by Paul A. Strassmann
The chairman of the Federal Reserve, Wall Street bankers and assorted chief business executives have explained the enormous gains in the 1990s stock market by an upward expansion of the US economy which is expected to deliver sustainable superior profits from steadily rising productivity gains. These productivity enthusiasts argue that the new electronic technologies have finally been harnessed by the workforce to deliver improved operating results. To make their case the optimists rely on a selected assortment of measures as well as on prolific anecdotal evidence from business magazines. In a poll of 200 CEOs and chief financial officers, the advances in information technology ranked third as the enablers behind the economic growth in the 1990s.
The productivity skeptics refer to official government statistics that flatly deny that productivity has been rising. According to government economists the annual rate of productivity increases since 1990 has been slowing down as compared with the prior three decades:
Figure 1 - Productivity growth
The stakes in these debates are enormous. The performance of the stock market, the prospects of achieving a balanced budget and the ability to finance Social Security plus Medicare all depend on the expectation of steadily rising productivity gains. The presumption that computers improve hard-to-measure productivity offers a legitimacy to yet another proposal to invest money on computers.
I believe that the prevailing differences of opinion about productivity gains cannot be resolved either with the existing government statistics and certainly not with the selective taking apart of data by the productivity enthusiasts. Only measures that can be explicitly related to corporate financial performance can settle the arguments whether the US is either losing or gaining on productivity growth in the private sector. Reliance on corporate financial data has also the advantage of relocating the speculations from unverifiable fantasies to provable productivity targets.
U.S. industrial corporations include in their financial statements an item known as Sales, General & Administrative Cost (S.G.&A). It represents the costs of coordinating, controlling, guiding, promoting, motivating, training and managing employees, customers and suppliers while making and delivering the goods. This number is a reasonable approximation of the costs of managing information, including the costs of computer hardware, software, networks and staff. The S.G.&A. largely accounts for a firm's overhead expenses. It also reflects the costs devoted to the generation and consumption of all data.
If IT would have increased the productivity of those involved in the handling of information, then it would now take less S.G.&A. to manage every dollar's worth of Cost of Goods Sold (C.O.G.) which are the expenses for materials and labor. A more information-productive firm would be able to process more goods with less money spent on information-related activities.
Take a case where a paper firm spends $100 million for materials and labor to produce boxes. It also needs $40 million worth of sales and administrative expense to manage production, distribution and selling. An advanced information system is then introduced. The paper company still spends $100 million to make boxes of the same quality, but now requires only $30 million worth of sales and administrative expenses to operate, including the cost of the new computer system. Clearly, the productivity of information management shows an improvement in this case.
Whether U.S. industrial firms, after a decade of intensive computerization and after spending about half a trillion dollars on computers, can now operate with less information management is something that can be tested. That requires tracking how much C.O.G. is created for every dollar spent on S.G.&A.
To give the IT optimists their best opportunity to look good, I examined the financial records, over the past nine years, of the sixty-six largest and most prestigious U.S. industrial corporations, with annual revenues of more than $10 billion :
This suggests that one dollars' worth of S.G.&A., which could support $4.10 dollars worth of C.O.G., could manage only about $3.20 dollars worth of C.O.G. after a decade of enormous spending on computers. The identical amounts of information inputs managed lower outputs a decade later. Since productivity is defined as the ratio of output divided by input, the following demonstrates a productivity decline in the uses of information:
Figure 3 - Productivity of information resources for largest US industrial corporations
To place the US banks' computerization into the most favorable possible light, I examined the financial records of the sixteen largest US banks with revenues over five billion. These have invested in computerization more heavily than any other segment of the economy, as judged by the ratio of computer budgets to payroll costs. These banks are:
US banks show the costs of staff on their financial statements. These costs are mostly for labor costs of the banks' information workforce. They account for most of a bank's overhead cost.
If information technology increased the productivity of the US banks' personnel, then one would find a declining ratio of staff expense for each dollar of revenue. Such an improvement would come from an effective substitution of more productive computers for less productive administrative labor.
