N obody Ever Gets Credit for Fixing Problems that N ever H appened: CREATING AND SUSTAINING PROCESS IMPROVEMENT N e lso n P Re p e n n in g Jo h n D Ste rm an H ow m u ch wou ld you r organ ization pay to develop m an u factu rin g capability equ al to Toyota’s? How m u ch wou ld a world-class, sixsigm a qu ality program be worth to you r com pan y? How abou t Harley-Davidson ’s ability to tap in to th e h earts an d m in ds of its cu stom ers or Dell’s ability to m an age its su pply ch ain ? Most firm s are workin g aggressively to develop th ese an d sim ilar capabilities th rou gh process im provem en t Th e com bin ed expen ditu re of U.S com pan ies on m an agem en t su ltan ts an d train in g in 1997 was over $100 billion , an d a sizeable fraction wen t towards efforts to develop operation al capabilities m atch in g th ose of th e best firm s in bu sin ess Wh eth er it’s an advan ced m an u factu rin g system or th e ability to respon d qu ickly to ch an gin g cu stom er n eeds, th e drive toward im provem en t h as becom e a way of life in corporation s today Th ere is on ly on e problem Despite th ese vast expen ditu res, an d n otwith stan din g dram atic su ccesses in a few com pan ies, few efforts to im plem en t su ch program s actu ally produ ce sign ifican t resu lts Con sider, for exam ple, Total Qu ality Man agem en t (TQM) In th e 1980s, spu rred by th e su ccess of m an y Japan ese firm s, TQM was all th e rage am on g U.S firm s Con su ltan ts an d bu sin ess sch ool facu lty preach ed its virtu es an d m an agers m ade pilgrim ages to com pan ies with award-win n in g qu ality program s By th e m id-1990s, h owever, TQM was sidered passé Academ ics h ad m oved on to oth er issu es, TQM received rare m en tion in th e popu lar bu sin ess press, an d articles th at did m en tion it u su ally did so in a n egative text TQM Work reported here was s upported by the MIT Center for Innovation in Product Development under NSF Cooperative Agreement Number EEC-9529140.For more information on the res earch program that generated this article,vis 64 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened h ad all th e earm arks of a m an agem en t fad: An in itial bu rst of en th u siasm , a flu rry of activity, an d th en a steady declin e as it was replaced by n ewer in n ovation s su ch as re-en gin eerin g It wou ld be easy to clu de th at TQM’s u n derlyin g valu e was m in im al However, wh en on e looks at th e experien ce a little m ore carefu lly, a differen t pictu re em erges A n u m ber of carefu l stu dies h ave n ow dem on strated th at com pan ies m akin g a seriou s com m itm en t to th e disciplin es an d m eth ods associated with TQM ou tperform th eir com petitors.1 Th ere is n ow little dou bt th at wh en u sed properly, TQM produ ces sign ifican t valu e to both organ ization s an d th eir cu stom ers Yet paradoxically, it rem ain s little u sed A recen t stu dy fou n d th at fewer th an 10% of th e Fortune 1000 h ad well-developed TQM program s; an d, in an oth er stu dy, TQM fell from th e th ird m ost com m on ly u sed bu sin ess tool in 1993 to 14 th in 1999.2 Th e situ ation is sim ilar for a wide ran ge of oth er adm in istrative an d tech n ological in n ovation s.3 Tech n iqu es tou ted as today’s “core com peten cies” all too often becom e tom orrow’s failed program s On ce an effort h as failed, th ere is an alm ost irresistible tem ptation to label it a fad or “flavor of th e m on th ” However, diggin g a little deeper sh ows th at m an y su ch tech n iqu es h ave u sefu l ten t It sh ou ld com e as little su rprise th en th at m an y cu rren tly popu lar in n ovation s are little m ore th an old ideas with n ew acron ym s Th e core disciplin es associated with statistical process trol an d varian ce redu ction becom e six-sigm a; wh at was on ce called a qu ality circle is n ow a h igh -perform an ce work team Th u s, today’s m an agers face a paradox On th e on e h an d, th e n u m ber of tools an d tech n iqu es available to im prove perform an ce is growin g rapidly Fu rth er, with advan ces in in form ation tech n ology an d th e ever-growin g legion s of m an agem en t su ltan ts, it is easier th an ever to learn abou t th ese tech n iqu es an d to learn wh o else is u sin g th em On th e oth er h an d, th ere h as been little im provem en t in th e ability of organ ization s to in corporate th ese in n ovation s in th eir everyday activities Th e ability to identify an d learn about n ew im provem en t m eth ods n o lon ger presen ts a sign ifican t barrier to m ost m an agers In stead, su ccessfu lly implementing th ese in n ovation s presen ts th e biggest ch allen ge Pu t m ore sim ply, you can ’t buy a tu rn key six-sigm a qu ality program It m u st be developed from with in To learn h ow firm s can overcom e th is “im provem en t paradox,” we h ave, over th e past decade, stu died process im provem en t an d learn in g program s, focu sin g on th e dyn am ics of im plem en tation an d organ ization al ch an ge We du cted over a dozen in -depth case stu dies in in du stries in clu din g telecom m u n ication s, sem icon du ctors, ch em icals, oil, au tom obiles, an d recreation al produ cts.4 We gath ered data th rou gh observation s, exten sive in terviews with participan ts, arch ival records, an d qu an titative m etrics We com plem en ted ou r field research with th e developm en t of a series of m odels captu rin g th e dyn am ics of im plem en tation an d im provem en t.5 Usin g system dyn am ics as th e basis for u n derstan din g im plem en tation h as yielded a n u m ber of in sigh ts in to th e im provem en t paradox In at least som e cases, th ese in sigh ts h ave proven CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 65 Nobody Ever Gets Credit for Fixing Problems that Never Happened in stru m en tal in h elpin g firm s ben efit from th e poten tial provided by available im provem en t tools an d tech n iqu es Most im portan tly, ou r research su ggests th at th e in ability of m ost organ ization s to reap th e fu ll ben efit of th ese in n ovation s h as little to with th e specific im provem en t tool th ey select In stead, th e problem h as its roots in h ow th e in trodu ction of a n ew im provem en t program in teracts with th e ph ysical, econ om ic, social, an d psych ological stru ctu res in wh ich im plem en tation takes place In oth er words, it’s n ot ju st a tool problem , an y m ore th an it’s a h u m an resou rces problem or a leadersh ip problem In stead it is a system ic problem , on e th at is created by th e in teraction of tools, equ ipm en t, workers, an d m an agers T he Structure of Improvement We presen t th e lesson s th at h ave em erged from ou r stu dy in th e form of a cau sal loop diagram Ou r m odel provides both a u sefu l fram ework for th in kin g abou t th e ch allen ges associated with im plem en tin g im provem en t program s an d practical su ggestion s to in crease th e ch an ces th at you r n ext su ch effort will su cceed Wh ile th e th eory reported h ere in itially em erged from th e stu dy of two im provem en t in itiatives in a m ajor au tom aker,6 th e resu ltin g m odel is qu ite gen eral an d can be applied to a ran ge of situ ation s We h ave observed th ese dyn am ics in alm ost every organ ization we h ave stu died Figu re begin s with th e basic “ph ysics” u n derlyin g process im provem en t Th e actu al perform an ce of an y process depen ds on two factors: th e am ou n t of Time Spent Working an d th e Capability of th e process u sed to th at work For exam ple, in m an u factu rin g, n et u sable ou tpu t is given by th e produ ct of labor h ou rs per day an d produ ctivity (u sable u n its per labor h ou r) Th e perform an ce of an y process can be in creased by dedicatin g addition al effort to eith er work or im provem en t However, th e two activities n ot produ ce equ ivalen t resu lts Tim e spen t on im provin g th e capability of a process typically yields th e m ore en du rin g ch an ge For exam ple, boostin g th e workweek 20% m igh t in crease ou tpu t 20% , bu t on ly for th e du ration of th e overtim e Gain s in process capability, h owever, boost th e ou tpu t gen erated by every su bsequ en t h ou r of effort Sim ilarly, overtim e devoted to reworkin g defective produ cts can boost n et u sable ou tpu t, bu t on ly as lon g as th e overtim e is tin u ed, wh ile elim in atin g th e root cau ses of th ose defects perm an en tly redu ces th e n eed for rework We captu re th is persisten ce by represen tin g Capability as a stock (den oted by a rectan gle), th at is, as an asset th at accu m u lates im provem en ts over tim e Specifically, Time Spent on Improvement in creases th e flow of Investments in Capability th at au gm en ts process capability Wh ile it often yields th e m ore perm an en t gain , tim e spen t on im provem en t does n ot im m ediately im prove perform an ce It takes tim e to u n cover th e root cau ses of process problem s an d th en to discover, test, an d im plem en t solu tion s, sh own in th e diagram as a delay between im provem en t activities an d th e 66 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened FIGU RE The “Physics”of Improvement Capability + Investment in Capability Capability Erosion DELAY Time S pent on Improvement + Time S pent Working + Actual Performance – Performance Gap + Desired Performance Note:Arrows indicate the direction of caus Signs (‘+’or ‘–’) at arrowheads indicate the polarity of relations :a ‘+’means ality hips that an increas in the independent variable caus the dependent variable to increas e es e,all els being equal (a decreas caus a e e es decreas imilarly ‘–’indicates that an increas in the independent variable caus the dependent variable to decreas (a e);s ,a e es e decreas caus an increas e es e).For more details ee J ,s Sterman,Bus s Dynamics ys ines :S temsThinking and Modeling for a Complex World (New Y ork,NY:Irwin/McGraw-Hill,2000) resu ltin g ch an ge in process capability Moreover, n o im provem en t in capability lasts forever Mach in es wear, processes go ou t of trol with ou t regu lar atten tion , design s becom e obsolete, an d procedu res becom e ou tdated Th u s, we also sh ow an ou tflow from th e stock captu rin g th e in evitable declin e of an y capability th at is n ot regu larly m ain tain ed Th e lag in en h an cin g capability depen ds on th e tech n ical an d organ ization al com plexity of th e process Stu dies sh ow th at th e delay in im provin g relatively sim ple processes su ch as th e yield of m ach in es in a job sh op is on th e order of a few m on th s, wh ile th e delay in im provin g h igh ly com plex processes su ch as produ ct developm en t can be several years or m ore.7 Sim ilarly, th e lifetim e of im provem en ts in capability will be sh orter in organ ization s with h igh rates of ch an ge in produ cts an d people Besides th e ph ysical an d in stitu tion al stru ctu res th at determ in e perform an ce, Figu re also sh ows th e goal for process th rou gh pu t set by sen ior m an agers (labeled Desired Performance) Th e goal cou ld be th e n u m ber of produ cts dem an ded by cu stom ers each day, th e rate at wh ich claim s n eed to be processed by an in su ran ce com pan y, or th e n u m ber of n ew produ cts th e firm seeks to lau n ch th is qu arter People com pare th at goal to th eir actu al perform an ce to determ in e th e Performance Gap Not su rprisin gly, in th e organ ization s we stu died CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 67 Nobody Ever Gets Credit for Fixing Problems that Never Happened FIGU RE The Work Harder Balancing Loop Capability + Investment in Capability Capability Erosion DELAY Time S pent on Improvement + Time S pent Working + + Actual Performance B1 Work Harder Pressure to Do Work + – Performance Gap + Desired Performance Note:The loop identifier,B1,indicates a negative (balancing) feedback.See J Sterman,op.cit it was rare to fin d a process perform in g above expectation s In stead, m an agers, workers, an d en gin eers u su ally faced h igh an d risin g dem an ds, som etim es despite down sizin g an d cu ts in resou rces Th ey were stan tly search in g for ways to im prove an d close th e perform an ce gap Sin ce m ost organ ization s are relu ctan t to in crease plan t an d equ ipm en t or h ire m ore staff, m an agers h opin g to close a perform an ce gap h ave on ly two basic option s First, th ey can try to in crease th e am ou n t of tim e people actu ally spen d workin g Figu re sh ows th is option , wh ich form s a balancing feedback, th e Work Harder loop B1 Th e process represen ted by th is loop works as follows: Man agers facin g a perform an ce gap are u n der pressu re to in crease perform an ce Th ey pressu re people to spen d m ore tim e an d en ergy doin g work An in crease in th e tim e spen t workin g in creases th e perform an ce of th e process an d closes th e perform an ce gap Th is stru ctu re is called a balan cin g feedback loop becau se it stan tly works to balan ce desired an d actu al perform an ce Pressure to Work in clu des, m ost obviou sly, direct m easu res su ch as tellin g people to work faster or pu t in overtim e, settin g m ore aggressive targets for th rou gh pu t, an d im posin g m ore severe pen alties for m issin g th ose targets Pressu re also in clu des m ore su btle action s design ed to extract greater effort from em ployees Th ese in clu de th e frequ en cy with wh ich perform an ce is reviewed, th e detail with wh ich th e reviews are du cted, an d th e sen iority of th ose doin g th e reviewin g At on e com pan y we stu died, it was n ot u n u su al for sen ior 68 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened FIGU RE The Work S marter Balancing Loop Capability + Investment in Capability Capability Erosion DELAY + Time S pent Working Time S pent on Improvement + + Actual Performance + B2 B1 Work Smarter Work Harder Pressure to Do Work + – Performance Gap + Pressure to Improve Capability + Desired Performance vice-presiden ts to review th e perform an ce of in dividu al m ach in es on th e factory floor Not su rprisin gly, su ch atten tion sen t a stron g m essage to all in volved: keep th e m ach in es bu sy at all costs Sim ilarly, a project m an ager we in terviewed recalled th at wh en a su bsystem for wh ich h e was respon sible fell beh in d sch edu le, h is boss requ ired h im to call in every hour with a statu s report u n til th e prototype m et its specification s A secon d option to close a perform an ce gap is to im prove th e capability of th e process In Figu re we represen t th is option as an oth er balan cin g feedback process, th e Work Smarter loop B2 Here, m an agers respon d to a perform an ce sh ortfall by in creasin g th e pressu re on people to im prove capability Th ey m ay lau n ch im provem en t program s, en cou rage people to experim en t with n ew ideas, an d in vest in train in g If su ccessfu l, th ese in vestm en ts will, with tim e, yield im provem en ts in process capability, boost th rou gh pu t, an d close th e perform an ce gap Of cou rse, everyon e kn ows th at it is better to work sm arter th an to work h arder: An h ou r spen t workin g produ ces an extra h ou r’s worth of ou tpu t, wh ile an h ou r spen t on im provem en t m ay im prove th e produ ctivity of every su bsequ en t h ou r dedicated to produ ction Yet, despite its obviou s an d docu m en ted ben efits, workin g sm arter does h ave lim itation s First, as sh own in th e diagram , th ere is often a su bstan tial delay between in vestin g in im provem en t activities an d reapin g th e ben efits Fu rth er, th e greater th e com plexity of CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 69 Nobody Ever Gets Credit for Fixing Problems that Never Happened th e process, th e lon ger it takes to im prove.8 Secon d, in vestm en ts in capability can be risky Im provem en t efforts don ’t always fin d th e root cau se of defects, n ew tools som etim es don ’t produ ce th e desired gain s, an d experim en ts often fail Wh ile in vestm en ts in capability m igh t even tu ally yield large an d en du rin g im provem en ts in produ ctivity, th ey little to solve th e problem s m an agers face right now Th u s, it is n ot su rprisin g th at m an agers frequ en tly u se th e Work Harder loop to both accom m odate variation s in daily workload an d solve pressin g problem s created by u n expected breakdown s or defects Wh en a m an u factu rin g lin e servin g an im portan t cu stom er goes down , a m an ager is u n likely to react by sen din g th e work team to train in g in reliability im provem en t In stead, th at m an ager is goin g to get th e lin e ru n n in g an d pu sh for overtim e u n til th e sh ipm en t is ou t th e door Of cou rse wh en th e lin e is back ru n n in g an d th e produ ct h as been sh ipped, th e m an ager sh ou ld retu rn atten tion to th e im provem en t activities th at will preven t fu tu re breakdown s, an d m ake u p for th e im provem en t tim e th at was lost du rin g th e cru n ch However, it doesn ’t u su ally h appen In stead, wh at we repeatedly observe, an d wh at is m ore difficu lt to u n derstan d, are organ ization s in wh ich workin g h arder is n ot m erely a m ean s to deal with isolated in ciden ts, bu t is in stead stan dard operatin g procedu re Rath er th an u sin g th e work h arder loop to occasion ally offset daily variation s in workload, m an agers, su pervisors, an d workers all com e to rely stan tly on workin g h arder to h it th eir targets an d, sequ en tly, n ever fin d th e tim e to in vest in im provem en t activities Wh at starts as a tem porary em ph asis on workin g h arder qu ickly becom es rou tin e The Reinvestment Loop To u n derstan d wh y, it is h elpfu l to sider h ow workin g sm arter an d workin g h arder are n ected Th e m ost im portan t in tercon n ection arises becau se organ ization s rarely h ave excess resou rces In creasin g th e pressu re to work leads people to spen d less tim e on n on -work related activities like breaks an d to pu t in overtim e (th at is, th ey u se th e Work Harder loop) For kn owledge workers su ch overtim e is often u n paid an d spills in to n igh ts an d weeken ds, stealin g tim e from fam ily an d com m u n ity activities Th ere are, h owever, obviou s lim its to lon g h ou rs After a wh ile th ere is sim ply n o m ore tim e If th e perform an ce gap tin u es to rise, workers h ave n o ch oice bu t to redu ce th e tim e th ey spen d on im provem en t as th ey strive to m eet th eir ever-in creasin g objectives Figu re adds th e n ection