusion a lies in regard to the use of P) analysis by #.4.A.s "ersus non$ #.4.A.s. *f anything, non$#.4.A.s. a ear to be slightly higher users of P) analysis. Table 8 shows the breakdown of the use of P) analysis by res ondents industry classifications. *n this case, the chi$s'uare test (see A endix 4) indicated a statistically significant difference between the categories. A null hy othesis of no relationshi between industry classification and the use of P) analysis could be the re%ected at the 8 ercent le"el of significance. *n this sam le, indi"iduals em loyed by mutual funds and

bank trust de artments a ear to be relati"ely high users of P) analysis whereas those working for brokerage firms, ri"ate money management grou s, and in"estment banking firms do not. 1ther in uts. The res ondents were also asked to determine the relati"e im ortance of other in uts in analy9ing securities. Table ; shows how the sur"ey artici ants ranked the im ortance of earning, cash flow, book "alue, and di"idends. The a"erage ranking for the in ut is shown in the far right column. (arning and cash flow are considered far more im ortant than book "alue and di"idends. The lack of im ortance these res ondents assigned to di"idends is interesting. As re orted in Table ;, only 7 of the 5-6 res ondents considered di"idends to be the most im ortant "ariable in "aluing a security. 1ne hy othesis is that such conclusion by analysts are linked to the irrele"ance of di"idends theory initially ostulated by #odigliani and #iller (2-;2)Dand debated e"er since. 4ut a far more likely cause of the low di"idends ranking is that in the momentum$dri"en en"ironment of 5.$7. ercent annual return of the mid$to$late$2--.s, di"idends do not count for much in the minds of analyst. Furthermore, the shar ly lower ca ital gains rates s ecified in the Tax ayer +elief Act of 2--6 all but wi ed out the e'uali9ation of taxing in"estment di"idends and ca ital gains that was and essential element of the +eagan Tax +eform Act of 2-3;. Finally, the desire by cor orations to buy back shares rather than increase cash di"idends a ears to be a distincti"e feature of the 2--.s. =ot all would agree with the lack of im ortance of di"idends. 4ernstein (2--3) made a strong case that management creates additional rein"estment and earning risk for shareholders when the com any retains a rogressi"ely larger ercentage of earnings. The unim ortance of di"idends to this sam le of analysts is further reflected, howe"er, in Table 6, in which the res ondents ranked the most significant in uts indetermining a stockEs P?(. 1nly 7 of the 5-6 res ondents ranked di"idend olicy first among the fi"e in uts listed>56; ranked it last. Although analysts might change the rankings shown in Table 6 when "aluing a real estate in"estment trust or a com any in the later stages of its life cycle, the classification of di"idends as unim ortant is clear in Table ; and 6. Also in Table 6, the growth otential for the com any has a strong F2 ranking as a determinant of a stockEs multi lier. The F5 ranking of 'uality of earning (abo"e 'uality of

management, risks and di"idend olicy) a ears to reaffirm the strong concern that racticing analysts ha"e for the legitimacy of re orted earning. *n another 'uestion related to "aluation, * asked the res ondents to rank the im ortance of the three in uts shown in Table 3 as art of the determination of whether a stock should be bought, sold, or held. The long$term outlook for the com any and the current "alue of the stock "ersus its historical trading range recei"ed to rankings> next 'uarterEs (P0 number was last by a larger margin. This res onse is somewhat sur rising> a click on the *nternet will bring a deluge of under$and o"er erfomance of 'uarterly earning against ex ected earning. Perha s the 28.7 years a"erage ex erience of the res ondents allows them to o"ercome the hy e of the moment.