The results of our examination are as follows:
Figure 5 - Staff productivity for largest US banking corporations
Expecting reductions in staff costs from computerization is a reasonable expectation for realizing favorable returns on investment. The banking industry tends to spend enormous amounts of money on computers relative to its payroll costs. The following table shows the relative importance of I.T. as compared with staffing costs for two leading banking firms:
Figure 6 - I.T. Cost ratios for highly computerized banking firms
US banks show the costs of non-interest costs on their financial statements. These costs include the costs of information technology, operating expenses, outsourcing, facility costs in addition to labor costs of the banks' information workforce. The non-interest expense reflects the bank's total overhead costs.
If information technology would reduce US banks' overhead, then one would find a declining ratio of non-interest expenses for each dollar of revenue. Such an improvement would come from employing computers such as lowering the costs of facilities through automatic tellers, substituting electronic fund transfer for costs of postage and extending the geographic reach of banking services without adding fixed costs.
The results of our examination are as follows:
Figure 7 - I.T. Cost ratios for leading banking firms
Contrary to expectations of the productivity proponents, the revenue per dollar of non-interest expense has not increased. It has declined for a period of steady productivity decline from 1989 through 1993. The recent reversals in this productivity have little to do with computerization. The fact is that in 1989 one dollars' worth of non-interest expense supported $3.65 dollars worth of revenue, whereas seven years later - after a period of sustained computerization - the same dollar supports only $2.80 of revenue.
The productivity decline, as measured by the staff productivity ratio, is steeper than the drop reflected by the non-interest cost ratio. If one expects to see a rise in banking productivity that would require either getting more revenues or reducing overhead costs. Neither effects took place in banking.
There are theoreticians who claim that in an information economy firms must spend more money on coordination, management and planning, to achieve increased profitability. The objective of computerization is then to drive for increased profits and not for cutting the costs of staff, now euphemistically called "knowledge workers."
A way of testing this proposition is to examine the profitability of banking, as measured by the Return-on-Shareholder Equity (ROE). Banking is relatively homogeneous. It is more consistently computerized than any other sector of the economy. Rising ROEs would offer a good argument favoring the theory that computers are an investment that would enhance a bank's "strategic position" and thereby its opportunity to earn higher profit margins. That presumption does not hold up. Despite an exceptionally favorable economic climate that made it possible for banks to earn the highest real interest incomes in recent history, the shareholder returns do not show a rising trend:
Figure 8 - Shareholder returns for leading banking firms
To function at present levels of productivity the US information workers in major corporations consume, on the average, $18,400 of information technology for per person. The significance of this number can be misleading. One of the most frequent errors in judging I.T. spending is to compare percent of revenue or per capita spending ratios for firms in the same industry. That has little merit if one considers that there is a wide range in average compensation, per employee, for example in banking:
Nevertheless, there is no correlation between I.T. spending per employee and banking profitability, or any other measure of productivity:
Figure 10 - IT spending per employee vs.
A number of financial analysts have explained the rapidly rising profits of US corporations since 1995 by depressed wages and not by productivity gains. Accordingly, average hourly wages have declined since 1973. Even with multiple earners the average household income has dropped appreciably since 1989. According to the Business Week magazine this amounts to "a squeeze on workers." As profits rose employee compensation declined.
I distrust most of the numbers that originate from government statisticians. Therefore, I decided to check up on wage levels, as reported in audited financial statements. Unfortunately, only US banks report consistently how much they spend on their staff.
The following is a tabulation of average annual increases in staff costs per employee for the leading US banking firms:
Figure 11 - Per person average banking pay increases per year - 1987-1996
Neither client/server, the Internet nor computer networks have so far materially improved the productivity of information handling by the premier U.S. industrial corporations. In 1996 $1,109 billions in cost of goods required the support from $301 billions in information management. In 1996 $207 billions of revenue of US banks consumed $75 billions in non-interest expense and $39 billions of staff costs.
These ratios are now lower than they were in the period from 1987 through 1993. I consider these declining ratios as proof that productivity of the information handling workforce, which now accounts for 59% of US total employment, has worsened in the last decade, not improved.
