Spero Consulting

How a Bud­get­ing Prob­lem May Not Really Be a Bud­get­ing Problem.

Else­where on our site, we crit­i­cize firm-​wide, over-​reliance on a sin­gle, uni­form per­for­mance mea­sure like profit-​sharing or share price. The inten­tion may be good, but the result need not be. The basis of our crit­i­cism is the pre­sump­tion that most employ­ees are risk averse, and for most of them there exists other, bet­ter per­for­mance mea­sure that are sta­tis­ti­cally more sig­nif­i­cant than a global one for the firm. Using more sen­si­tive and pre­cise mea­sures allows the orga­ni­za­tion to either: (1) induce the same types of effort or deci­sions at a lower com­pen­sa­tion cost (by reduc­ing the nec­es­sary risk pre­mium in com­pen­sa­tion) or (2) induce more goal-​compatible actions and deci­sions at the same expected com­pen­sa­tion cost.

Com­bin­ing that crit­i­cism with this cur­rent essay reminds us of Goldilocks and of the ben­e­fits of mod­er­a­tion — i.e., not too few, not too many.

We recall a meet­ing and con­ver­sa­tion with a senior finance offi­cer a few years ago, where he com­plained about the bud­get­ing process at his firm, Mega­Corp, with its obvi­ously made-​up name.

In his indus­try, merg­ers and con­sol­i­da­tions were fre­quent because there were sub­stan­tial economies of scale for most ser­vice pro­cess­ing activ­i­ties. In fact, the promised sav­ings from those economies were the main jus­ti­fi­ca­tions for the wave of merg­ers. By cen­tral­iz­ing the ser­vices, if Mega­Corp had accu­rate demand plan­ning, it could real­ize huge cost sav­ings and equally large increases in prof­its and mar­ket value.

Unfor­tu­nately, the firm’s actual demand was rather volatile and often very dif­fer­ent than planned. There were peri­ods of sub­stan­tial excess capac­ity and peri­ods of severe capac­ity con­straints, and in both extremes the ben­e­fits of con­sol­i­da­tion were elim­i­nated if not greatly reduced. To say that coör­di­na­tion was lack­ing was an understatement.

Our acquain­tance blamed the firm’s regional pres­i­dents (RPs) for this predica­ment. Many RPs had run the smaller orga­ni­za­tions that had been acquired by Mega. Because of their equity stakes in the acquired com­pa­nies, they had ben­e­fited from the merg­ers, but most other Mega­Corp employ­ees hadn’t, yet.

He noted that the RPs were par­tic­u­larly poor pre­dic­tors of both new cus­tomers and of the resource usage growth due to exist­ing customers.

He had his own the­ory about why they were poor prog­nos­ti­ca­tors. He stated that many of these indi­vid­u­als had become quite wealthy when they sold their small, regional firms to Mega­Corp and had assumed their cur­rent posi­tions. He also men­tioned that many of these folks were vis­i­ble mem­bers of their com­mu­ni­ties. He argued that since they had acquired their wealth, their goals had changed. They were now focused on main­tain­ing their per­sonal pres­tige rather than run­ning their busi­nesses as ded­i­cated Mega­Corp employees.

Curi­ously, he did admit that per­sonal rela­tion­ships were cru­cial to the firm’s suc­cess in these smaller mar­kets, and that caused us to won­der whether any of his incen­tive hypoth­e­sis was based upon envy. At the time, the regional pres­i­dent had made their mil­lions, and our acquain­tance had not.

Given his the­ory, he was con­vinced the bud­get­ing process was bro­ken and needed to be fixed. He wanted our advice about how to fix it, and we agreed to help because we knew that regard­less of the valid­ity of his the­ory, the bud­get­ing process could still be broken.