Valuation Models. *n addition to 'uestiona about the in uts to stock

e"aluation, the 'uestionnaire asked res ondents about their use of three "aluation models. Panels A and 4 of Table - ro"ide the results for two traditional modelsDthe di"idend "aluation (di"idend discount) model and the ca ital asset ricing model (CAP#). =either model fared well in the sur"ey. The di"idend model was "iewed as "ery im ortant or moderately im ortant by <5 ercent of the res ondents, and the same two o inions totaled 72.2 ercent for the CAP#. The model that recei"ed the highest number of "ery or moderalety im ortant o inion, as indicated in Panel C of Table -, is the economic "alue added (()A) model de"elo ed by 0tern 0tewart and Com any. 0trictly s eaking, ()A is not a "aluation model, but it does ha"e im lications for describing stock rice beha"iour. 4ased on these sur"ey results, ()A may take on increasing im ortance for analysts. Ahether the res ondents understood that ()A is rimarily a method for s litting earning between re'uired returns and excess returns is not e"ident from the 'uestionnaire. Further in'uiry about how analysts use ()A would thus be useful.

Portfolio Management
The issues discussed so far ha"e dealt with "aluing indi"idual securities. The three items tabulated in Table 2.Dbeliefs about market timing, the a eal of global in"esting, and near$term re"ersion to the meanDrelate more to ortofolio management.

Panel A of Table 2. indicates that only 53.; ercent of the res ondents belie"ed that attem ts at market timing are likely to enhance ortfolio returns (the "alue is 75.6 ercent if only those with o inions are included). The consistency of this res onse with the result s shown in Panel C will be discussed shortly. Panel 4 of Table 2. deals with global in"esting. A ma%or henomenon ortfolio managers ha"e witnessed in the mid$to$late 2--.s is the s eed at which international financial markets react to each other. #arket erformance in the /nited 0tates on a gi"en day a ears to start a chain raction in Gondon, Tokyo, and other ma%or markets. The se'uence may also mo"e in the other direction. The internationali9ation of the world economy through reduced trading barriers and the increased merger acti"ity between financial institutions in "arious countries a ears to add to this chain reaction. The res onses to ,uestion 2< re orted in Panel 4 gi"e strong su ort to the notion that global in"esting may ha"e lost some of its a eal in the closely linked markets as a means to achie"e better risk$return outcomes through di"ersification. 0lightly more that 36 ercent of res ondents belie"ed there has been some loss or substantial loss of a eal. Finally, Panel C of Table 2. addresses a 'uestion that all ortfolio managers and analysts a ear to be asking in the financial ressDwhether there will be a re"ersion to the mean for P?(s and di"idend yield within the next decade. Aith the P?( for the 0HP 8.. *ndex in the 5<$53 range and di"idend yield in the 2.;$2.3 ercent range in late 2--3, this 'uestion is timely and great interest to the rofession and in"estors. Among the res ondents, as indicated in Panel C, 86.; ercent ex ected a re"ersion to the mean. This statistic suggest that many belie"e e'uity "alues will be lower in the future, but res onses to ,uestion 6 (not re orted here) indicate that res ondents belie"e high "alues may be sustainable as long as interest rates and inflation remain low. The re"ersion is erha s most likely to come when these mitigating "ariables are not longer in lace. The totality of information in Table 2. may re"eal an inconsistency on the art of res ondents. The ma%ority did not belie"e in market timing but did belie"e in a coming re"ersion to the mean Presumably, a re"ersion to the mean has im lications for the timing of decisions.

Mar et !fficiency
The res ondents were asked to indicate their acce tance or re%ection of the efficient market hy othesi ((#@), which in its broadest (semistrong) for suggests that ublic information is im ounded it the current rice of the stock and that any additional analysis by an indi"idual analyst is likely the roduce little or nothing in the way of added "alue. The (#@ was initially ostulated in the 2-;.s, and it has been under se"erate attack e"er since as researchers claimed to identify anomalies almost e"ery area of in"estments. As shown in Table 22, close to 2.. ercent of racticing analysts in this sur"ey were neutral or strongly disagreed with the (#@. The res onses to an allied to ic are resented in Table 25. in answering a 'uestion about the most im ortant "ariable in determining ortfolio returns, more than ;. ercent of the res ondents chose the skill and training of the ortfolio manager as most im ortant. :es ite the em hasis on the risk com onent often found in the academic literature, risk in the ortfolio came in at about half the ercentage of skill and training. And the amount of trading in the ortfolio came in a oor third. These res onses are generally in line with the re%ection of the (#@ re orted in Table 22 but at "ariance with the res onses to the usefulness of the CAP# shown in Table -. A number of res ondents who indicated that skill and training was the most im ortant "ariable in determining ortfolio return suggested that ego might ha"e layed a role in their o inion. 0uch a suggestion would be consistent with the em irical research in this area in the ast decades (Fama 2--2> Iandel and 0tambaugh 2--;). Perha s ho e trium hed o"er realityfor the ma%ority of res ondents. To in'uire into analysts attitudes toward anomalies that tend to dis ro"e the (#@, the res ondents were gi"en four market strategies from which to choose (,uestion 25). These four were by no means inclusi"e of all the ossible strategies, and in s ite of research in this area, no one answer can be assumed to be correct. The answers are resented an Table 27. Table 27 shows that the low$P?( effect and the small$firm effect recei"ed the greatest allegiance. This res onse to the small$firm effect is of articular interest because the small$firm effect has been called too time$ eriod s ecific and o"erly de endent on the month of January for high returns. As an exam le of the time$ eriod s ecificity, research