The time has come to face up to the facts: the stock-market analysts and the over- optimistic CEOs are misled. It's a myth that computers have measurably increased overall US productivity of information. Whatever productivity gains may have happened to increase profits took place in the factories and the warehouses. Such realization will lead to placing IT expenses under much closer financial scrutiny to make sure that costs are not only contained, but that the computerized work creates an innovative stream of new profits.
Whatever productivity gains may have been achieved through computerization of office work in the last decade have been squandered by the profligate waste of human and technological resources. The bureaucratization and complexity of business processes have indeed increased, thereby creating the demand for more information processing to get anything accomplished. However, increased amounts of unnecessary work does not create wealth, whether done faster than before or not. If prosperity is to continue, we need to fulfill the promise of the Information Age. We must deploy IT so the workforce can consistently deliver more value with less effort.
The term "paradox" is used when reasoned acts deliver results which are counter to expectations. I think that therein lies the fallacy, and perhaps arrogance, of thinking about computers as a paradoxical phenomenon. Computers are indeed a miraculous technology. Therefore, management has been led to expect delivery of fantastic gains in productivity. The press, the vendors and the consultants have reinforced such convictions.
The facts are that:
Time has come to face up to the evidence that often computer productivity is not a mythical, intangible benefit but more often than not a productivity detractor. That means placing I.T. expenses under a close financial scrutiny to make sure that costs are not only contained, but that the computerized work creates an innovative stream of new profits.
Whatever productivity gains in the application of information may have been achieved through computerization in the last decade must have been squandered by the rising bureaucratization of the US organizations and the profligate waste of resources - human as well as technological. If prosperity is to continue, as the optimists hope, we need to take advantage of what is the promise of the information age. We must deploy information technologies to create a sustained ability of the information workforce to manage the delivery of more valuable results with less effort.
Information costs have been rising relative to other production costs, not declining. In accounting terms this means that hard-to-control and largely fixed overhead costs have been displacing variable direct costs. The impacts on planning, budgeting and controlling of information technologies are far reaching. The pressure to reduce fixed overhead costs will mount, especially during a downside swing of an economic cycle. The likely managerial countermeasures will be: increase outsourcing of information services; license software based on usage and shift computing to where capacity utilization can take advantage of economies of scale.
Within ten years the structure of the existing corporate information technology budgets will see radical changes. The enormous fixed cost of computing assets configured to meet local peak loads will be replaced by computing-on-demand from commercial utilities. Company-specific networks will be phased out in favor of carriers that will auction off bandwidth based on real-time demands for capacity. Custom-made applications requiring huge support staffs will be phased out in favor of systems configured out of standard functional modules and billed in microtransaction units. Computing will cease to be a firm-level cottage industry aspiring to self-sufficiency. Computing services will become a globally traded commodity.
In the case of industrialization, the evolution from local guild-like monopolies to globally traded markets took over three hundred years and is right now in its final phase of historical development. In the case of the information economy, the evolution from local guild-like monopolies has just begun. However, the evolutionary pace will be much swifter because telecommunications technology now makes it possible to bypass the local institutional barrier to progress by offering instant cost reductions. With little effort the customer can reap the benefits of obtaining a global marginal price instead of having to pay for all of the local fixed costs.
The acceleration in the acceptance of global trading in computerized services also enjoys unique advantages which the marketing in industrial goods could never realize. Network-based trades of information services can approach the conditions of "perfect competition" as closely as a classical economist could conceive. The variable transaction cost of any network services is extremely low, which means that price can be set by real-time bidding. The buyers can acquire perfect information about what they are purchasing, since the cost of trying out a service is trivial. The costs for each transaction are also very low because the metering and billing can be totally automated for less than a few microcents per event. No intermediaries are needed to assure global market access by buyers to sellers or vice versa, which eliminates most of the overhead expenses for brokers, wholesalers, insurers, bankers, customs officials and legions of purchasing agents who have always acted as a brake on trade. If endowed with a sufficiently intelligent electronic device, any consumer of services can connect directly to any supplier of services, anywhere, anytime.
When the new evolutionary changes in the conduct of information management will materialize, we will be able to observe and measure the long heralded rise in the productivity of the information resources. It will finally take less information inputs to deliver more and better goods and services outputs.