We under­stand that there are dif­fer­ent types of peo­ple: high types, low types — effec­tive and inef­fec­tive ones. We aren’t known for view­ing the world through rose-​colored glass, and we can eas­ily attribute any level of suc­cess — out­side of our own — to luck rather than to abil­ity. How­ever, many of the regional pres­i­dents had suc­cess­fully run their smaller orga­ni­za­tions in the past with­out the ben­e­fit of the low costs result­ing from national economies of scales. So, it seemed unlikely that the inac­cu­rate fore­casts were due to sheer incom­pe­tence. More­over, if they were wealthy enough not to care about work, they could likely quit; so, neglect did not seem to be a com­pelling rea­son either.

It could have been an issue with the bud­get­ing process. Poorly designed cost allo­ca­tion schemes or trans­fer pric­ing sys­tems can cause wild swings in inter­nal demand esti­mates — even in sta­ble envi­ron­ments. Plus, the indus­try was grow­ing slightly, not shrink­ing. We won­dered about the usual prob­lems of elic­it­ing pri­vate infor­ma­tion, but decided to start where we always do — at the begin­ning — by ask­ing, “What do you want the regional pres­i­dents to do?”

We thought that given the jus­ti­fi­ca­tion for MegaCorp’s mul­ti­ple recent merg­ers and acqui­si­tions, the answer would be short and sim­ple (and cru­cial): make accu­rate forecasts…of the vol­ume of shared pro­cess­ing resources, either directly or indi­rectly by fore­cast­ing sales.

We were wrong — very wrong — by two orders of mag­ni­tude. (No, we were not wrong about how he should have answered the ques­tion, but our pre­dic­tion of his answer was incorrect.)

His ini­tial answer was, “It depends.” Hav­ing sat through a few of our sem­i­nars, he thought it was clever response — his mim­icry of our class­room mantra. As it turns out, his response was truth­ful but his rea­son­ing was not wealth-​maximizing.

Mega­Corp used a management-​by-​exception pol­icy for the regions. Every month cor­po­rate head­quar­ters would inform the regional pres­i­dents of the three-​to-​five per­for­mance mea­sures that appeared most out-​of-​control in their loca­tions and would instruct them to fix the prob­lems caus­ing these variations.

We asked about the total num­ber of per­for­mance mea­sures. He replied that he hadn’t counted them all, but there were many. So, we asked to see a com­pre­hen­sive list of per­for­mance measures.

We received six pages with almost 70 rows of mea­sures (of small fonts) on each page. ¹ Yes, there were approx­i­mately 400 “per­for­mance” mea­sures, and many of these mea­sured var­i­ous aspects of the region’s oper­at­ing per­for­mance and effi­cien­cies — there were many ratios and vari­ance calculations.

Now, before con­tin­u­ing it is worth not­ing that the prob­a­bil­ity is very low that one could find 400 inde­pen­dent and rel­e­vant vari­ables mea­sur­ing dis­tinct aspects of a person’s per­for­mance. It was more likely that many of these vari­ables were highly cor­re­lated with one another or that many vari­ables were actu­ally depen­dent vari­ables — lin­ear com­bi­na­tions of oth­ers, par­tic­u­larly if they were extracted from account­ing reports. How­ever, this point isn’t cru­cial to the case.

We asked, “Which of these do you emphasize?”

His response: “That depends, there are usu­ally three or four — some­times five — that we empha­size each month. Usu­ally, we won’t have four dif­fer­ent issues. Instead, we’ll get mul­ti­ple sig­nals of one or two under­ly­ing problems.”

“Are they the same sig­nals for everyone?”

“No.”

“Are they the same mea­sures for the same peo­ple every month?”

“Uh, no”

“So, every month, you are telling most of these folks to focus on dif­fer­ent aspects of their region’s performance?”

“Well, yes, we’re just show­ing areas for improve.”

We asked, “How do the regional pres­i­dents know what you want them to do?”

We con­tin­ued, “There is a good chance that they are con­fused by head­quar­ters monthly draw­ing of con­cern fac­tors. More­over, there is a good chance that they would do what you want them to do if they knew what it was, but to them it seems that the orga­ni­za­tion can­not make up its mind.”

