CO o >-■ ■ ^ m W i P ^^ 'x* " The Service Bulletin of the Bureau of Personnel Research H. G. KENAGY, Editor VoL 4 MAY, 1922 No. 7 In this Issue : ARE LABOR TURNOVER RECORDS WORTH WHILE By David R. Craig The Division of Cooperative Research Carnegie Institute of Technology, Pittsburgh, Pa. "I ' ... • .. . 1 ;,. ;. , > ... ARE LABOR TURNOVER RECORDS WORTH WHILE A Criticism of the Present Methods of Recording and Interpreting Ttirnover Date - With Suggestions for a More Reliable and Valuable Plan By David R. Craig Assistant Professor, PersoHnel Administration, Carnegie Institute of Technology. The Bureau of Personnel Research May, 1922. » » FOREWORD Labor turnover is R popular subject in omployment manegement circles. Since the tine some years ago when a few interested individuals draw the attention of employees to the huge cost of high turnover among employers, employ- ment managers have co ncerned themselves very earnestly with the causes and remedies. In fact most such officials owe . their positions to the fact of labor tvirnover. Discussions, speeches, magazine articles, even books, have dealt with the problem in its many phases - from the social effects of rapid shifting to highly technical formulae for meoBuring the phenomenon* Rarely, very rarely, has the study of labor turn- over been looked at from the pragmatic standpoint of the total value of results obtained, professor Craig has established this fact very clearly in the discussion which follows. He points out the fallacies and insufficiencies of ordinary turnover data, r..rd at the same time the waste of effort and money involved in keeping records which have no real value. From an analysis of the actual turnover data in a large department store, he illustrates his criticisms of ordinary records and suggests valuable changes irtiich can be made. H. G. Kenagy, Assistant Director. TABLE OF COKTEKTS / tabor Tvirnover Records Todsiy «.«.* 1 What Do The Percentages Mean? ....4 3 What Do The Causes of Lator Turnover Mean .. 3 Comparing Labor Turnover Percentages 4 , When Is Labor Turnover High? 5 Reliability of the Causes of Lnbor Turnover. 6 ' 0n6 Company's Results 8 Conclusions < 12 Other Ways of Analyzing Labor Turnover ..... 12 Digitized by the Internet Archive in 2007 with funding from IVIicrosoft Corporation http://www.archiYe.org/details/arelaborturnoverOOcrairich ARE LABOR THHKO^ER RECORDS WORTH 7JHILB Labor Turnover Records To-^day. Five or six years ago labor ttirnover records began to gain the wide currency -which they now enjoy. It "waB important to notice, especially during the war situation, the costly mobility of labor, and to take every precaution that this shifting should be prevented as far as possible. It was delightful to bo able at last to express something about human beings in a mathematical formula . At first the statement of the turnover in simple per- centage figxzres was the standard method. After raising, the question of the method of computing this figure, the standard \vas adopted of dividing the total number of separations by the average working force; although there were mary who thought that the numerator of the fraction should be the number of replacements rather than the number of separations. The separations standard waa adopted by the U. S. Department of Labor. Labor turnover percentages are today commonly computed by this method. Hot long afterwards the desirability of analyzing labor turnover figures became apparent. Forms were drafted to simplify the recording of turnover data by causes, and the form now most widely prevalent is the one which v/as put out by the Library Bureau. It has since been modified by a large number of companies to suit their particular heeds. Almost every company that keeps labor turn- over records today is using some modification of this original form The causes of labor turnover are usually classified in some such manner as the following: -2- R© signed or left of own accord Better jot Salary or wages Dissatisfied with work Unable to get along with superior Leaving city Steadier employment 