has found that between 2-68 and 2-37, small$ca itali9ation stocks a"eraged a 78.7 ercent annual return, more than twice the 28.6 ercent return of large$ca stocks. :uring the same time eriod, com ounded total returns on small$ca stocks exceeded 2,<.. ercent. @owe"er, from 2-3< to 2--6, small$ca stocks (as defined by *bbotson and Associates 2--3) increased by 85;.- ercent while large$ca stocks (0HP 8..) were u -.5.3 ercent. Ahen one stri s the 2-68$37 eriod out of the *bbotson and Associates data, small$ca stocks fell one$third below large$ca stocks from 2-5; through 2--6. The intent here is not to castigate small$ca stocks> clearly, such stocks as #icrosoft, *ntel, and @ome :e ot had to start as small$ca stocks. Furthermore, for the articularly astute analyst, smaller com anies may re resent es ecially good areas for study, in that e"en the strongest ad"ocates of the (#@ would admit that small com anies ro"ide o ortunities. The im ortant oint is that the strong su ort for the small$firm (and low$P?() anomaly in this study may indicate that many racticing financial analysts maintain a belief in these conce ts and a belief that a different market en"ironment may bring the o ortunity for strong small$ca erformance to rea ear. Also, the loyalty that some in"estors ha"e shown to large$ca high$P?( stocks (such as Coca Cola and Keneral (lectric) is not necessarily felt by res ondents in this study, who a ear to be more "alue$ stock than growth$stock oriented.

"onclusions
The most im ortant conclusion from this sur"ey is that P) techni'ues are not as widely used in ractice as they are in theory. 1nly 8<.7 ercent of the res ondents said they use P) analysis as art of their normal analytical rocess. The cause may be that the difficulties of ro%ecting future cash flows and selecting an a ro riate discount rate sim ly make use of P) analysis a ear to be too difficult for real$life decisions. Although the length of forecasting eriods was not s ecifically co"ered in the 'uestionnaire, my obser"ation is that few analysts ro%ect earnings or di"idends more than two (or at most three) years into the future because of uncertainty. Also, they rarely ro%ect future P?(s. The industry ractice is to di"ide the current rice by future earnings to create a multi le of future earnings to create a multi le of future earnings. This

a roach is, of course, "ery different from ro%ecting a future P?( that can be used to discount a future stock rice back to the resent. Answers to a number of 'uestions indicate that the di"idend$ aying olicy of a com any is relati"ely unim ortant in the analytical rocess. This attitude may be related to the current en"ironment. *n addition, although 'uarterly earnings announcements ha"e recei"ed much attention in the financial ress, 5-5 of the 5-6 analysts said 'uarterly earnings carry less weight than the long$term outlook for the com any or its current "ersus historical trading range. The res ondents ga"e high marks for im ortance to the ()A a roach to "aluation and low marks to the di"idend "aluation model and CAP#. The res ondents adhere to the notion that the most im ortant "ariable in determining return on a ortfolio is the skill and training of the ortfolio manager and that this consideration o"erweight theories about stock market efficiency. Finally, res ondents belie"e that global in"esting has lost some a eal as a risk$retun o timi9er in a world that a ears to be increasingly integrated.