Has Business Squandered The IT Payoff?

Issue Date: 01-06-97
Issue Number: 8.01
Category: IT COSTS

Are IT costs getting too out of control? Is there any way to tame them? The Squandered Computer, the latest book from IT economics authority Paul Strassmann, pops some of the myths of IT budgeting. We assess some of Strassmann's latest ideas and discuss some budget trends.

In 1995, large corporations spent more on IT than they earned in profits, according to Paul Strassmann. Globally, the overall profit for nearly 13,000 organisations totalled $750 billion; he estimates that IT expenditures for the same group amounted to over $1 trillion. The numbers beg the question, are organisations getting their money's worth out of their computers?

Strassmann, a retired CIO for three major multinational corporations and the US Department of Defence, has written several books regarding planning, managing, and justifying information technology investments. In his newest work, The Squandered Computer, Strassmann argues that arbitrary control over expenditures is not enough to guarantee decent results from IT investments. Instead, he calls for organisations to measure the true value that IT projects and acquisitions deliver over their useful - not their economic life, using real figures, not guesses. The overriding theme? 'The era of easy and generous spending on computers is coming to an end.' In this report, we review Strassmann's latest work.

WHERE'S THE BEEF?

IT investments have hardly ever been money winners, according to Strassmann. Prior to the 1980s, most systems were divorced from the revenue stream of the corporation, consisting primarily of financial systems that automated repetitive tasks, saving administrative costs. In other words, the net contribution of such systems to the bottom line was relatively minor.

During the 1960s and 70s, few systems automated, or improved, activities that added value, such as improving product, or contact with the customer. Naturally, there are exceptions, such as systems which processed syndicated market research data; however, it could be argued that even these tended to save money in tabulating the returns, rather than making possible new ways of using or finding intelligence in the data. Stacks of print-outs may have been more thorough, but they were hardly amenable to ad hoc decision support.

Today, IT expenditures are exploding. According to a 1994 Morgan Stanley report, since the early 1980s, IT expenditures have surpassed that for basic industrial equipment. On one hand, that's a sure sign of the emergence of the information economy. Another sign, Strassmann says, is that IT expenditures keep growingÉ and growing. In corporate America, IT spending is $4,970 per employee - a figure that is over 60% higher than 1988, and one that in corporate budgets is dwarfed only by employee health and pension benefits.

Has anything been gained by this? Strassmann cites Steven Roach, a Morgan Stanley economist, who notes that while corporations appear to have higher productivity, the numbers tell only half the story. In reality, the figures are more a reflection of the extent to which companies are now outsourcing key business processes. Anecdotally, Strassmann relates that the Big Three US automakers now outsource between 55% - 65% of their parts.

More precise data on the consequences of outsourcing come from two studies cited by Strassmann, include a measure of revenue per employee (which Strassmann compiled from his own database of 138 major US corporations) and the level of economic value added (EVA) by the largest industrial firms, compiled by Stern Stewart (who devised the EVA metric) in 1996. According to Strassmann's data, revenue per employee increased 26% from 1988-1994; meanwhile, EVA declined by $150,000 per $1 million revenue during a similar (1987-92) period. Clearly, companies are producing more revenue per employee, but adding less value, due to outsourcing.

Therefore, Strassmann concludes, computerisation is not necessarily responsible for recent productivity gains in US industry.

Meanwhile, IT investment life cycles are compressing. The toll is especially bloody on the desktop where most hardware grows obsolete after 3 years. Competition is driving many firms to make unwise IT investments - a form of 'keeping up with the Joneses' syndrome. For instance, says Strassmann, a bank feels compelled to install new computing applications because its competitors already have. Strassmann counters that the problem with this logic is that one would have to prove that profits would have declined without the project. The true measure, he says, is whether the firm could have boosted profits with alternative, lower cost measures.

Software 'decays' just like hardware. Although the process isn't physical (software doesn't break down just because it gets old), the changes in business processes, organisational structures, hardware, and operating environments take their toll on software. Furthermore, the pressure for rapid returns on IT projects results in a plethora of short-term point solutions that are especially prone to obsolescence.

The results are twofold: an accumulating inventory of obsolete, incompatible software, or software that must be constantly updated.

According to Strassmann, the average age of the software portfolio of a Fortune 500 firm is 6 - 8 years. Replacing or upgrading such software would easily double or triple the typical Fortune 500 organisation's annual IT budget, he claims.