111 health Chagge of vocation Discharged Careless Incompetent Liquor Trouble breeder Misconduct Stealing or dishonesty Unreliable Insubordinate ?liyeiccl reasons (sometimes classified under "laid off") Laid off Decrease of fore© Lack of T/ork Unavoidable Injured Death T»i^nsferred The str.ndr.rd form for recording turnover data usually prints these reasons across the top of the card (horizontnl axis), and down the side (vertical axis) the department numbers oi" names, (for the monthly and yearly summarjr by departments), the names of the months (for the monthly recapitulations), or the days of the month, (for the daily recording)* The results obtained by the use of this form have apparently been satisfactory; at any rate, they have continued in use. It is possible to say, as a result of these records, that a certain number of men left department No. 30 because of dissatisfaction with their work, or because of finding a better job, or because of changing their vocations. This classifies the labor turnover and reduces it to simple elements. The causes can then b© attacked, and the tvirnover reduced. Of cotirse, this is the aim of keeping labor turnover records. -3- If they do not servo this purpoeo, then they are like any other use- less office enctonbrance^^ So much for the theory of this form. How about the practice? What Do The Percentages Mean? It is proper to raise some questions regarding the simple percentages. Is it always fair to say to Foreman Jones that his labor turnover is higher than that of Foreman Brown, and to hold him to account for that fact? Does this prove that Foreman Brown is a better foreman? May-it not possibly be explained that Foreman Jones lost less men than Foreman B»ovm, but since he had a much smaller department the denominator of his fraction brought his psrcentage up beyond Foreman Brown's. Riot . Dt> - The Causes of Labor Turnover i l xa? Al-so^^lt is proper to^ raise the question: Have labor turnover records, as currently kept, served their purpose faithfully? Has the employment manager been able to point out to the General Manager the \veak spots of the personnel policy? Has the information which the employment manager must have from these records been reliable? Have the resigning employees always stated the real reasons for resigning? Have foremen always given the real reasons to the employment manager when discharging an employee? These are serious questions to raise, because they strike at the only t^-vo aspects of labor turnover that are currently studied. If they can be ansv/ered satisfactorily, then labor turnover can proceed evenly along its 'my, if not, then we must consider some revision of our method of studying this industrial phenomenon. Let us take up first the original sspect of labor turnover. -4> and secondly the aspect which was added shortly afteirvmrds* Comparing Labor Turnover Percentages. The comparison of percentages bet'.^een departments and be- tween companies is fraught with dangers. The reduction of the fig\ires to a percentage basis lends them n superficial canparability. But unless the denominators of the fractions are nearly comparable, then the quotients are not strictly comparable. Suppose for instance that ■S7e have tvro depo.rtments, one of 100 men, one of 25 men. Such depart- ments exist in every large plant. Each department loses 10 men. The labor turnover is 10 per cent fbr the large department, and 40 per cent for the small department. Is the foreman of the small department to be charged ■-jd.th a ttirnover four times as large as the turnover in the large department? This is hardly fair. Furthermore, responsibility for labor turnover hinges on the causes that are given for it and the foreman of the 'small department may not be wholly responsible. Comparison of labor turnover bet'A'een companies is more accurate; with large denominators the quotients become comparable in greater degree. But comparison of labor turnover bet^A^een companies can serve no conceivably useful purpose unless it be to add to the prestige of one employment manager and