between pressu re to work an d th e am ou n t of tim e spen t on im provem en t Th e addition al lin k creates th e Reinvestment loop Un like th ose described so far, th e Reinvestment loop is a positive feedback th at ten ds to rein force wh ich ever beh avior cu rren tly dom in ates An organ ization th at su ccessfu lly im proves its process capability will experien ce risin g perform an ce As th e perform an ce gap falls, workers h ave even m ore tim e to devote to im provem en t, creatin g a virtu ou s cycle of im proved capability an d in creasin g atten tion to im provem en t 70 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened FIGU RE The Reinvestment Reinforcing Loop Capability + Investment in Capability Capability Erosion DELAY R1 + Time S pent Working Reinvestment + Actual Performance + Time S pent on Improvement – + B1 Work Harder Pressure to Do Work + – Performance Gap + B2 Desired Performance Work Smarter Pressure to Improve Capability + Note:The loop identifier,R1,indicates a pos itive (reinforcing) feedback.See J Sterman,op.cit Con versely, if m an agers respon d to a th rou gh pu t gap by in creasin g work pressu re, em ployees in crease th e am ou n t of tim e spen t workin g an d cu t th e tim e spen t on im provem en t Capability begin s to decay As capability erodes, th e perform an ce gap grows still m ore, forcin g a fu rth er sh ift towards workin g h arder an d away from im provem en t Here th e rein vestm en t loop operates as a viciou s cycle, drivin g th e organ ization to ever-h igh er degrees of work pressu re an d m in im al levels of process capability Not su rprisin gly, su ch a viciou s cycle qu ickly drives ou t m ean in gfu l im provem en t activity Here, for exam ple, is th e way a m an ager in an electron ics assem bly plan t explain ed th e persisten t failu re of th e organ ization to en gage in process im provem en t: “Su pervisors n ever h ad tim e to m ake im provem en ts or preven tative m ain ten an ce on th eir lin es th ey h ad to spen d all th eir tim e ju st tryin g to keep th e lin e goin g, bu t th is m ean t it was always in a state of flu x, wh ich , in tu rn , cau sed th em to wan t to h old lots of protective in ven tory, becau se everyth in g was so u n predictable A qu ality problem m igh t n ot be discovered u n til we h ad produ ced a pile of defective parts Th is of cou rse m ean t we didn ’t h ave tim e to figu re ou t wh y th e problem h appen ed in th e first place, sin ce we were n ow really beh in d ou r produ ction sch edu le It was a kin d of sn owball effect th at ju st kept gettin g worse.” CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 71 Nobody Ever Gets Credit for Fixing Problems that Never Happened Shortcuts and the Capability Trap Th e Reinvestment loop m ean s a tem porary em ph asis on on e option at th e expen se of th e oth er is likely to be rein forced an d even tu ally becom e perm an en t Organ ization s th at in vest in im provem en t will experien ce in creasin g capability an d fin d th at th ey h ave m ore tim e to allocate to workin g sm arter an d less n eed for h eroic efforts to solve problem s by workin g h arder In th e su ccessfu l in itiatives we stu died, leadersh ip often worked to stren gth en th e rein vestm en t process by explicitly allocatin g th e resou rces freed u p by produ ctivity gain s to fu rth er im provem en t Un fortu n ately, h owever, th ese in itiatives were th e exception rath er th an th e ru le In m ost of th e organ ization s in ou r stu dy th e rein vestm en t loop worked as a viciou s cycle an d preven ted im provem en t program s from gettin g off th e grou n d Even wh en im provem en t program s yielded in itial resu lts, cost an d sch edu le pressu res soon tem pted m an y organ ization s in to down sizin g or h igh er perform an ce goals th at drain ed resou rces away from im provem en t, weaken in g th e rein vestm en t loop an d cau sin g capability to stall or even fall.9 Un derstan din g wh y th e rein vestm en t loop typically worked in th e down ward, viciou s direction rath er th an th e u pward, virtu ou s direction requ ires th at we add a fin al lin k to th e m odel (see Figu re 5) As discu ssed above, cu ttin g in vestm en ts in m ain ten an ce an d im provem en t in favor of workin g h arder erodes process capability an d h u rts perform an ce However, capability does n ot drop righ t away It takes tim e for process in tegrity to depreciate In th e m ean tim e, th e decision to skim p on im provem en t—skippin g im provem en t team m eetin gs, n eglectin g to take m ach in es down for sch edu led m ain ten an ce, or ign orin g docu m en tation requ irem en ts—boosts th e tim e available to get work don e righ t n ow We captu re th is in tercon n ection by addin g a n egative lin k between Time Spent on Improvement an d Time Spent Working Wh en th e perform an ce gap rises an d m an agers resort to in creased work pressu re, overworked people cu t back im provem en t activity to free still m ore tim e for produ ction Th e perform an ce gap falls, closin g a th ird feedback th at works to balan ce desired an d actu al perform an ce We label th is th e Shortcuts loop (B3) to captu re th e idea th at in creased th rou gh pu t com es at th e cost of departin g from stan dard rou tin es an d processes, cu ttin g corn ers, an d redu cin g th e tim e spen t on learn in g an d im provem en t Sh ortcu ts are tem ptin g becau se th ere is often a su bstan tial delay between cu ttin g corn ers an d th e sequ en t declin e in capability For exam ple, su pervisors wh o defer preven tive m ain ten an ce often experien ce a “grace period” in wh ich th ey reap th e ben efits of in creased ou tpu t (by avoidin g sch edu led down tim e) an d save on m ain ten an ce costs On ly later, as equ ipm en t ages an d wears th ey begin to experien ce lower yields an d lower u ptim es (see section 4) Sim ilarly, a software en gin eer wh o forgoes docu m en tation in favor of com pletin g a project on tim e in cu rs few im m ediate costs; on ly later, wh en sh e retu rn s to fix bu gs discovered in testin g does sh e feel th e fu ll im pact of a decision m ade weeks or m on th s earlier.10 Th u s, th e Shortcuts loop is effective in closin g th e th rou gh pu t 72 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened FIGU RE The S hortcuts Balancing Loop Capability + Investment in Capability Capability Erosion R1 DELAY Reinvestment Time S pent on Improvement – + + Time S pent Working – + Actual Performance + B3 B1 Shortcuts Work Harder Pressure to Do Work + – Performance Gap + B2 Desired Performance Work Smarter Pressure to Improve Capability + gap on ly becau se capability does n ot ch an ge im m ediately wh en th e tim e dedicated to learn in g an d im provem en t declin es To illu strate th ese dyn am ics, Figu re sh ows two sim u lation s of th e m odel in wh ich we sh ow h ow a h ypoth etical process reacts to workin g h arder versu s workin g sm arter Both sim u lation s begin in th e sam e equ ilibriu m state Th e first sim u lation sh ows th e respon se to an in creased em ph asis on workin g h arder As m ore effort is dedicated to work, gross th rou gh pu t im m ediately rises Tim e spen t im provin g falls im m ediately, bu t capability does n ot Perform an ce th erefore rises Th e ben efit of workin g h arder is, h owever, sh ort-lived With less tim e devoted to im provem en t, capability gradu ally erodes, even tu ally m ore th an offsettin g th e in creased tim e spen t workin g Workin g h arder creates a “betterbefore-worse” situ ation Con versely, as seen in th e secon d sim u lation , in creasin g th e tim e spen t on im provem en t redu ces ou tpu t in th e sh ort ru n Even tu ally, h owever, capability rises m ore th an en ou gh to offset th e drop in work effort an d perform an ce is perm an en tly h igh er, a “worse-before-better” dyn am ic Th e in teraction between th e balan cin g Shortcuts loop an d th e rein forcin g rein vestm en t loop creates a ph en om en on we call th e Capability Trap an d h elps explain wh y organ ization s often fin d th em selves stu ck in a viciou s cycle of declin in g capability Man agers an d workers in n eed of an im m ediate perform an ce CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 73 Nobody Ever Gets Credit for Fixing Problems that Never Happened FIGU RE Simulations of the Working Harder and Working Smarter Strategies SYST EM RESPO N SE TO : W O RKI N G H A RD ER W O RKI N G SM A RT ER Actual Performance Actual Performance → Time ؄ Effort → Time ؄ Effort Time Spent Improving Time Spent Improving Time Spent Working Time Spent Working → Time ‫؂‬ Capability → Time ‫؂‬ Capability → Time → Time boost can get it by skim pin g on im provem en t an d m ain ten an ce However, capability even tu ally declin es, cau sin g th e Reinvestment loop to work as a viciou s cycle Man agers wh o rely on workin g h arder an d sh ortcu ts to m eet im m ediate th rou gh pu t n eeds soon fin d th e process fallin g sh ort of its objectives, requ irin g a fu rth er sh ift towards workin g h arder an d away from im provem en t To see th e capability trap in action , sider h ow a m an u factu rin g su pervisor in an au to 74 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened com pan y explain ed th e in ability of h er organ ization to m ake a com m itm en t to regu lar im provem en t activities: “In th e m in ds of th e [operation s team leaders] th ey h ad to h it th eir pack cou n ts Th is m ean t if you were h avin g a bad day an d you r yield h ad fallen you h ad to ru n like crazy to h it you r target You cou ld say, ‘you are m akin g 20% garbage, stop th e lin e an d fix th e problem ,’ an d th ey wou ld say, ‘I can ’t h it m y pack cou n t with ou t ru n n in g like crazy.’ Th ey cou ld n ever get ah ead of th e gam e.” By keepin g th e lin e goin g rath er th an stoppin g to fix th e problem , th ese team leaders relied on th e Shortcuts loop to h it th eir th rou gh pu t objectives However, by “ru n n in g like crazy” th ey also cau sed th e Reinvestment loop to operate as a viciou s cycle, drivin g th e lin e to a m in im al level of capability an d forcin g th em to ru n ever faster Th e capability trap is n ot lim ited to m an u factu rin g—we h ave observed it in firm s ran gin g from fin an cial services to stru ction For exam ple, th e capability trap preven ted a produ ct developm en t organ ization we stu died from developin g n ew processes th at wou ld h ave in creased produ ctivity Like m an y firm s, th ey sou gh t to create an en gin eerin g library or “booksh elf” of reu sable design s an d software However, as described by an en gin eerin g m an ager, “An en gin eer m igh t n ot take th e tim e to docu m en t h er steps or pu t th e resu lts of a sim u lation on th e booksh elf an d becau se of th at sh e saved en gin eerin g tim e an d did h er project m ore efficien tly Bu t in th e lon g ru n it preven ted u s from bein g able to deploy th e reu sability cepts th at we were lookin g for.” Ju st as m ach in e operators an d su pervisors in th e first exam ple faced a basic trade-off between produ cin g an d im provin g, developm en t en gin eers were forced to trade off gettin g th eir assign ed tasks don e again st docu m en tin g wh at th ey learn ed so th at oth ers m igh t ben efit En gin eers cou ld m ake m ore rapid progress towards th eir objectives by takin g sh ortcu ts an d ign orin g th e booksh elf, bu t doin g so preven ted th em from in itiatin g th e self-rein forcin g rein vestm en t loop th at wou ld h ave led to im proved process capability T he Persistence of the Capability Trap Becau se workin g h arder an d takin g sh ortcu ts produ ce m ore im m ediate gain s an d h elp solve today’s problem s, m an agers u n aware of th e in h eren t “better before worse” trade-off are likely to ch oose th em over workin g sm arter Un fortu n ately, th ese tem porary gain s com e at th e expen se of th e lon g-ru n h ealth of th e process By pressu rin g people to work h arder, m an agers often u n wittin gly force th eir organ ization s in to th e capability trap wh ere ever-in creasin g levels of effort are requ ired to m ain tain perform an ce Of cou rse, th is ph en om en on is n ot lim ited to large organ ization s Man y readers will recogn ize th is dyn am ic in differen t aspects of th eir person al lives In situ ation s ran gin g from learn in g h ow to u se a n ew software package to com m ittin g to a n ew exercise CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 75 Nobody Ever Gets Credit for Fixing Problems that Never Happened program , we often fail to th e th in gs th at will im prove ou r lon g-ru n produ ctivity an d well-bein g du e to th e sh ort-ru n stresses of oth er obligation s A qu estion n atu rally arisin g at th is poin t is: “Wou ldn ’t m an agers even tu ally figu re th is ou t?” Wh ile it is u n derstan dable th at, on occasion , people get cau gh t in th e capability trap, wou ldn ’t th ey even tu ally realize th e tru e sou rce of th eir problem s an d rebalan ce th eir efforts between workin g h arder an d workin g sm arter? Un fortu n ately, th e data su ggest th at overcom in g th e capability trap is rare Man agers often n ot realize h ow deeply th ey are trapped in it In stead, th e lesson s th at people learn wh en cau gh t in th e capability trap often lead to action s th at m ake th e situ ation worse Faulty Attributions Su ppose you are a m an ager faced with in adequ ate perform an ce You r operation is n ot m eetin g its objectives an d you h ave to som eth in g abou t it As we h ave ou tlin ed so far, you h ave two basic ch oices: get people to work h arder or get th em to work sm arter To decide, you h ave to m ake a ju dgm en t abou t th e cau se of th e low perform an ce If you believe th e system is u n derperform in g du e to low capability, th en you sh ou ld focu s on workin g sm arter If, on th e oth er h an d, you th in k th at you r workers or en gin eers are a little lazy, u n disciplin ed, or ju st sh irkin g, you n eed to get th em to work h arder How you decide? Research su ggests th at people gen erally assu m e th at cau se an d effect are closely related in tim e an d space: To explain a pu zzlin g even t, we look for an oth er recen t, n earby even t th at m igh t h ave triggered it People also ten d to assu m e each even t h as a sin gle cau se, u n derestim ate tim e delays, an d fail to accou n t for feedback processes How th ese cau sal attribu tion s play ou t in a work settin g? Con sider a m an ager observin g a m ach in e operator wh o is produ cin g an u n u su ally h igh n u m ber of defects Th e m an ager is likely to assu m e th at th e worker is at fau lt: Th e worker is close in space an d tim e to th e produ ction of defects, an d oth er operators h ave lower defect rates Th e tru e cau se, h owever, m ay be distan t in space an d tim e from th e defects it creates Perh aps th e defect is actu ally th e resu lt of an in adequ ate m ain ten an ce procedu re or th e poor qu ality of th e train in g program In th is case, th e delay between th e tru e cau se an d th e defective ou tpu t is lon g, variable, an d often u n observable As a resu lt, m an agers are likely to clu de th at th e cau se of low th rou gh pu t is in adequ ate worker effort or in su fficien t disciplin e, rath er th an featu res of th e process Th e attribu tion of a problem to th e ch aracteristics—an d ch aracter flaws—of in dividu als in a system rath er th an to th e system in wh ich th ey fin d th em selves is so pervasive th at psych ologists call it th e “fu n dam en tal attribu tion error.”11 Su ppose m an agers clu de th at people, n ot th e process, are th e sou rce of low perform an ce Havin g m ade su ch an attribu tion it m akes sen se to in crease produ ction pressu re As discu ssed above, an in crease in produ ction pressu re h as two effects Worker effort im m ediately rises, closin g th e perform an ce gap as th e m an ager in ten ded However, workers are n ow less able to ach ieve th eir 76 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened objectives by in creasin g th e tim e th ey spen d workin g To tin u e to h it th eir ever-in creasin g targets, th ey even tu ally resort to sh ortcu ts, cu ttin g th e tim e spen t on im provem en t However, as h igh ligh ted above, th e Shortcuts loop, wh ile h avin g th e desired effect in th e sh ort ru n , yields a lon g-ru n side effect With less effort dedicated to im provem en t, capability begin s to declin e Perform an ce falls, offsettin g th e in itial gain s By tin u ally in creasin g th rou gh pu t objectives in th e pu rsu it of better perform an ce, m an agers wh o m istaken ly attribu te low perform an ce to th e attitu des an d disposition s of th eir workforce in adverten tly force th e system in to th e capability trap Superstitious Learning Th e bias towards blam in g people rath er th an th e system in wh ich th ose people are em bedded m ean s m an agers are pron e to pu sh th eir organ ization s in to th e capability trap As workers spen d m ore an d m ore of th eir tim e on th rou gh pu t an d cu t back on fu n dam en tal im provem en t, sh ou ldn ’t m an agers realize th at th e tru e cau se of su b-stan dard perform an ce is low process capability rath er th an u n m otivated workers? Un fortu n ately, in m an y situ ation s m an agers learn th e opposite lesson Man agers can n ot observe all th e activities of th e workers Hen ce, after th ey apply produ ction pressu re, th ey can n ot easily determ in e h ow m u ch of th e resu ltin g rise in th rou gh pu t is du e to in creased work effort (th e Work Harder loop) an d h ow m u ch to cu ttin g back on train in g, im provem en t or m ain ten an ce (th e Shortcuts loop) For exam ple, su ppose th ere is a perform an ce gap requ irin g an addition al six h ou rs of produ ctive effort per person per week Man agers, believin g em ployees are sim ply n ot workin g h ard en ou gh , in crease produ ction pressu re Workers bu ckle down , cu ttin g back on breaks, web-su rfin g an d oth er n on produ ctive tim e Su ppose th ese respon ses yield on ly two h ou rs per person per week To close th e rem ain in g th rou gh pu t gap, workers resort to sh ortcu ts an d gradu ally redu ce th e tim e th ey spen d on process im provem en t, train in g, an d experim en tation u n til th ey free th e n eeded fou r h ou rs per week Man agers observe th at th rou gh pu t rises by th e equ ivalen t of six h ou rs of produ ctive effort Becau se m an agers n ot fu lly observe th e redu ction in train in g, experim en tation , an d im