Are 'best practices' the answer? Strassmann is cynical. 'Programmers who programmed calendar routines over the past 30 years believed they were applying the best software practices.' Since memory and disk were expensive, those years were coded as compact two-digit fields. According to Gartner Group, the results of these 'best practices' will be a $200 - $300 billion repair job to fix the year field problem before the year 2000. The same goes with adhering to ISO 9000 quality guidelines or the principles of the software engineering Institute, added Strassmann. 'It does not always follow that superior software will result from compliance with checklists that experts believe to be the guarantors of excellence in the software process', he noted.

IT EXPENDITURE PATTERNS SHOW

The only clear pattern is that organisations are spending more on IT than they earn in profits. Strassmann's figures - nearly $1.1 trillion for world-wide IT expenditures versus $750 billion in estimated profits - were culled from several sources, including a 1996 Wall Street Journal report on global IT purchases, Strassmann's estimates of additional internal spending extrapolated from his sample of 220 corporations, and corporate profit and loss figures on 13,000 organisations from Disclosure Inc.

Strassmann comments that, given the growing size and scope of IT projects, failures could seriously dent profits. 'Although the belief still persists that computer applications may give companies competitive advantage, what has changed is the recognition that computer investments are indeed risky', he says. Consequently, CFOs are raising the financial bar for IT projects, in growing cases requiring paybacks in as little as 12-18 months. The result, says Strassmann, is a growing short-term orientation to IT projects, which he believes reduces the likelihood that any strategic benefits will be delivered. The more probable result: companies will end up with a proliferation of incompatible systems that rapidly grow obsolete as the business or organisation changes (he calls them 'build and junk' solutions).

Not surprisingly, Strassmann doesn't believe IT expenditures are improving corporate bottom lines. That runs counter to macro figures of national economic growth. Between 1987-94, the US led the world, both in GDP growth (at 8.1%), and the share of GDP spent on IT (2.8%). By comparison, the closest runners up - Australia, Canada, and New Zealand - enjoyed GDP growth rates of 7.0% and IT spending rates of 2.0% of GDP. Initially, IT expenditures appear to be consistent with national GDP growth. Strassmann found no such correlation.

For instance, during Strassmann's experience as CIO at Xerox in the 1970s, different business units with similar products, similar competitors - and similar computing budgets - still had highly variable financial results. Since then, Strassmann has built a database comprising over 500 US companies, including public firms, client organisations from his consulting practice, and other firms for which he was able to obtain figures. He also referred to a study of IT expenditures from the 200 largest Finnish firms. Neither sample showed any connection between profitability and revenue - although Strassmann did not provide specific data.

But data can also play tricks. He cites a comparison of per capita IT spending in the food industry; Quaker Oats, which in 1995 had the highest return on equity (nearly 70%) of the firms studied, spent less than 25% per capita compared to H.J. Heinz or Archer Daniels Midland (which hovered at the 20% -30% ROE levels). It is difficult to draw conclusions regarding Quaker's 'wise' use of IT resources, given that the CEO recently resigned after a disastrous 1996.

In Quaker's case, a major culprit was a disastrous acquisition of a beverage subsidiary; the company lost touch with distributors who marketed the product. The tempting question is: could an effective sales and marketing decision-support IT investment have maintained Quaker's profitability and preserved the CEO's job?

ATTACKING THE OLD MYTHS

Sacred Laws: Strassmann pokes holes in several popular premises regarding IT economics. For instance, 'Grosch's Law', named after a former IBM researcher and National Bureau of Standards computer technology director, stated that computing power increases at the square of its costs - a concept that promoted economies of scale. It was a concept that was especially popular until the 1980s. Strassmann responded that the concept merely reflected IBM's pricing schemes, which during the heyday of the mainframe, were practically unchallenged. He dismissed the 'law' as 'nothing but a reflection of IBM's marketing strategy.'

Back to the future, today's better-known 'Moore's Law', which holds that microprocessing power doubles every 18 months, has become a Trojan Horse for software vendors, claims Strassmann. As computing power increases, what otherwise seems like a bargain is quickly consumed by larger, more complex software. Admittedly, Strassmann does not elaborate here, although a quick comparison of the storage and computing requirements of Office 95 and Office 97 adequately proves Strassmann's point. 'Intel gives and Microsoft takes away', he concludes.

