Bain's Blow to Berkeley

Nothing kills performance like distress.  The same goes for fear, anger, powerlessness, injustice, and a sense of having been screwed over by ignoramuses. The remarkable conversation going on via the recent comments on this blog suggest that the Bain-Berkeley analysis for UCB’s Operation Excellence has dug Berkeley into a deeper operational hole than it was already in.

Why is this report so disturbing? Is there anything in it that might make Berkeley administration better?  The answer to the second question is yes, but the report rests on several foundational errors. These are serious enough to suggest that implementation of the report as it is will impose an outdated, centralized organizational model that will in turn undermine Berkeley’s key strengths as a university.

This post describes the report’s analysis, and shows that its solutions do not fit with this analysis. I argue that the report will, if implemented, make UCB administration less efficient than it already is. I also offer one simple reason that the recommendations will also make faculty’s relations to staff worse rather than better, meaning that faculty should help staff find alternatives to the Bain-OE blueprint rather than vainly hoping that it will solve some of their local staffing issues.

Bain was paid $3 million, did much no doubt unpleasant digging in financial records, and in probable conceptual deference to its a campus working group partners (8) produced 205 slides on potential savings on the Berkeley campus. The report identifies “five opportunity areas” for savings (15): Procurement, information technology (IT), energy management, student services, and a non-“area” called organizational simplification.  They produced some useful data: Berkeley works with over 18,000 separate vendors, mostly lacks the equipment standards that would allow discounts via aggressive negotiations on bulk purchases, has such a diffuse IT system that server farms can be found in more than 50 different buildings, doesn’t actually meter and manage energy consumption across campus, and has spread more than 50 student services across five “control points” (slides 16-20).  By improving these functions while also implementing organizational simplification, Bain says that the campus could cut cost in those areas by around 6-10% each (23), thus achieving annual net savings of $70 million by year four of the OE project (24).  Some slides offer sensible price negotiation and consolidation strategies (e.g. 39-42).  Since UCB’s annual budget is $1.65 billion (2007-08), the savings will not be transformative.  But a bit more than 4% annual savings at the end of 4 years seem realistic, useful, and responsible, given UC’s ambitions and obligations as a public institution.

The relatively modest cost savings are coupled with the absence of innovative thinking in this report, and the question arises about why this report had to exist at all. Weren’t these management problems obvious to Berkeley’s own administration?  Didn’t anyone in the Vice Chancellor for Administration’s office count vendors and plan some remedies in 2006, when “strategic sourcing” was a management fad in UCOP, or back in 1996, when the vendor count probably hit 10,000?   So one naturally wonders what is it about UCB’s administration that required it to pay Bain $3 million to tell it how to do some fairly humble portions of its job.

Clues appear in two of Bain’s slides, and bring us back to the fifth opportunity area, the organization itself.  First, Berkeley is not a “flat” or networked organization, but an old-school military pyramid with 11 separate layers of management (56). It is at the same time a fragmented pyramid (62) in which communication is less than ideal. The combination no doubt makes Berkeley administration at times the worst of both worlds: inefficient and yet top-down authority, which stifles creativity in the middle and lower layers without offering clear mobilization orders based on some greater vision of the whole.  In addition, supervisors have fairly few direct reports, or “narrow spans” (55-57).

So far, the data would suggest to most readers that Berkeley’s problem is that it is too hierarchical, has a heavy bureaucratic burden, partitions or actively hides information to the detriment of efficiency, and intimidates or inhibits lower-level staff rather than informing and empowering them.  This would explain why the VC for Administration might now know just how fragmented procurement had become, and why local units would try to operate as much as possible on their own.  In other words, Berkeley has too many supervisors and they do too much supervising, which makes supervising less efficient. Too much staff effort is spent negotiating authority and ensuring compliance and too little in practical problem-solving, process innovation, open communication, and direct service to faculty and student “customers.”  Both efficiency and innovation depend on informal knowledge, constant information exchange, easy collaboration, and high levels of interest and trust.  Berkeley’s stratification and opacity inhibits all of these.

The second clue is confirmation of the effects of UCB’s elaborate hierarchy on organizational morale.  Bain’s survey found that Berkeley has a very low “net promoter score” (114), meaning that a majority of employees would not recommend Berkeley as a place for a friend or relative to work.  Employees also take a dim view of the organization’s mode of decision-making because it is insufficiently “participatory” (115-16).