provem en t effort (th ey fail to accou n t for th e Shortcuts loop), th ey overestim ate th e im pact of th eir get-tou gh policy, in ou r exam ple by as m u ch as a factor of th ree Th e feedback m an agers receive does n ot correct th e error To th e trary, m an agers qu ickly learn th at boostin g produ ction pressu re works—th rou gh pu t rose wh en th ey tu rn ed u p th e pressu re Th e gain s resu ltin g from produ ction pressu re provide powerfu l eviden ce firm in g th eir su spicion s th at workers were n ot givin g th eir fu ll effort We call th is syn drom e th e Self-Confirming Attribution Error: On ce m an agers decide th at th e workforce is th e sou rce of th eir difficu lties, th ey take action s th at provide vin cin g an d im m ediate eviden ce firm in g th is erron eou s attribu tion Th e cycle of self-con firm in g attribu tion s drives th e organ ization to h igh er levels of produ ction pressu re an d fewer resou rces dedicated to process CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 77 Nobody Ever Gets Credit for Fixing Problems that Never Happened im provem en t Far m ore im portan tly, h owever, it gradu ally ch an ges th e m en tal m odels of th e m an agers by providin g th em with in creasin gly com pellin g eviden ce th at th e sou rce of low th rou gh pu t can be fou n d in th e poor attitu des an d weak ch aracter of th e workforce Recall th e project m an ager discu ssed above wh o was requ ired to provide h ou rly statu s reports on a balky prototype Soon afterward th e problem was solved, firm in g th e boss’s belief th at h e h ad acted appropriately, in deed h ad decisively taken ch arge of th e situ ation , even th ou gh th e team was already workin g arou n d th e clock an d h is in terferen ce drain ed preciou s tim e from th eir efforts to solve th e problem More su btly, th e lon g-ru n effects of produ ction pressu re also rein force m an agers’ belief th at workers are th e problem Th e delay between in creased produ ction pressu re an d in creased th rou gh pu t (via th e Work Harder an d Shortcuts loops) is sh ort, an d th e n ection between work effort an d ou tpu t is u n am bigu ou s In trast, th e erosion of process capability cau sed by produ ction pressu re is delayed, gradu al, an d diffu se It is distan t in tim e an d space from its cau se Man agers are u n likely to attribu te th e cau se of a th rou gh pu t gap to th e pressu re th ey placed on workers m on th s or even years before In stead, th ey are likely to clu de th at th e workers h ave on ce again slacked off, requ irin g an oth er in crease in produ ction pressu re Workers often u n wittin gly spire in stren gth en in g th e m an agers’ attribu tion s Faced with in ten se produ ction pressu re, people are n atu rally relu ctan t to tell su pervisors th ey can ’t m eet all th eir objectives Th e m ore effectively workers cover u p th e sh ortcu ts th ey take to m eet th eir th rou gh pu t targets, th e less aware m an agers will be of th e lon g-ru n costs of produ ction pressu re Un aware th at im provem en t activity, m ain ten an ce, an d problem solvin g h ave been cu t, th rou gh pu t appears to rise with ou t requ irin g an y sacrifices, rein forcin g m an agem en t’s attribu tion th at th e workers really were n ot workin g h ard en ou gh Wh en m an agers even tu ally discover th ese sh ortcu ts, th eir view of workers as u n tru stworth y is firm ed Man agers are th en , as th ey see it, forced to m on itor worker effort even m ore closely (e.g., m ore frequ en t statu s reports, stiffer pen alties for m issin g targets, software for m on itorin g key-stroke rates of data en try operators) Wh at starts as an erron eou s attribu tion abou t th e skills, effort, an d ch aracter of th e workers becom es tru e Man agers’ worst fears are realized as a sequ en ce of th eir own action s Con sisten t with ou r th eory, we are n ot attribu tin g th ese dyn am ics to u n skilled, in experien ced, or ill-in ten tion ed m an agers Rath er, th e stru ctu re of th e system in adverten tly leads even m an y talen ted an d dedicated m an agers in to th e capability trap, wh ile at th e sam e tim e providin g com pellin g eviden ce th at th e sou rces of th eir difficu lties lie in factors beyon d th eir trol, su ch as lazy workers, a “difficu lt” u n ion , fau lty m ach in ery, or fickle cu stom ers Man agers are u n likely to escape th e capability trap becau se th ey rarely realize th ey are in it In stead, as capability stagn ates despite repeated attem pts at im provem en t, th ey slowly, perh aps relu ctan tly, bu t with in creasin g viction , com e to believe th at th eir problem s lie in th e attitu des an d ch aracter of th e people th at work for 78 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened th em Havin g m ade su ch an attribu tion , th e action s th ey take, wh ile ration al from th eir perspective, m ake th e situ ation worse How Superstitious Learning Thw arts Improvement Programs Wh at h appen s wh en an organ ization stu ck in th e capability trap attem pts to im plem en t an im provem en t program ? Perform an ce is low an d work pressu re in ten se In su ch en viron m en ts, im provem en t program s add to th e workload— th e organ ization is so far beh in d th at it can n ot afford to cu t back th rou gh pu t In deed, in m an y organ ization s, m an agem en t im poses aggressive stretch objectives for both th rou gh pu t an d im provem en t in th e belief th at aggressive goals are n eeded to sh ake th in gs u p an d m otivate people In on e firm we stu died, th e gen eral m an ager laid ou t h is goals for im provin g th e produ ct developm en t process by sayin g: “We n eed a developm en t process th at is fast, is th e best in th e in du stry, an d it n eeds to in crease th rou gh pu t by 50% in two years An d everyon e m u st adh ere to th e sam e process.” At th e sam e tim e, th ey lau n ch ed m an y n ew developm en t projects in an ticipation of th e expected produ ctivity gain s Viewed th rou gh th e len s of m an agem en t’s m en tal m odel th ese decision s were en tirely ration al However, th at m en tal m odel, dition ed by th e self-con firm in g attribu tion error dyn am ics discu ssed above, led th em to th e erron eou s belief th at th e delay between im provem en t effort an d resu lts was sh ort an d th at th eir en gin eers were u n deru tilized, u n disciplin ed, u n m otivated, an d u n willin g to adh ere to th e specified process Th e com pan y spen t m illion s an d in vested cou n tless person -h ou rs to create a n ew produ ct developm en t process Th e n ew process in clu ded better tech n ical tools, su ch as im proved CAD/ CAE/ CAM system s, bu t also in creased m on itorin g, in clu din g a stru ctu red stage-gate review process an d m an dated u se of project m an agem en t software Wh ile th ere were som e pockets of su ccess, in m ost cases th e effort h ad little im pact Th e leaders of th e ch an ge effort often attribu ted its failu re to th e en gin eers’ lack of disciplin e: “En gin eers—by trade, defin ition , an d train in g—wan t to forever tweak th in gs It’s a Wild West cu ltu re.”—Man ager A “We wen t th rou gh a period wh ere we h ad so little disciplin e th at we really h ad th e ‘process du jou r.’ Get th e job don e an d h ow you did it was u p to you ” —Man ager B “A lot of th e en gin eers felt th at [th e n ew process] was n o valu e-add an d th at th ey sh ou ld h ave spen t all th eir tim e doin g en gin eerin g an d n ot fillin g ou t project worksh eets It’s bru sh ed off as bu reau cratic.”—Man ager A “It was fair to say th at a lot of en gin eers viewed th is as a n eat way to get som e fan cy tools an d to h ell with process.”—Man ager C CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 79 Nobody Ever Gets Credit for Fixing Problems that Never Happened Yet, wh en we asked en gin eers wh y th e effort failed, we got a differen t story: “We n ever h ad tim e to take th e cou rses an d get th e equ ipm en t we n eeded to really m ake th is stu ff work It was really exh au stin g tryin g to learn h ow to u se th e tools an d th e design at th e sam e tim e ”—En gin eer A “People h ad to th eir n orm al work as well as [u se th e n ew project m an agem en t system ] Th ere ju st weren ’t en ou gh h ou rs in th e day, an d th e work wasn ’t goin g to wait.”—En gin eer B “Un der th is system th e n ew workload was all in crease In som e cases you r workload cou ld h ave dou bled.”—En gin eer C “How did we catch u p? We stayed late Most of th e team was workin g from 7:00 a.m to 8:00 p.m an d on weeken ds A lot of people worked righ t th rou gh th e Ch ristm as vacation ”—En gin eer D “Th e n ew process is a good on e Som eday I’d like to work on a project th at actu ally u ses it.”