It could be argued that more compact predecessors to Office 97 carried similar price points ($300 - $500/seat, individual street price, depending on whether the full suite was ordered and whether the suite is a new purchase or competitive upgrade); it could be even argued that new, desktop suites are for all practical purposes cheaper, since they are bundled with major brand Pentium machines. Furthermore, new capabilities, such as Microsoft's incorporation of native TCP/IP capability, could be said to make networked computing cheaper. Strassmann does not address these arguments; he would probably consider them irrelevant anyway, given his reference to Gartner Group's $8,000 - $10,000 annual PC fat client cost figures (which reflect the fact that fat clients are complex to configure, maintain, and while the fat client software requires significant end user training and hand holding).

Budgeting Guidelines: Target number one is personnel. He cites the dilemma of companies that outsource, such as the manufacturer which primarily resells but does not outsource its information workers (e.g., sales, marketing, budgeting, and corporate management). If the company was apportioning IT expenses based on employee counts, it would have unnecessarily reduced IT budgets.

There is one minor problem with Strassmann's analogy. As manufacturing gets more computerised, more and more manufacturing workers are joining the information workforce. Nonetheless, this is a case of the right point, wrong example. Strassmann is stating that information technologies be apportioned, not on a per capita basis, but by the value that is added by organisation business processes. In so doing, it will be easier to estimate how IT increases the efficiency of processes, that in turn contribute to the bottom line.

He strikes down another metric - that of apportioning IT investments as a percentage of revenue - to make the same point. By pegging IT as a percent of corporate revenue, it is essentially positioning IT as a tax without justifying what it is being spent for. And, he singles out trade magazines, such as Information Week and Computerworld, which annually publish profiles of the top corporate users of IT based, in part, on the company's percent of revenue invested in IT. He believes that such metrics glorify spending for its own sake. Therefore, a company which is caught in an exorbitant outsourcing or client-server migration contract might look favourable in the magazine rankings because it is spending a higher percent of revenue on IT than a similar firm which negotiated its contracts more wisely.

Another way not to invest is by looking at discounted cash flows. Hardware and software vendors, he argues, want to get paid now, and not on an inflation-adjusted basis. In large part, this is due to the extreme volatility and short life-cycle of too many computer investments, especially at the desktop level. Admittedly, inflation eats into the value of an asset; in the IT world, obsolescence eats at it more. Therefore, the market value of a short-lived IT asset will probably be less than its inflation-adjusted counterpart.

Put another way, according to a 1994 Morgan Stanley report, in actual dollars, US business is spending more on IT than on basic industrial equipment. This has been the case since 1981-2. 'The gross IT put in place in the last decade had an extraordinarily short life, whereas modern basic industrial equipment now has greater longevity because of its greater reliability', Strassmann notes. After a lull in the late 80s, the trend grew even more pronounced with the emergence of client-server technology in 1991. (For the record, Strassmann claims that client-server is a bit of a misnomer, since, 'there have always been clients and serversÉ The client-server surge is nothing more but the continuation of a much older trend toward dismemberment of the centralised data processing organisation.' He terms the emergence of so-called client-server as 'a testimonial that the worshippers of Moore's Law have displaced the adherents to Grosch's Law.')

Strassmann adds that many IT expenses are either unnoticed or get swept under the carpet. First, there is the issue of 'stealth spending.' These 'stealth costs' may include under-utilised PCs that are installed as employee perks (for personnel of given status), informal end user support (what some analyst groups call 'shadow IT'), training hours for business (non-IT) personnel, and so on. Strassmann cites Gartner Group estimates of $5,000 for end user computing operations (see Figure 1). (Note: One of the areas cited by Gartner, the 'Futz Factor', is the target of a major Microsoft service offering, 'AnswerPoint'. The costs might be petty cash, for instance $55 - $195 per incident, or part of a major line item expenditure, where a 'package' of 75 such incidents paid in advance costs $9,995.)

Some costs seem obvious. Telecommunications is often handled outside the IT organisation, or by non-IT specialists. Consequently, costs get underreported. For instance, when Computerworld co-operated with Network World to publish a list of the top 25 users of telecommunications technologies, it found that average telecommunications costs (at 26% of IT budgets) were roughly four times higher, on average, than figures reported in Computerworld's own 'Premier 100' company IT profiles. The difference? While the 'Premier 100' figures were primarily reported by CIOs, the Network World figures were provided by telecommunications managers.

In fact, Strassmann argues that IT expenditures should not be restricted to computing. To gain a full picture, it should encompass all 'information handling' functions, many of which are conducted outside IT. While telecommunications is the obvious example, others include copying, printing, videoconferencing, and so on.