detract from that of another. L'.bor turnover vdll never be reduced by comparing the figures for last year bete/een c ompa nie s , either in government reports or in chambers of commerce. To be sure, it is helpful to know whether the adoption of special methods by a particular company has been successful in reducing its labor turnca'er, and the comparisons betT.veen companies may possibly shcrr which company's methods have been successful. But in most cases the usual thought of the employment manager is "I was better or worse than so-and-so." -5- Lator Turnover Htgff? The comparison of labor turnover percentases, then, even ■when they are arrived at in similar ways, is not specially useful. What can be said of the percentages taken by themselves? The percentage figure must always be interpreted with reforenco to scroe standard. If the turnover is 75 per cent, is that high or low? Is it high or low with reference to what standard? The ideal labor turnover figure is not zero, in spite of the fact that many industrial managers have made a fetich in the past five years of the words "reducing labor turnover". No manager vdll say that ho would be pleased to see his labor turnover disappear entirely, for that xvould imply a stagnant condition of his business v/hich would be worse, possibly, than a high labor turnover. No standard has ever been set for labor turnover, though we may put down five or ten per cent per year as a tentative ideal, with reasonable aspurnace that not one company in a thousand will reach it. Yet this ideal refers only to times of prosperity. In the industrial depression from -wAiich we are just emerging, labor turnover, on the basis of separations, rose enormously. Men wet© being laid off every day, because there was no work for them to do. Only the best men were kept, and sometimes they too had to be released. Then came the period of starting up again, and in this process, v/ith so many good men out of work andapplying for work every day, many companies took advantage of the opportunity to re- place even the good men in their employ with better men from the outside, Trtienever they could be secured. Is it right to say that a low labor turnover, of five to ten per cent, is ideal in such a time? -6- If a comparer 0!?.n better itself by repl-^.cing poor men '-/ith good men, or replacing good men with better men, it is idle to ta^k about "reducing labor turnover". In such n time a high labor turnover is an excellent thing — for the company, if not immediately or remotely so for the men replaced. Viewing the labor turnover in times of prosperity, the same interpretation holds. I^ a company, by dint of careful search and selection, discovers a good man, that company is ill-advised if it thinks twice of labor turnover before seciiring him. Labor turnover percentages, then, mean nothing by them- selves. They are ^igh with reference to one standard, low with reference to another. The standard which happens to be in operation at any given time depejfids on general business conditions and the resulting con-"*." dition of the labor market. Labor tiirnover should be fairly high in times of depression, lov/ in times of prosperity, if the ideal is to be reached. This v/ill be true until depressions and prosperous periods are brought within control. Since there is no such thing as "normal times" (normal times, you will notice, are always "next year", or "last year"), there is no norm that can be a fixed reference for labor turnover. Reliability of the Pauses of Labor Tnrjiover . What of the causes of labor turnover? There are two kinds of reasons for human actions. One is the kind which the individual gives as his reason for behaving as ho does. Sometimes this is also the real reason. But it is easy to demonstrate that our motives are not slways known to us. For example, the business man says that the Soldier Bonus should be defeated because it is wrong to put a price -7- on patriotism; the ex-eoldier points out that the country ought to Bhavf its gratitude to its heroes. These are "good" reasons. The other