Putting these pieces together, we might assume, would mean two remedies. 1) Delayer management, specifically, the middle and upper-middle layers that supervise supervisors rather than offering direct support to teaching and research.  2) Rebuild morale by moving to greater participation by all employees in key decisions. (116).  Bain and the OE working groups would then define “organizational simplification” as moving towards a relative flat organization where a smaller number of supervisors and senior executives redefine their role as coordinating participatory decision-making taking place across a decentralized “learning organization.” Efficiency would be sought through informational transparency rather than through bureaucratic authority. Knowledge and trust would increase, self-protection and suspicion would decrease, zero-sum selfishness would decline, and units would be more likely to accept and adapt to painful changes because the participatory process allowed them to understand and to influence the process, even if the outcome was not what they wanted.

This interpretation would move Berkeley towards a flatter and genuinely participatory management structure. But it requires a non-corporate understanding of the university as a necessarily distributed innovation system.  The university’s core missions – instruction and research – do in fact occur in frontline units rather than in central administrative offices.  Its mission intelligence, in other words, is bottom-up, not top down.  The university’s overall output occurs through parallel processing, in which units work on specific activities – only partially shared with others – in relative autonomy.  The core activities range from computer engineering to studio art and cannot be standardized, nor can their administrative support be standardized. Efficiency is compatible with – indeed, depends on -- unit autonomy, which enables units to deliver services in the way that is most effective for their particular “customers” (first-generation students, or IT needs in physics as opposed to economics).  Unit activities and “outputs” are also therefore modular. The university’s decentralization is an expression of the way its work actually gets done – and must get done.  Though logistical and administrative support can certainly be improved, it cannot be improved by violating the bottom-up, semi-autonomous, modular, and decentralized architecture that has been developed over decades to support academic work as it actually takes place.  A university is more like an open source software development community than it is like a production line.

Unfortunately, the OE report’s solutions get this upside-down.  They rest on two “critical enablers” – a “higher performance operating culture” and a “financial management model” (21).

At first these look like no-brainer clichés, but in fact they form a belief system that the report does not derive from its campus analysis but that it imports as pre-conceived ideas about modern organizations.  If you look at column A on slide 21, you will see that “high-performance operating culture” means linear communication and harder decision paths coupled with more measurement of employee output. Column B puts unit finances more directly in charge of those who speak for “pan-university priorities.”  Both “enablers” seek increased control of units and employees. They envision the use of financial audits (and merit pay) to ensure compliance.  These two empirically unsupported axioms about efficiency are what have provoked the staff concerns we’ve been seeing on this blog.  They are structured to default to top-down power, and to operationalize the authority of higher-ranking units over frontliners.  The report’s “enablers” convert Berkeley administration to what British scholars of higher education call “audit culture,” which predictable increases in administrative costs due to escalations in assessment and control.

It follows logically that the report solves the procurement problem with centralization. The future is described in slogans like the following. “Vendor relationships: Central procurement owns all vendor relationships” (47). In this model, financial savings will come only when central procurement reduces unit autonomy and choice.  One could imagine the relief of dozens of units in giving up vendor relations to a central office if and only if the units knew that the central office saw itself as a support function that would honor unit views about their own (distinct and only partially standardizable) needs.  Art history may require graphics cards or processors that are not part of the standardized package.  But centralized, functionally uniform service centers not only lack the kind of local knowledge that makes the unit’s activities efficient, but will define the overriding of local knowledge as part of their job.

The same move toward reinforcing UC’s existing top-down, hierarchical structure appears throughout the organizational simplification section, which is worth reading carefully (52-71).  The report mixes “increasing supervisory spans” with “improving front-line productivity” (54) – meaning it mingles together cutting supervisors with cutting front-line employees.  The data showing that more than half of all supervisors (1000+ people) have 3 or fewer reports (58) supports the interpretation that Berkeley staff suffer from too many supervisors.  The “illustrative example” shows a laborious reshuffling of units with the net result that 22 employees are reduced to 21 (68). Since OE is looking to cut Berkeley’s $700 M in staff payroll by 6-8% (23) the actual layoff rate would be double this.  In addition, the report’s stress on centralization will push losses to the periphery and the bottom rather than letting them occur in the upper-middle, where the salary and efficiency savings would be greater.