Intangible Benefits: Strassmann rejects the notion that IT projects should be justified on 'soft' criteria. For instance, he attacks Information Economics, which relied almost entirely on subjective scoring of the benefits and risks of prospective projects. He also labels as 'astrology' methods such as 'Analytical Hierarchies,' which dispenses entirely with numerical scoring, replacing it with subjective opinions (e.g., poor, fair, good, very good).

The only justifiable cost justification methods, says Strassmann, are those techniques that quantify risks and benefits in hard dollars. Be wary of claims that IT projects will lower operational costs. In many cases, he says, the IT solution may not eliminate the cost, but shift it somewhere else inside (or in some cases, outside) the organisation. For instance:

  • Replacing printouts with electronic distribution of routine reports may actually increase printing costs, as more end users gain access to the data and print portions on a more accessible LAN printer.

  • Distributed computing often shifts data centre costs to local costs (e.g., informal technical support from colleagues) that are not captured in overhead budgets.

  • Encouraging trading partners to manage operations (e.g., Wal-Mart's practice of having suppliers manage its inventories).

However, cost shifting can sometimes be beneficial - especially if the business unit or organisation which receives the added responsibilities is able to perform the function more efficiently. For instance, Wal-Mart's supplier managed inventories reduce the overall cost of keeping Wal-Mart's retail shelves stocked.

Fantastic ROI: Strassmann critiques published competitions that often tend to hype the results. He cited the example of a 1995 CIO Magazine award to K Mart for a computerised inventory system that supposedly saved $365 million. Strassmann believes that smart inventory managers, not the computer project, deserved the award. (Incidentally, Strassmann adds, immediately following the award, the CIO resigned, and 25% of the IT staff was made redundant after massive deficiencies in the system were discovered.)

Instead, Strassmann says, the ROI of an IT investment is the difference in costs between the solution and a less expensive alternative. Don't attribute benefits which were due to other factors.

Outsourcing: Strassmann is very cynical about the benefits of outsourcing. 'Corporations that outsourced more than 60% of their IT budget tended to be economic losers', he said, citing his own informal studies of well-known outsourcing contracts, which showed that most of the firms involved had negative EVAs for at least three years prior to their outsourcing contracts. He concludes that large-scale outsourcing, as practiced, is more a fancy term for downsizing, and not a miracle cure. Strassmann argues for outsourcing contracts that charge by transaction types and levels; he believes such contracts would introduce real competition between outsourcing providers and internal IT organisations and contribute positively to the bottom line.

A PROACTIVE BUDGETING APPROACH

Given the range of problems which Strassmann has described, his solutions are almost anti-climatic. They sound more like ordinary common sense that should be obvious to any IT executive. For instance, Strassmann says there is no best way or magic recipe to IT budgeting. Every company and every sector is different. Regardless of whether a company relies on centralised mainframes or distributed client-server networks, buys off-the-shelf applications or develops in house, no single strategy is a formula for success. His recommendations include:

  • Follow the Money: Rather than react to the competition, fears of obsolescence, or technology fads of the moment, Strassmann argues that IT investments should be aligned with value-added business processes.

  • Consider Cheaper Alternatives: Less expensive alternatives should be considered first - even if they don't entail computerisation. For instance, better inventory management practices might be preferable to installing a new application. Even if the resulting automation delivered deeper savings, streamlining processes alone might deliver most of the benefit for a fraction of the cost.

  • Use EVA as Guide for Evaluating IT Expenditures: Economic value added (EVA), a gauge of a company's value-added output, is a better yardstick for evaluating IT expenditures. He notes that EVA is becoming the most popular metric for long-range financial planning in the corporate world, citing a CFO Magazine survey indicating that 65% of US corporations now use it for this purpose. EVA equates to the profits, minus the product of shareholder equity times the cost of capital. Theoretically, says Strassmann, if there is any relationship between IT costs and EVA, IT budget reductions should have a favourable impact on EVA. Unfortunately, his point would have been more effective with a fuller explanation of the utility and logic of EVA.

  • Outsourcing Should be Competitively Priced: Shifting problems or expenses to a third-party firm will not deliver competitive benefit. Competitive pricing, where the outsource provider is forced to compete with internal IT, should deliver better bottom line results.

  • Avoid Documenting Operational Benefits of Projects: Ultimately, end users, not IT staff, control the operational benefits realised from an IT project. IT staff can only influence the degree to which the system performs; they cannot guarantee that the application will be properly used. Leave the documentation of benefits to customers or business units that must justify and use the system. Confine IT budget promises, budget to systems capabilities, and project schedules and staffing.