kind of reasons is the "real" reason. The business man will have to pay a tax if the bonus goes through; the soldier will lose money if it doesn't. How often will "good" reasons and "renl" reasons coincide in labor turnover records? Quite often, perhaps; very seldom, perhaps. It is not possible to rely on all of them. ■When an employee resigns from a company, he may give any one of the reasons that are listed, and possibly some others. If he is an ordinary employee, he is likely to remember that employment is nowhere very secure, that he may need to come back to his compary at a future time. He will therefore want to leave open every avenue of return. Now, if he remarks as he quits that he is resigning because he is dissatisfied with his wages, or dissatisfied with his work, or because he is xanable to get a long with his superior, these facts are likely to be remembered and held against him the next time he appears at the gate looking for work. It is very easy for him to hide these "real" reasons and say that he "is leaving the city", "has a better job", "is in ill-health", or give any of the other "alibi" reasons that are listed or that occur to him epontaneously. When a foreman wishes to discharge an employee, he may givB any number of reasons in the same manner. A great many of the reasons are verifiable from the records. "Attendance" is one of them. "Dishonesty" is another. The writer recalls a case in which the real reason was not given, and a substitute reason supplied. In a company employing 4500 men, the labor turnover of one department was more than 1000 per cent. It was found, on analyzing the records, that a good -8- percentage of the men had been discharged for incompetence, yet the employment manager had been very careful to supply the foreman with tradesman of good ability and considerable experience; some of them had had twenty years experience at their trade, which vras longer than the foreman's own relationship to it. It was found, on questioning the men more closely, that the foreman was trying to get non-vmion men in a trade which was completely unionized. It was an impossibility. In order to avoid the appearance of discrimination, the foreman had set down incompetence as the reason for discharging able men who held union cards. The employment manager can hardly be ^tiided by this type of cooperation. "Insubordination" is not always a choice "alibi" reason, for the reason that there are always two sides to insubordination, and if a foreman is constantly releasing men because of this reason, he is likely to be criticized keenly on his methods of supervision. If this reason is given, then, it is likely to be the right one. One Company's results. Let us examine the records of a particular compary for one year, and analyze the causes of separations. The table belov/ is taken from the 1921 records of a depsirtment store, and supplies the desired data : (See next page) ■■" ■ ••. • 1 . > (J , : ^^.y-i. - ■ ^ ' , ^ .'." i \ , -9- TABLE NO. I Reasons for Number Per cent Per cent Per cent Leaving left Resigned Dismissed total Married 24 5.7 2.1 Moved away 56 13.0 8.6 Return to School 22 5.2 2.0 Hea 1th 44 10.3 4.0 Diiable to get along with s\;5)erior 5 1.1 .5 Dissatisfied with v7ork 14 2.3 1.3 Other position 110 26.8 9.8 Salary 2 5 ' .2 Ho notice 81 19.2 7.1 Stay home 68 15.9 6.1 [Total resigned) (426) 100 .OJS (38.7) Attendance 21 3, .1 1.9 Incompetence 84 12 .2 7.4 Dishonesty 8 1 .2 .7 Insubor d in?, t ion 14 2. .0 1.3 Wrong type 10 1 .5 .9 Retired on pension .0 0.0 Reducing force 248 36 .0 22.1 Temporary 302 44 .0 27.0 (Total dismissed) (687) • • lOQff. (61.3) Total loss 1113 100. 