Faculty members should also pay close attention, because their activities will not be helped by the report’s definition of optimization as conforming local units to “pan-university” goals (6, 132). Faculty dissatisfaction with staff can be traced to the fact that staff members already do not work for faculty or for their local unit but for the pan-university.  Your department’s bookkeeper may help you to squeeze the last extra value out of your limited grant money, but this is in fact not his job. His job is to provide the university’s budget office with accurate financial information, to enforce university regulations, and to protect the university from audit problems. The same is true for the IT specialist, who is already charged with minimizing costs and maximizing standardization, cross-university connectivity and interoperability.  She is not charged with optimizing your office’s or your lab’s particular system. Staff are already there to fuel the university’s administrative metabolism, to the frustration of faculty and also of most staff.

Bain’s plan is to make IT people report directly to central and senior IT people rather than to your department chair and/or her IT colleagues in related units.  In other words, local, horizontal, cross-functional relationships will be gutted, and replaced by vertical single-function reports.  Those who think this siloing and verticalization will increase efficiency are about thirty years behind in their reading of the management literature, which abounds with tales of the effectiveness of cross-functional teams and horizontal collaborations based on local needs.  Similarly, Bain-OE promulgates the dumb idea that “generalist” supervisors are a problem (52, 76), although in fact they are the indispensible coordinators in any distributed, necessarily non-standardized innovation system. The first victims of this OE regression towards something that looks even more like a 1950s multidivisional corporation will be staff, and the second victims will be the faculty members who have built up local staff collaborations over time on which, in reality, their productivity depends. But the goal is not to enhance faculty productivity, but to optimize the pan-university.  The big losers will be individual academic departments – their staff and faculty alike.  Staff will be pulled away from frontline operational reports and into the authority field of the central office controlling their specific function.

The slide that best summarizes OE’s regressive solutions is 117.

The image is the bureaucratic version of the alien ship in Independence Day, a looming, inverted pyramid in which top dwarfs bottom and threatens to swallow it whole.  Everything flows top-down, and the administrative content takes the form of goals-metrics-evaluation which are communicated to units (metrics –assessment of performance), on down to supervisors (accountability functions) and then finally to individuals (performance metrics tied to unit goals.) Relationships are reduced to the abstract modalities of compliance embodied in assessment procedures.  Management is not support for the university’s necessarily diverse creative functions but is a state of permanent evaluation.  There is no respect here for the autonomy of the units – departmental staff, student services, and technical staff for laboratories – that are close to the “customer” (cf. “autonomous culture” as a source of inefficiency in procurement, slide 35). The tone is of control through communication, through finance, through even more of the endless audit and evaluations to which UC employees are already subject.  The implicit diagnosis is that Berkeley’s employees are inefficient because they are insufficiently assessed, measured, and financially incentivized.  The diagnosis is anti-humanist, at odds with current literature about both human motivation (intrinsic) and effective organizational behavior (collaboratively organized).  It is also ungrounded in evidence from the Berkeley campus.  The predictable effect, as I noted at the start, is that the model contained in the report is already making staff efficiency worse.

I don’t have first-hand knowledge of how the Berkeley campus process is unfolding this week or this month, but the public documents are not promising. There is the OE czar, central process managers, hand-picked committees making implementation decisions in smoke-free rooms, and roving HR bands hired from the outside.  There is nothing there about collaborative implementation, protections for productive autonomy, bottom-up integration, non-intrusive coordination of the decentralization on which organizational creativity depends.

I see three things that would help:
A freeze on threats of staff layoffs tied to the OE process.
Stated commitment to viewing the distributed nature of the university as a strength, and to seeing management as a coordinative support function.
Full participatory involvement in implementation, starting with rank-and-file faculty and front-line staff.

Only a bottom-up process will keep Bain-OE’s new centralized units and new administrators (more HR consultants, more “experienced sourcing staff” (49) from bogging Berkeley down even further. Industry also used to know this, and research managers from Kodak’s legendary Kenneth Mees in the 1920s to the “skunkworks” at Lockheed during World War II to Bell Labs in its heyday to the Internet (all top-down managed “intranets” were commercial failures) to Apache and similar open source software projects– all managed innovation through mutual respect, good communication and the participatory negotiation of policy and resources. The best way to wreck innovation in organizations is to micromanage the local units.  Unless they want to spend even more time and money encouraging UC’s decline, the OE groups need to turn the OE blueprint on its head.