“But that is not true. We know what we want. We want them to make accu­rate fore­casts and take care of prob­lems that arise.”

“You might say that, but your actions speak louder than words. Twelve times per year you are ask­ing the RPs to (pos­si­bly) redi­rect their ener­gies to “new” prob­lems. In fact, if there are no glar­ing weak­nesses in a region, then it is more likely that you are induc­ing them to run willy-​nilly to dif­fer­ent illu­sory prob­lems each month. (We’ll explain that later with our recommendations.)”

He dis­agreed. “That doesn’t make sense. We spent a lot of time and money devel­op­ing our reports, and every month we employ sub­stan­tial resources to mon­i­tor per­for­mance. In fact, we have sev­eral ana­lysts com­pile and review the reports.”

“The Soviet Union did, too, but that’s not the point. First, note that if there were glar­ing prob­lems, then you would likely see ser­ial cor­re­la­tions among the out-​of-​control measures. That means that prob­lems would likely per­sist through time; this month’s trou­bles would be like last month’s.”

“We con­cluded that we saw no per­sis­tence because we were man­ag­ing effec­tively. In other words, through our data min­ing, we would iden­tify prob­lems, and the regional man­agers would rapidly fix them.”

“Well, sup­pos­ing that your con­clu­sion is true, your mea­sure­ment sys­tem is still suboptimal.”

“How can you say that?”

“Very eas­ily by ask­ing our first ques­tion, again: what do you want them to do?”

“Make accu­rate fore­casts and fix problems.”

“So, among the three-​to-​five per­for­mance vari­ables, where is the empha­sis on forecasts?”

“Well, it is not explicit, but it is in there.”

“Yeah, it prob­a­bly is in there, but with 400 vari­ables every­thing else is in there, too, and it appears that you are giv­ing those other fac­tors equal if not more-​than-​equal weight.

As you know, many of the fac­tors relate to oper­at­ing effi­cien­cies. Let’s sup­pose that each month your ana­lysts iden­tify real prob­lems, and as we will explain later, that is a sup­po­si­tion at this point, but let’s go with it.

At the mar­gin, cor­rect­ing those prob­lems will increase Mega’s prof­itabil­ity but only mar­gin­ally. Your cur­rent scheme makes the firm pen­ny­wise, but pound foolish.”

He inter­rupted, “You know, some­one did men­tion we don’t seem to make a lot of progress despite all of the activity.”

“Well, yeah.“ We con­tin­ued, “Could the acqui­si­tions have been jus­ti­fied on the basis of elim­i­nat­ing all regional inef­fi­cien­cies — exclud­ing the gains from the cen­tral­iza­tion of pro­cess­ing services?”

Hav­ing been involved in the analy­sis of each of the acqui­si­tions, he knew the answer was no, and replied, “No, the cru­cial sav­ings involved the cen­tral­iza­tion of ser­vice pro­cess­ing, but the regions still need to be well-​managed. So, what do you rec­om­mend? How do we get accu­rate fore­casts for cen­tral activ­i­ties and still man­age regional performance?”

“We would split the RP posi­tions into two jobs: the one called regional pres­i­dent deal­ing with sales, fore­cast­ing, and com­mu­nity involve­ment and one called regional chief oper­at­ing offi­cers (RCOOs) devoted to day-​to-​day oper­at­ing issues. Given what you told us, it’s likely that most regions already have some­one serv­ing in this role, albeit informally.”