09? In this table, as in the classification of causes of turn- over given on page 2, two general divisions are shown. These coxirespond to "Reasons given by employees for resigning or quitting" and "Reasons given by foreman or department heads for dismissing or firing employees". These divisions must again be divided into reasons which are reliable and reasons v;hich are not reliable. The resigning employee has every reason not to say that dissatisfaction vdth wages, dissatisfaction with work, and inability to get along with superior are determining causes for his resignation. If, therefore, he gives any one of these reasons, it may be presumed to be true. He may give aur of the other -10- reasons instead, if he likos. Of these other reasons, which we shall refer to ns "nlibi" reasons, sojno may t>e correctly stated and may be found to be justified in any ?iven case. But it is important to note that the three reasons v/hich usually operate to aause resignations art- the very ones which an employee hesitates to give. These we have every reason to beliovo, then, if ho gives them. We have some cause to doubt the others, and less reason to believe in their reliability. They may be classified as follor/s: TABLE NO. 2. Reasons which are reliable Married (verifiable) Unable to get rilong tdth superior Dissatisfied with work Salary Reasons which are not reliable Moved away Return to school Health Other position No notice Stay at home In the same manner, scjtb of the reasons vrhich foremen give are sometimes verifiable and some ere not. The reliable and the "alibi" reasons may be classified as follows: TABLE no. 3. Reasons vrhich are reliable Attendance (verifiable) Dishonesty 7/rong type Retired on pension Insubordinr>. t ion Reasons which are not reliable Incompetence Reduc ing ' force Temporary -li- on the basis of this discussion, let us reclassify the first Bianmary of cousos shown in Table No. 1 above. In order to make clearer the point at issue, the two causee called "reducing force" and "temporary" have been placed in a separate class in this table, under the heading "partly reliable". The business of a department store is inherently so seasonal that we cannot say these reasons are absolutely unreliable. Btit they make such good "alibi" reasons that they cannot be called thoroughly reliable. TABLE ITO. 4,_ Per cent of total Z.lfo .5 1.3 .2 1.9 .7 1,3 .9 0.0 5.6 2.0 4.0 9.8 7.1 6,1 7.4 42.0f„ Reasons for leaving Number left (RELLA.BIS) Married 24 Unable to get along with superior 5 Dissatisfied with work 14 Dissatisfied ivith salary 2 Attendance 21 Dishonesty 8 Insubor dina t ion 14 Wrong type 10 Retired on pension Total (reliable) S8' (OTIRELIABLE) Moved away B6 Retvirn to sbhool 22 Health 44 Other position 110 No notice 81 Stay at home 68 Incompetence 84 Total (iinreliable) 46S' (PARTLY RELIABIE) Reducing force Temporary employmenrb Total (partly reliable) Tota Is 248 302 550 1113 22.1 27.0 49.1?5 100.0^ Notice especially those figures which isrhow the total number of separations caused by the reliable reasons, and those caused by the I<'eu/^bility /^naly3i5 of Reasons f^of^ lu/^NOvEf^. 2 <» e 6 IP It /^ [t> la 20 ^^ 24 Zi 26 so 32 34 3fc 30 ^/? 4e 44 4 6 40 50 "T — r I^tUP^QLS. Mar ^ied. , Xnt > Uifa) o rL:y/A|<7 ^ vb/ "> I 1 O \v/ T^tal. rh l5o 'arL. S^iptficir. I I ^iyN/^tL-lAlSLE. f.)os t'Hon T<;>-r/»4- FV^fSTlLY ^EiJ/^ BI_E 'Tentpartlru ernyo/oyrrjcrn*; »n Force rjT-^c -12- unreliable ones. 8.9 per cent of the separations are caused by reasons of which we are ipfe certain* To be stare, a large proportion of the labor turnover in this particular case has been caused by lay-offs and the employment of temporary help. We do not kno^v how large that proportion really is, and how much the stated figvire must be reduced for unreliability, possibly 90 per cent is caused by the release of temporary employees and layoffs. Conclusions Our labor turnover forms, then, do not record the causes in a trustv/orthy mr.nner. If the percentages of labor turnover are not to be compared, '?nd are subject to differing interpretations at different seasons, and if the causes given on ovir forms and by o\ar employees are not as reliable a s we should like, is it not proper to ask ourselves, "Is it worth while to keep labor turnover records?" Other Ways of Analyzing Labor Turnover. It is v/ithout ' doubt important to record labor tvirnover; with this view we have no quarrel. The thesis of this paper is that the labor ttirnover records, as ordinarily kept, cons\ane a great deal more time and clerical labor tlian the results warrant. There are other ways in which the labor turnover record may be kept which will save much of this time and effort, and yield fully as important results. The fundamental record includes only those facts regarding labor turnover which are indispensable. These are: 1. The average working force in each department. 