  • Avoid ROI Metrics: Evaluating returns based on purchase price of an IT asset within an arbitrary time period (generally determined by corporate hurdle rates) fail to account for the full life-cycle costs and benefits of a project.

  • Quantify Residual Value: Evaluate an IT investment based on its useful, not its economic life. The notion here is that, after an item may have been depreciated, or after a new model or software version renders an IT product obsolete, it may still have useful life, aiding the execution of value-added business processes within the firm. Otherwise, as mentioned before, a typical Fortune 500 firm would have to double or triple its IT budget just to replace or upgrade 'obsolete' or fully depreciated software.

    A good example is older IBM mainframes and storage subsystems. Using IBM's new SysPlex architecture, older models don't have to be retired, they can be diverted to the processing of low-priority tasks. Similarly, older 386 and 486 computers may have a useful life performing dedicated, off-line functions for users (or tasks) which do not require Windows capabilities. This approach can counteract pressures for short-term results, which otherwise produce rapidly obsolescent software.

  • Focus on Reusable Software: Strassmann is bullish on approaches, such as Java, which make applications more reusable due to platform- and database-independence. Such reusability in the long run boosts the residual value of software because business processes can be decomposed and reconfigured (to reflect organisational changes) or operated with different databases or platforms (to reflect technology changes). (Unfortunately, Strassmann does not deal with the costs of designing for reuse.)

  • Evaluate 'Open Systems' Based on Avoided Costs: With the 'costs' of client-server well documented, open systems should be evaluated, not by technology criteria (e.g., the ability to operate on multiple platforms), but by the savings they really deliver. Strassmann argues that such savings are possible only if an organisation enforces standards.

    Future IT projects, or project budget increases, should be examined warily, and approved only if they will reduce long-term support costs.

DON'T TRY THIS AT HOME

True to the book's title, The Squandered Computer provides sufficient shock value to make IT executives question the usual assumptions regarding IT budgeting, cost justification, and project management. And, in fact, Strassmann presents compelling data to demonstrate that, based on conventional criteria, IT projects as they have been traditionally managed are extremely poor investments.

In most cases, they are treated as a necessary cost of doing business, and in some cases, as a me-too strategy for keeping up with the competition.

There is little question that traditional IT practices and technologies, with their rate of obsolescence, poor history of documentation, and alarming tendencies to foster scorn for 'not invented here' solutions, have significantly dented the profitability of many enterprises.

Strassmann cites a variety of studies and sources; the book's greatest fault is that the sources of the data, the means by which the data was analysed, and the results, are not always clear.

The most valuable concept that Strassmann introduces is the notion of residual value. He justly attacks 'build and junk' short-term solutions, and in fact, the evidence that older applications and platforms still work fine, thank you, is not hard to find in most organisations.

The age-old question is how to quantify residual value. Strassmann ventures a formula (see Table 1) which takes the initial cost of hardware, software, application development, and training, and subtracts depreciation for each. Identifying the upfront value is the easy part; the challenge is assigning an appropriate depreciation value. The accounting profession and government tax codes offer plenty of depreciation schedules based on the notion of reselling assets, but not on their innate usefulness to the organisation itself.

Strassmann provides some sample depreciation charts. However, the book could have used a more detailed explanation. We believe that it would have revolved something around the notion of the avoided cost of rebuilding applications, retraining personnel, or buying new hardware. Towards that end, we would love to see a Strassmann sequel that explains exactly how to do it.

Table 1. Strassmann's residual value formula.

Change in Information Technology Assets =

(Equipment Acquisition - Equipment Depreciation)
+ (Development Acquisition - Development Depreciation)
+ (Software Acquisition - Software Depreciation)
+ (Training Acquisition - Training Depreciation)

Order from:
The Information Economics Press
P.O. Box 464
New Canaan, CT 06840-0264
USA
Tel: (800) 800-0448
Fax (203) 966-5506
http://www.strassmann.com
426 pages, 66 illustrations

Price $49 plus $4.50 for shipping and handling via UPS in the continental US.
Any other shipping at cost.


This article originally appeared in Computer Finance. For subscriptions, please contact:

North America:
Teresa White
+1 203 973 6700
teresa.white@metagroup.com
http://www.computerwire.com/cfus/

Rest of the World:
Dominic Sharp
+44 (0)171 468 0800
doms@computerwire.co.uk
http://www.computerwire.com/cf/