In addi­tion, we rec­om­mended that the cur­rent RPs remain in their cur­rent posi­tions, but that RCOOs report directly to Mega’s head­quar­ters. We did rec­om­mend that the RCOOs report on a dotted-​line basis to the RPs, too. We viewed this for­mal­ity as an overt and nec­es­sary method to ensure that the RPs had access to any rel­e­vant infor­ma­tion pos­sessed by the RCOOs and to ensure that the RCOO’s rec­og­nized their ongo­ing respon­si­bil­ity along this dimen­sion. ²

The regional pres­i­dents could then focus on their key tasks, and we could develop ways to elicit their pri­vate infor­ma­tion that included a care­ful analy­sis of cost behav­ior, exist­ing capac­ity uti­liza­tion, and antic­i­pated changes in tech­nol­ogy to deter­mine how Mega­Corp wanted the RPs to act. Elim­i­nat­ing most, if not all, oper­at­ing per­for­mance mea­sures already greatly sim­pli­fied the plan­ning prob­lem. Lastly, with respect to the RPs, we men­tioned that Mega could offer dif­fer­ent rewards depend­ing upon the mar­ket char­ac­ter­is­tics of the regions, e.g., in smaller and stag­nant mar­kets, gain­ing the RPs’ dif­fer­en­tial infor­ma­tion was gen­er­ally less valu­able and likely not worth the cost.

In addi­tion, we rec­om­mended a more care­ful analy­sis of the impor­tant tasks per­formed by RCOOs and a stream-​lining of their per­for­mance measures: “Suppose that the exist­ing 400 oper­at­ing per­for­mance mea­sures are inde­pen­dent ran­dom vari­ables and each one is influ­enced by a sep­a­rate, inde­pen­dent action taken by the employee per­form­ing 400 dif­fer­ent, worth­while, non-​repetitive actions per month — seems ambi­tious, but just sup­pose. Fur­ther­more, sup­pose that the employee takes the 400 actions that Mega­Corp prefers. ³

If the monthly review inves­ti­gated any mea­sure that was excep­tion­ally bad, say, in the worst one per­cent of that measure’s pos­si­ble val­ues, then, on aver­age, there will be four excep­tional mea­sures per month. ⁴ So, the worker is asked to change his action despite behav­ing appro­pri­ately in the first place. ⁵

Now, sup­pose there are only 100 inde­pen­dent ran­dom vari­ables and the other 300 are deter­min­is­tic func­tions of these vari­ables. For sim­plic­ity, assume that each of the 100 has three related, depen­dent vari­ables. In this case, there would be one true excep­tion per month and three vio­la­tions of related mea­sures. The addi­tional three pro­vide no addi­tional infor­ma­tion; so, 75% of your mea­sures are a waste of time. If the rela­tion­ships aren’t deter­min­is­tic, there is a pos­si­bil­ity of more noise and worse infer­ences, which we would observe as rec­om­men­da­tions to fix non-​existent problems.

So, the cur­rent shot­gun approach may be com­pre­hen­sive, but a more care­ful approach that is con­sis­tent with Mega’s pri­or­i­ties will be more prof­itable. That means start by ask­ing: what does Mega want the RCOOs to do and what sig­nals exist of this effort?

Also, given dif­fer­ences in the mar­kets that Mega serves, there is no rea­son to believe that the answer to this ques­tion is con­stant across regions. Mega­Corp should pre­fer the flex­i­bil­ity of using dif­fer­ent mea­sures or dif­fer­ent weights on the same mea­sures depend­ing upon regional mar­ket characteristics.

For­tu­nately, with 400 exist­ing mea­sures, we have a solid plat­form from which to start.

So, we are now able to solve your two most press­ing prob­lems. The first involves elic­it­ing the RPs’ pri­vate infor­ma­tion for plan­ning pur­poses and that is a much smaller prob­lem now — in the sense that we can now trade effi­ciency gains of man­ag­ing cen­tral­ized capac­ity against oppor­tu­nity costs of insuf­fi­cient capac­ity and our esti­mates will be more pre­cise. The sec­ond involves pre­vent­ing oper­at­ing prob­lems within the regions and quickly fix­ing those that do occur. By ele­vat­ing indi­vid­u­als to for­mal RCOO roles, your com­pen­sa­tion costs may increase slightly, but that mar­ginal cost is a small frac­tion of the ben­e­fits that should be derived by improv­ing the coör­di­nated use of shared resources.”