2. The nTjmber "'ho leave each department (if the turnover is kept on the basis of separations); or the number who are actually replaced -within a department (if the turnover is kept on the basis of replacments) . 8. The percentage figure vrhich results from dividing (2) by (l). >" w -13- ThiB will give the fundamentals. In matty cases the departmen'tiB vrill have a reasonoble labor tulrnover, and it will not b© necessary to consider those departments further. If the ttornover in other departments is considered to be too high or too lo7/, depending on the position of the business ciarve at the time, this can then be investigated by other means. The feature of this plan is that a great deal of unnecessary work is avoided, and only necessary v/ork is done. In order to analyze further the labor turnover in those departments ivhere it is unreasonably high or lev/, as the case mfty be, two or three methods may be adopted. They are as follows: 1. An informal inquiry could be made of the department manager or foreman, asking him to explain ar^ deviation from desirable turnover. This method would be satisfactory only if the foreman can be trusted to cooperate successfully 'Adth the employment manager, and is not personally responsible for the tiirnovor because of some trait of character or temperament which makes him an undesirable boss to those under him. 2. The second method tal C 1 N 1 1 C g c 1 1 r > 1 M M f 0- N 1 o 10 O m 2 N ffU O 1 1 % 1 in 1 m t- C «i L 8 > 1 ■+• c «l 1 c •» -♦- 1 • 2 M m ^ X > 1 c • it 3) I 9i C L (I 1 Oiasa+iafied wi+h pay • " ••" ^: •• •• •"• ... •• • »... \V ... • :•:• :"• •••• •.;•• ... :^" • •*■ •' • :::•. *• .... *■•• 19 Olaaatiofiad withFctreman ::;' ••■ • • ' • ••» ;... •• * ' •• .... ... :::• .... «... •• * ••* • ■• • • •*** ■ ... *•■ 12 Beffarjob * "•' i*" •••* • ■ »• • •** .... *•* " • •> I' I! .... • "IZ :::: •*** ..*• *. . .... • •** ■M# ••• Un — •**• la Lea\'ir-)ff C//y * M '• • • •*• -•■■ »• • *" »•• •• •■• :;:• '■ •• *• • »• • •• V" " ■•• u- /o Reason no-f known :::| >«•• •••• •• ::•• .... • ... -• !«• :u* J'" 5?" «• ••*• .— • • :r.: »i»« •••• "*' ;- • .... 22 Jnc o n->;0 e /wn ■/• •• — • *• .... — •• ■•• V** :::: ~ • • •«•« ... •••• •-• • L"' |o Insubondina-^iort n** ••• • „ • ■•• i" • • • • •• • •M •• :u* ... ... .... - £r* ..• • r.*" •• • • le / O - / VV«e>< (VVee/c- 4 Weeks tMoni-h-3f^on-l-ha 4*ro3. -6/Vfo9, 7Af«»s. -l2Mo«. /(/ear- -2t/eai-3 3yeara -5 Ueoi-a r:it *■■• :::: H:: :~" III* Ji? • ••• «•• isi iV,' W rst "At •• nu ::« p; iU* •IS* 3«» M.. • ••* • "• *••! •••• •• P •*• •••• 15** ;ir • • •• ::"' l«» Lengfh ' • ••• •• • ••■ M* •••• •• .... "i- :u- • •••• •• It** •• 17 of < ■•«• r::: w* x; • ••« •••• ••"- — *; "" ••• >••• )««■ •••• • •••■ •• r.:- ••■• 20 d«rv/ce '" ' • « • ••• • ■• * »« *• • 3 • • « • « / N • « • • • * / <"« - 'o • - "-• ;••• • •:•• iU: • •••< «« • • • • • • 13 l• 1*^ v." ••• n • 27 - 30 31 - 35 36 -40 *•• • • " ■• :;•• •• ■ •.« ••• r" •-** 16 ■ •• ' •*** **■ * • ••« •• •• 4 k • « • • • / <* b-^ 2o ^.: » • It; • .... • •«•• • • 15 2/ - ZS 26 - 30 3/ - 35 36 - 40 1^ 41-45 1^ ••• jS is • •* ...• •*•• •fc^ •M U~ •• 31 Poif < • ••• ill? « 111! .«.« •«*• ».»* • >•• •^7 • !••• • •• •" 39 • • >• ::;• :•■■•■ .... •7* •— •••• .^ 16 **" •" •• •••• • * • • 4 o «••* • ■£' *• P r* 40 , / i./V» vv <^*. Parf-nH mTc ^- .■If • • il? 3J 1 Boai-d Home Condirions < • Rcni Home •«* rs^ :H1 '**• •• S« 43 »»M **'• • i»T »- « r~ 14 V Own Horn* • • •• 2 < HoO»p*ndenf, [ST M«« >••• ::•• ••t « / • ••• 30 t »•• »••* r*' ■• le 3 :•." • • 8 ^ • 1 7b yw/, Sourc m C e/wm /-» 35 12 do ?i (ffe/Of-oefue 1^ _^ L£ »r-r «/'• aio "f roi «" ro< ^« fnt «// Vha no ?•' -ne r,y" iai lw< r e f ' ^i 0' 7