—En gin eer E Wh ile m an agers felt th e en gin eers h ad little in terest in followin g th e process, en gin eers becam e in creasin gly fru strated with leaders th ey felt h ad n o u n derstan din g of wh at was really requ ired to develop n ew produ cts Faced with th e dou ble bin d of h ittin g aggressive perform an ce targets an d equ ally aggressive im provem en t targets, th ey were forced to cu t corn ers wh ile still appearin g to follow th e process As on e en gin eer rem arked, “In m an y ways we worked arou n d th e [n ew] system Good, bad, or in differen t th at’s wh at h appen ed We h ad a du e date an d we did wh atever it took to h it it.” As m an agem en t discovers th e en gin eers’ sh ortcu ts an d workarou n ds, th eir view th at th e en gin eers can ’t be tru sted is firm ed, an d th ey are forced to step u p th eir m on itorin g Faced with sim ilar difficu lties in its effort to im plem en t a n ew produ ct developm en t process, a differen t firm even created a cadre of “com plian ce m an agers” wh ose sole job was to en force adh eren ce to th eir n ew developm en t process Workers in su ch organ ization s qu ickly learn to h ide problem s from oth ers In all of th e organ ization s we stu died, en gin eers rou tin ely n eglected to reveal th e existen ce of seriou s design issu es for fear of retribu tion from m an agers In on e firm , th e m otto of th e developm en t en gin eers was “n ever reveal you h ave a problem u n til you also h ave th e solu tion ” In an oth er, en gin eers called th e weekly progress review m eetin gs th e “liars’ clu b”—each participan t overstated th e progress of h is su bsystem an d h id kn own defects from oth ers in th e h ope th at oth ers wou ld be discovered first, givin g th em tim e to catch u p.12 Th e sequ en ce is lon g delays in th e discovery of n eeded rework, greatly in creasin g costs, delayin g lau n ch , an d, often , com prom isin g qu ality Th e capability trap goes beyon d low capability an d h igh work pressu re Even tu ally it gets em bedded in deeper stru ctu res, in clu din g in cen tives an d 80 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened corporate cu ltu re As organ ization s grow m ore depen den t on firefigh tin g an d workin g h arder to solve problem s cau sed by low process capability, th ey reward an d prom ote th ose wh o, th rou gh h eroic efforts, m an age to save trou bled projects or keep th e lin e ru n n in g Con sequ en tly, m ost organ ization s reward lastm in u te problem solvin g over th e learn in g, train in g, an d im provem en t activities th at preven t su ch crises in th e first place As an en gin eer at an au to com pan y told u s, “Nobody ever gets credit for fixin g problem s th at n ever h appen ed.” Over tim e, sen ior m an agem en t will in creasin gly sist of th ese war h eroes, wh o are likely to groom an d favor oth er can -do people like th em selves As described by a project leader we in terviewed, “Ou r [com pan y] cu ltu re rewards th e h eroes Fran kly, th at’s h ow I got wh ere I’ve gotten I’ve delivered program s u n der du ress an d difficu lt situ ation s an d th e reward th at com es with th at is th at you are recogn ized as som eon e th at can deliver Th ose are th e opportu n ities for advan cem en t.” Th u s in cen tives an d cu ltu re n ot on ly rein force th e ten den cy toward sh ort-ru n th in kin g an d workin g h arder, bu t also are th em selves sh aped by th at very sh ortterm focu s an d work-h arder m en tality, creatin g an oth er rein forcin g feedback th at in ten sifies th e capability trap.13 An organ ization su fferin g from th e self-con firm in g attribu tion error is poorly position ed to escape th e capability trap Im provem en t program s add stress to th e organ ization , triggerin g greater work pressu re th at preven ts people from in vestin g in im provem en t, an d en cou rages sh ortcu ts In su ch organ ization s m an y im provem en t program s n ever get off th e grou n d If, despite th e work pressu re, people su cceed in allocatin g m ore tim e to im provem en t, th e resu lt is a sh ort-term drop in perform an ce as tim e spen t workin g falls before th e in vestm en ts in im provem en t bear fru it Observin g th at perform an ce is n ot im provin g, m an agers clu de th e particu lar im provem en t m eth od is n ot workin g an d aban don it Sin ce th e n eed to im prove rem ain s, th ey search for an oth er, m ore prom isin g tool, on ly to fin d it too su ffers a sim ilar fate Th e resu lt is growin g cyn icism am on g em ployees abou t “flavor of th e m on th ” program s More in sidiou sly, th ese dyn am ics stren gth en stereotypes an d flicts th at n ot on ly h u rt organ ization al perform an ce bu t dam age society Con sider, for exam ple, h ow a sen ior m an ager explain ed wh y th e produ ct developm en t im provem en t effort h e ran h ad failed: “Program m an agem en t an d th e disciplin es associated with it tin u e to be a problem in m y opin ion in m ost western cu ltu res Th e people th at are particu larly rigorou s an d disciplin ed, th e Japan ese an d th e Germ an s, ten d to be so by cu ltu ral n orm s I can ’t tell you if it’s h ereditary or society or wh ere it is th ey get it bu t th e best en gin eers are th ose th at ten d to be th e m ost disciplin ed, n ot as in dividu al tribu tors bu t as team -based en gin eers So th ere’s a stron g pu sh back from th e western type of en gin eer for m u ch of th is.” Th ere is n o m en tion of th e stru ctu ral featu res of th e system or th e pressu re, felt th rou gh ou t th e organ ization , to deliver am bitiou s projects on tim e CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 81 Nobody Ever Gets Credit for Fixing Problems that Never Happened In stead, th is m an ager blam es th e failu re on th e u n disciplin ed ch aracter of “Western ” en gin eers Su ch attribu tion s, h ere gen eralized to en tire n ation al grou ps, an d in vokin g a distu rbin g racial an d eth n ic su btext, are typical of th e fu n dam en tal attribu tion error As th ese attribu tion s are sh ared an d repeated th ey becom e in stitu tion alized Th ey becom e part of th e corporate cu ltu re, an d, as su ggested by th e qu ote above, can stren gth en pern iciou s stereotypes an d preju dices in society at large Overcoming the Capability Trap So wh at can be don e? Th e m ost im portan t im plication of ou r an alysis is th at ou r experien ces often teach u s exactly th e wron g lesson s abou t h ow to m ain tain an d im prove th e lon g-term h ealth of th e system s in wh ich we work an d live Su ccessfu l im provem en t m u st in clu de a sign ifican t sh ift in th e m en tal m odels of th ose both leadin g an d participatin g in an im provem en t effort Th is in sigh t was captu red su ccin ctly by on e m an ager in a su ccessfu l im provem en t effort: “Th ere are two th eories On e says, ‘th ere’s a problem , let’s fix it.’ Th e oth er says ‘we h ave a problem , som eon e is screwin g u p, let’s go beat th em u p.’ To m ake im provem en t, we cou ld n o lon ger em brace th e secon d th eory, we h ad to u se th e first.” On ce th e cycle of self-con firm in g attribu tion s is broken , an y n u m ber of process im provem en t tools an d m eth ods can h elp im prove capability With ou t th is sh ift, n ew tools an d tech n iqu es, n o m atter h ow great th eir poten tial, are u n likely to su cceed Breakin g th e cycle of self-con firm in g attribu tion s is n ot easy, bu t it can be don e Th e followin g are exam ples from two organ ization s th at overcam e th ese difficu lties an d in trodu ced su ccessfu l im provem en t efforts.14 Du Pont In 1991, an in -h ou se ben ch m arkin g stu dy docu m en ted a gap between Du Pon t’s m ain ten an ce record an d th ose of th e best perform in g com pan ies in th e ch em icals in du stry Th e ben ch m arkin g stu dy revealed an apparen t paradox: Du Pon t spen t m ore on m ain ten an ce th an in du stry leaders bu t got less for it Du Pon t h ad th e h igh est n u m ber of m ain ten an ce em ployees per dollar of plan t valu e, yet its m ech an ics worked m ore overtim e Spare parts in ven tories were excessive, yet th ey relied h eavily on costly expedited procu rem en t of critical com pon en ts Overall, Du Pon t spen t 10-30% m ore on m ain ten an ce per dollar of plan t valu e th an th e in du stry leaders, wh ile overall plan t u ptim e was som e 10-15% lower An experien ced m an ager, Win ston Ledet, an d a team ch arged with im provin g m ain ten an ce operation s, developed a system dyn am ics m odel of th ese issu es Th e m odelin g process in volved exten sive h an ds-on worksh ops in 82 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened wh ich th e team , assisted by an experien ced m odeler, discu ssed, tested, an d ch an ged th e m odel as th ey iden tified areas n eedin g im provem en t Usin g th e m odel as a laboratory to design an d test differen t policies, th e team gradu ally developed an appreciation for th e capability trap an d th e paradox of h igh m ain ten an ce costs an d low reliability To see h ow th e capability trap arose in th e ch em icals in du stry, im agin e th e effects of cost cu ts on m ain ten an ce, su ch as th ose begin n in g with th e oil crisis of 1973 an d su bsequ en t recession In ch em ical plan ts, wh en critical equ ipm en t breaks down , it m u st be fixed Hen ce m ain ten an ce m an agers requ ired to redu ce costs m u st cu t preven tive m ain ten an ce, train in g, an d in vestm en ts in equ ipm en t u pgrades Th e drop in plan n ed m ain ten an ce even tu ally cau ses breakdown s to in crease, forcin g m an agem en t to reassign m ore m ech an ics from plan n ed m ain ten an ce to repair work Breakdown s th en rise even m ore As u ptim e falls, operators fin d it h arder to m eet dem an d an d becom e less willin g to take equ ipm en t down for sch edu led m ain ten an ce, leadin g to m ore breakdown s an d still lower u ptim e More breakdown s sim u ltan eou sly strain reven u e (by lowerin g produ ction ) an d in crease costs (du e to overtim e, expedited parts procu rem en t, th e n on rou tin e an d often h azardou s n atu re of ou tages, collateral dam age, an d so forth ) More su btly, lower u ptim e erodes a plan t’s ability to m eet its delivery com m itm en ts As it develops a repu tation for poor delivery reliability, bu sin ess volu m e an d m argin s fall fu rth er Th e plan t slowly slides in to th e capability trap, with h igh breakdown s, low u ptim e, an d h igh costs Policy an alysis sh owed th at escapin g th e capability trap n ecessarily m ean t perform an ce wou ld deteriorate before it cou ld im prove: Wh ile tin u in g to repair breakdown s, th e organ ization h as to in vest addition al resou rces in plan n ed m ain ten an ce, train in g an d part qu ality, raisin g costs Most im portan tly, in creasin g plan n ed m ain ten an ce reduces u ptim e in th e sh ort ru n becau se operable equ ipm en t m u st be taken off-lin e for th e plan n ed m ain ten an ce to be don e On ly later, as th e Reinvestment loop begin s to work in th e virtu ou s direction , does th e breakdown rate drop Fewer u n plan n ed breakdown s give m ech an ics m ore tim e for plan n ed m ain ten an ce As m ain ten an ce expen ses drop th e savin gs can be rein vested in train in g, parts qu ality, reliability en gin eerin g, plan n in g an d sch edu lin g system s, an d oth er activities th at fu rth er redu ce breakdown s For exam ple, u pgradin g to a m ore du rable pu m p seal im proves reliability, allowin g m ain ten an ce in tervals to be len gth en ed an d in ven tories of replacem en t seals to be cu t High er u ptim e also yields m ore reven u e an d provides addition al resou rces for still m ore im provem en t All th e positive feedbacks th at on ce acted as viciou s cycles draggin g reliability down becom e virtu ou s cycles, progressively an d cu m u latively boostin g u ptim e an d cu ttin g costs Now th e ch allen ge facin g th e team was im plem en tation Th ey kn ew n oth in g cou ld h appen with ou t th e willin g participation of th ou san ds of people, from th e lowest-grade h ou rly m ech an ic to region al vice presiden ts Th ey also realized th at th eir views h ad ch an ged becau se th ey h ad participated in th e CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 83 Nobody Ever Gets Credit for Fixing Problems that Never Happened m odelin g process Som eh ow th ey h ad to facilitate a sim ilar learn in g process th rou gh ou t th e plan ts Th e team verted th e m ain ten an ce m odel in to an in teractive roleplayin g sim u lation th ey called th e Man u factu rin g Gam e.15 Th e gam e is closely based on th e m odel an d realistically captu res th e tim e delays, costs, an d oth er param eters ch aracterizin g typical plan ts Th ey em bedded th e gam e in an in teractive worksh op design ed to create an en viron m en t for learn in g th at addressed em otion al as well as cogn itive issu es Th e process at Du Pon t’s Wash in gton Works com plex in Parkersbu rg, West Virgin ia, was typical: Th e team was in itiated with a two-day learn in g lab learn in g th e cepts of defect elim in ation an d experien cin g th e Man u factu rin g Gam e Th e m aterial is presen ted in th e form of lectu res, skits an d participative exercises in an off-site en viron m en t Posters an d m u sic are u sed Th e atm osph ere is m u ch differen t th an rou tin e plan t m eetin gs or train in g, to open u p th eir th in kin g Th rou gh in teractive exercises, th e team develops th eir person al aspiration s for im provin g th e area wh ere th ey h ave ch osen to work [Th en ] th ey develop an action plan to im m ediately start workin g.16 Despite its m an y sim plification s, th e gam e qu ickly becom es in m an y ways a real plan t with real em otion s an d flicts am on g players In itialized with h igh breakdown s an d low u ptim e, th e people playin g th e role of operation s m an agers face in ten se pressu re to keep equ ipm en t ru n n in g an d often rebu ff attem pts to in crease plan n ed m ain ten an ce, ju st as in th e real world Players wh o stick with th e prevailin g cost-m in im ization , work-h arder, reactive m ain ten an ce policies can keep costs low for a wh ile However, as defects accu m u late, u ptim e slowly sin ks wh ile costs rise Team s wh o follow a plan n ed m ain ten an ce strategy first fin d costs rise wh ile u ptim e falls Soon , h owever, costs begin to fall an d u ptim e rises Th e gam e allows people to experien ce th e worse-before-better dyn am ic in a few h ou rs in stead of a few m on th s For m an y, th e gam e was th e first tim e in th eir careers th ey experien ced th e possibility th at im provem en t was actu ally possible Th e gam e an d learn in g laboratory proved popu lar However, playin g it on ce was n ot en ou gh Th e team fou n d th at th ey h ad to ru n several worksh ops for a given plan t before a critical m ass em erged to lead action team s an d pu t proactive m ain ten an ce policies in to practice In dividu al plan ts n eeded th e capability to ru n th e gam e so th eir own people, with th eir site-specific experien ce an d legitim acy, cou ld ru n it on dem an d By th e en d of 1992, som e 1200 people h ad participated in th e worksh op, an d m ore th an 50 facilitators h ad been certified At plan ts th at im plem en ted th e program by th e en d of 1993, th e m ean tim e between failu re (MTBF) for pu m ps (th e focu s of th e program ) rose by an average of 12% each tim e cu m u lative operatin g experien ce dou bled Direct m ain ten an ce costs fell an average of 20% In 23 com parable plan ts n ot im plem en tin g th e program th e learn in g rate averaged ju st 5% an d costs were up an 84 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened average of 7% Wash in gton Works boosted produ ction capability 20% , im proved cu stom er service 90% , an d cu t delivery lead tim e by 50% , all with m in im al capital in vestm en t an d a drop in m ain ten an ce costs For th e com pan y as a wh ole, servative estim ates exceed $350 m illion / year in avoided m ain ten an ce costs alon e However, su ccess creates its own ch allen ges On e issu e related to th e persisten ce of th e cost-savin g m en tality A m em ber of th e m odelin g team com m en ted, “As soon as you get th e problem s down , people will be taken away from th e effort an d th e problem s will go back u p.” In fact, m an dated corporate cost-cu ttin g program s did cau se sign ifican t down sizin g th rou gh ou t th e en tire com pan y, weaken in g th e rein vestm en t feedback an d lim itin g th eir ability to expan d th e program Win ston Ledet took early retirem en t an d began workin g with oth er com pan ies in terested in th e gam e an d learn in g lab Th ese firm s in clu de oth er ch em icals m an u factu rers alon g with firm s in th e en ergy, au tom otive, an d h igh -tech sectors British Petroleum On e of th e organ ization s Ledet worked with after leavin g Du Pon t was British Petroleu m ’s refin ery in Lim a, Oh io Fou n ded in 1886 by Joh n D Rockefeller, an d on ce “Qu een of th e Fleet,” th e refin ery en gaged in cost cu ttin g du rin g th e 1980s th at triggered th e viciou s cycle of in creasin g breakdown s, h igh er m ain ten an ce costs, an d less plan n ed m ain ten an ce, pu sh in g it in to th e capability trap By th e early 1990s, Lim a lagged well beh in d oth er U.S refin eries BP began to th in k abou t sellin g or closin g th e facility In 1994, th e Lim a facility in trodu ced th e m ain ten an ce learn in g lab an d oth er system dyn am ics tools It was n ot a top m an agem en t in terven tion : Th e origin al ch am pion s were an equ ipm en t specialist, a m ain ten an ce train in g su pervisor, an d an en gin eer Su ccessfu l pilot projects led to favorable word of m ou th ; even tu ally 80% of all em ployees participated in th e program Soon dozen s of im provem en t team s were in place Du rin g th e first six m on th s, m ain ten an ce costs balloon ed by 30% Havin g experien ced it in th e gam e, m an agem en t was prepared for th e worse-before-better dyn am ic, an d focu sed on th e im provem en ts gen erated by th e action team s In Jan u ary 1996, BP an n ou n ced th at it in ten ded to sell th e Lim a refin ery an d stepped u p its cost cu ttin g an d down sizin g A few m on th s later BP stu n n ed th e em ployees by an n ou n cin g th at it cou ld n ot fin d a bu yer at a satisfactory price an d wou ld th erefore close th e refin ery Th e an n ou n cem en t was a deep blow to th e workers an d th e com m u n ity On e of th e m ost im portan t bu sin esses in th e com m u n ity, th e refin ery em ployed 450 people an d pu m ped m ore th an $60 m illion per year in to Lim a’s depressed econ om y Som e em ployees becam e discou raged an d qu estion ed th e valu e of th e learn in g lab an d im provem en t program A few tran sferred to oth er BP facilities or left altogeth er Win ston Ledet described wh at h appen ed n ext: CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 85 Nobody Ever Gets Credit for Fixing Problems that Never Happened TABLE Improvement at the Lima Refinery • Pump MTBF up from 12 to 58 months (failures down from more than 640 in 1991 to 131 in 1998) Direct s avings :$1.8 million/year • Hydrocarbon flare-off down from 1.5% to 0.35%,s aving $0.27/barrel and improving environmental quality • On-line analyzer uptime improved from 75% and not trus to 97% and trus ted ted,permitting real-time optimization of product flow.Savings :$0.10-0.12/barrel • Safety incidents and los hours cut by a factor of t • Thirty-four production records s et • Cas margin improved by $0.77 per barrel of oil proces ed h s • Total new value created:$43 million/year.Total cos t:$320,000/year.Ratio:143:1 • Learning initiative under way for other BP facilities around the world Source:Paul Monus ,“Proactive Manufacturing at BP’s Lima Oil Refinery” pres , ented at National P etroleum RefinersAs ociation s Maintenance Conference,May 20-23,1997,New Orleans Griffith,D.Kuenzli,and P ;J Monus ,“Proactive Manufacturing:Accelerating Step Change Breakthroughs in P erformance,” NPRA Maintenance Conference,MC-98-92,1998;Paul Monus ,pers onal communication “For th ose wh o decided to stay with th e sh ip, a n ew spirit em erged Th ey realized th at th ey n eeded a fu tu re in Lim a an d sh ou ld take respon sibility for creatin g th at fu tu re Th e first step was to en su re th at th e exit of m an y experien ced people did n ot th row th em back in th e reactive m ode It actu ally created a clearer focu s for th e people wh o rem ain ed Th ey were all th ere becau se th ey h ad ch osen to be th ere.” Soon th e im pact of th e n ew m ain ten an ce policies an d attitu des was clearly visible (Table 1) Th ese dram atic im provem en ts did n ot go u n n oticed On Ju ly 2, 1998, th e ban n er h eadlin e of th e Lima News an n ou n ced “Oil Refin ery Rescu ed.” Clark USA, a privately h eld Fortune 500 com pan y with refin in g an d distribu tion in terests, agreed to bu y th e Lim a refin ery for $215 m illion an d keep it operatin g Th e Du Pon t an d BP cases illu strate th e power of a sh ift in m en tal m odels Th e m odel, gam e, an d worksh op n ot teach an yon e h ow to m ain tain equ ipm en t For exam ple, a BP team redu ced bu tan e flare-off to zero, savin g $1.5 m illion / year an d redu cin g pollu tion Th e effort took two weeks an d cost $5000, a retu rn on in vestm en t of 30,000% / year Mem bers of th e team h ad kn own abou t th e problem an d h ow to solve it for eigh t years Th ey already h ad all th e en gin eerin g kn ow-h ow th ey n eeded, an d m ost of th e equ ipm en t an d m aterials were already on site Wh at h ad stopped th em from solvin g th e problem lon g ago? Th e on ly barrier was th e m en tal m odel th at th ere were n o resou rces or tim e for im provem en t, th at th ese problem s were ou tside th eir trol, an d th at th ey cou ld n ever m ake a differen ce Th e m odelin g process an d th e resu ltin g gam e were effective becau se th ey elim in ated m an y of th e im pedim en ts to learn in g in th e real system Dyn am ics 86 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 Nobody Ever Gets Credit for Fixing Problems that Never Happened su ch as th e progressive slide in to th e capability trap th at n orm ally play ou t over years or even decades cou ld be experien ced in ju st a few h ou rs Un like th e real world, people cou ld take differen t roles: A m ech an ic playin g th e role of plan t m an ager m igh t fin d h im self with low u ptim e an d th en cu t preven tive m ain ten an ce to avoid equ ipm en t takedown s an d cu t costs Seein g people from differen t fu n ction s an d backgrou n ds en actin g th e sam e beh aviors h elped break th e viciou s cycle of self-con firm in g attribu tion errors an d blam e Th e system s th in kin g process en abled people to experien ce for th em selves th e lon g-term , organ ization -wide sequ en ces of th eir action s Th ey discovered h ow to u se in itial su ccesses to create resou rces for fu rth er im provem en t an d h ow to su rvive th e sh ort-ru n drop in perform an ce Th ey saw h ow sm all action s cou ld sn owball in to m ajor gain s Most im portan tly, th ey learn ed th at th ey cou ld, after all, m ake a differen ce N o te s See G Easton an d S Jarrell, “Th e Effects of Total Qu ality Man agem en t on Corporate Perform an ce: An Em pirical In vestigation ,” Journal of Business, (1998): 253307; K Hen dricks an d V.R Sin gh al, “Qu ality Awards an d th e Market Valu e of th e Firm : An Em pirical In vestigation ,” Management Science, 43/ (1996): 415-436 See Easton an d Jarrell, op cit.; Darrell Rigby, “Man agem en t Tools an d Tech n iqu es: A Su rvey,” California Management Review, 43/ (Win ter 2001): 139-159 See J Pfeffer an d R Su tton , The Knowing-Doing Gap (Boston , MA: Harvard Bu sin ess Sch ool Press, 2000); K Klein an d J Sorra, “Th e Ch allen ge of In n ovation Im plem en tation ,” Academy of Management Review, (1996): 1055-1080 Pu blish ed su m m aries can be fou n d in : J Sterm an , N Repen n in g, an d F Kofm an “Un an ticipated Side Effects of Su ccessfu l Qu ality Program s: Explorin g a Paradox of Organ ization al Im provem en t,” Management Science, 43/ (1997): 503-521; N Repen n in g an d J Sterm an , “Gettin g Qu ality th e Old Fash ion : Self-Con firm in g Attribu tion s in th e Dyn am ics of Process Im provem en t,” in R.B Cole an d R Scott, eds., Improving Theory and Research on Quality Enhancement in Organizations (Th ou san d Oaks, CA: Sage, 2000), pp 201- 235; E Keatin g an d R Oliva, “A Dyn am ic Th eory of Su stain in g Process Im provem en t Team s in Produ ct Developm en t,” in M Beyerlein an d D Joh n son , eds., Advances in Interdisciplinary Studies of Teams (Green wich , CT: JAI Press, 2000); R Oliva, S Rockart, an d J Sterm an , “Man agin g Mu ltiple Im provem en t Efforts: Lesson s from a Sem icon du ctor Man u factu rin g Site,” in D Fedor an d S Gh osh , eds., Advances in the Management of Organizational Quality (Green wich , CT: JAI Press, 1998), pp 1-55; J Carroll, J Sterm an , an d A Marku s, “Playin g th e Main ten an ce Gam e: How Men tal Models Drive Organ ization Decision s,” in R Stern an d J Halpern , eds., Debating Rationality: Nonrational Elements of Organizational Decision Making (Ith aca, NY: ILR Press, 1997) For form al m odels of im plem en tation see: Sterm an , Repen n in g, an d Kofm an op cit.; N Repen n in g, “Drive Ou t Fear (Un less You Can Drive It In ): Th e Role of Agen cy an d Job Secu rity in Process Im provem en t Efforts,” Management Science, 46/ 11 (2000): 1385-1396; N Repen n in g, “A Dyn am ic Model of Resou rce Allocation in Mu lti-Project Research an d Developm en t System s,” System Dynamics Review, 16/ (2000): 173-212; N Repen n in g, “A Sim u lation Based-Approach to Un derstan din g th e Dyn am ics of In n ovation Im plem en tation ,” Organization Science (forth com in g) Repen n in g an d Sterm an (2000), op cit CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 87 Nobody Ever Gets Credit for Fixing Problems that Never Happened See A Sch n eiderm an , “Settin g Qu ality Goals,” Quality Progress (April 1988), pp 55-57; J Sterm an , N Repen n in g, an d F Kofm an , op cit Sch n eiderm an , op cit For an exam ple, see Sterm an , Repen n in g, an d Kofm an , op cit 10 Th e dyn am ics of project m an agem en t an d software developm en t h ave been treated exten sively in system dyn am ics See, for exam ple, T Abdel-Ham id, “Th e Econ om ics of Software Qu ality Assu ran ce: A Sim u lation -Based Case Stu dy,” MIS Quarterly, (1997): 395-411; D Ford an d J D Sterm an “Dyn am ic Modelin g of Produ ct Developm en t Processes,” System Dynamics Review, 14 (1998): 31-68 11 For an excellen t su m m ary of attribu tion th eory, see S Plou s, The Psychology of Judgment and Decision Making (New York, NY: McGraw-Hill, 1993) 12 See also George Roth an d Art Klein er, Car Launch (New York, NY: Oxford Un iversity Press, 2000) 13 See N Repen n in g et al., “Past th e Tippin g Poin t: Th e Persisten ce of Fire-Figh tin g in Produ ct Developm en t,” pu blish ed in th is issu e [California Management Review, 43/ (Su m m er 2001)] 14 Th ese cases are discu ssed in depth in J Sterm an , Business Dynamics: Systems Thinking and Modeling for a Complex World (New York, NY: Irwin / McGraw-Hill, 2000), ch apter 15 See Win ston Ledet, “En gagin g th e En tire Organ ization : Key to Im provin g Reliability,” Oil and Gas Journal, 97/ 21 (1999): 54-57 16 R Tewksbu ry an d R Steward, “Im proved Produ ction Capability Program at Du Pon t’s Wash in gton Works,” Proceedin gs of th e 1997 Society for Main ten an ce an d Reliability an n u al feren ce 88 CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 ... effect th at ju st kept gettin g worse.” CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 71 Nobody Ever Gets Credit for Fixing Problems that Never Happened Shortcuts and the Capability Trap... ediate perform an ce CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 73 Nobody Ever Gets Credit for Fixing Problems that Never Happened FIGU RE Simulations of the Working Harder and Working... dedicated to process CALIFORNIA MANAGEMENT REVIEW VOL.43,NO.4 SUMMER 2001 77 Nobody Ever Gets Credit for Fixing Problems that Never Happened im provem en t Far m ore im portan tly, h owever, it gradu