Arnold Kling, "Arguing in My Spare Time", No. 15

"Oh, there will still be programmers. Most will work at software vendors, creating components. A few still will be around your IS shop, keeping legacy systems limping along until they can be replaced."

--Frank Hayes, Computerworld, May 25, 1998

This quote is a better fit for some of my earlier essays. It succinctly describes the changes that I foresee in corporate computing, as traditional proprietary systems are replaced by applications assembled from generic components.

When I think of the Dilbert sector, I think of large, bureaucratic companies. Is there a chance that the change in software development methodology will lead to companies that are smaller or less bureaucratic? This raises issues about economies of scale and the efficiency of alternative decision-making processes. This in turn requires us to define carefully the concepts of scale, scope, and bureaucracy.

The pure issue of scale concerns fixed cost vs. variable cost. When fixed costs are high relative to variable costs (a classic example would be a nuclear power plant), then the advantage accrues to a large firm.

The effect of technology on economies of scale can be ambiguous. New technology can lower fixed costs, and thus reduce economies of scale. On the other hand, technology could reduce variable costs, thereby enhancing economies of scale.

For example, suppose that electronic banking were to eliminate the need for people to visit tellers in branch offices. This would reduce fixed costs associated with building branches. However, it also would reduce the marginal cost of handling additional transactions and additional customers. Whether the combination of these cost reductions favors large or small banks is unclear.

If the value of proprietary systems as strategic assets has fallen, this also has an ambigous effect on scale. It may make more sense now for two large Dilbert-sector companies to merge, because a merger eliminates one of their proprietary systems. If proprietary systems still had value, then it probably would be more economical for the firms to remain separate.

On the other hand, if proprietary systems no longer represent valuable assets, then smaller firms may enter where previously they could not. They may be able to compete using generic software. The handful of "pure" Internet banks that have come on line might be indicative of this possibility.

The strong prediction is that the number of proprietary systems will decrease. The open issue is whether this will occur because new (small?) companies drive old companies out of business, or whether it occurs because large companies merge, or both. Obviously, the range of outcomes for industry structure is quite wide.

A second issue, which often gets confused with scale, is scope. Increasing the scale of the firm means increasing the number of customers for the same product. Increasing the scope of the firm means increasing the variety of products or services that it sells. The recent merger between Citicorp and Travelers increases scope by combining bank products with insurance products. If the companies succeed in "cross-selling" their customers, the merger also could increase scale.

If it costs less to produce jointly multiple types of goods and services than to produce them separately, then there are economies of scope. If not, then there are diseconomies of scope. If doing one thing well teaches you how to do other things well, that is an economy of scope. If doing one thing well is more difficult when you are distracted by trying to do something else, then that is a diseconomy of scope.

People who believe that "one-stop-shopping" is the holy grail for financial services firms implicitly believe that there are economies of scope in marketing and/or providing financial services to individuals. Others among us are skeptical of this hypothesis.

The guessing here is that new technology does not promote economies of scope from the consumer's perspective. An Internet user who can quickly obtain price quotes from several online stores would probably see very little value in being able to obtain price quotes on different products from a single store. "One-stop shopping" seems relatively meaningless as a value proposition when consumers can do all the shopping that they require without leaving their desks.

From the producer's perspective, economies of scope might consist of using knowledge and experience gained in designing and selling one product to venture into a related area. It is difficult to assess how technology might increase or decrease the ability of firms to exploit such economies.

Related to the issue of scope is the issue of bureaucracy. What I mean by bureaucracy is the use of a lot of resources in the process of making a decision. In a previous essay, I cited John Kenneth Galbraith, who suggested that bureacracies arise to solve problems in which decisions involve major long-term commmitments of capital. Proprietary computers systems used to fit this model. Applications assembled from generic components should require less bureaucracy.

Firms that attempt to achieve broad scope also may require bureauracy. Just as a firm with large capital investments must make decisions across time periods carefully, a firm with many related products tries to make decisions across divisions carefully. Indeed, the perceived need for bureaucratic co-ordination may represent an important diseconomy of scope.

Concerning management organization, it is interesting to note that in theory, a large firm can operate as if it were a collection of small firms. In that sense, the difference between a large firm and a set of small firms is that the large firm has an additional degree of freedom. The large firm can choose to operate exactly as if it were nothing but a set of small firms, but it also can try to take advantage of scale or scope economies that might arise by combining different components.

Given this additional degree of freedom, one might expect large firms to be the dominant form of industry structure, regardless of the relative costs and benefits of bureaucracy. The large firm could choose to discard its bureaucracy and operate as a set of small firms if necessary.

Bureaucracy works by increasing the number of "eyes" that look at a decision before it is implemented. This tends to reduce mistakes. Operating without bureaucracy involves more trial and error.

Some Guesses

We have reached the point in this essay where conclusions ought to be drawn. Unfortunately, I do not have powerful enough evidence for my conclusions to present them as anything more than guesses. Here are the guesses:

1. Economies of scale are on an upward trend, because of reduced costs of managing operations and dealing with customers from far away. This is a result of better information management and communication. Hence, Walmart is displacing local drugstores. More generally, "category-killers" are emerging, because what they can achieve in one location they can achieve everywhere.

2. Economies of scope are on a downtrend, or at least not trending upward as rapidly as economies of scale. The evidence for this is anecdotal, at best. For example, compare Microsoft today with IBM 25 years ago. IBM was in hardware, software, and service. Microsoft only is in software. Their efforts to find other profit opportunities, such as in the area of content, have not been particularly successful.

3. The costs of bureaucracy are increasing relative to its benefits. The thinking here is that a large planning apparatus for systems is dysfunctional in the current environment.

Having said that, the companies that appear to be most threatened by technological change are the small companies. Studies indicate that small firms have the slowest rate of adoption of new technology. (My guess is that if one thinks in terms of a statistical distribution, there is a segment of small firms that is faster than large firms at adapting to new technology, but that segment does not represent the highest frequency point in the distribution.) The bureaucracies may not be the winners at adopting technology, but they are not the worst losers.

It may be that if the object of the game is to achieve large scale, then bureaucracies are helpful. In an earlier essay, I argued that in competition no one plays perfectly, and that over a long period of time the player who makes the most correct moves will win. If a large bureaucracy can make good decisions in an environment with economies of scale, then the cost of the bureaucracy in terms of overhead and sluggishness will be offset by the benefits of more accurate management.

Consider two evolutionary processes that could lead to a winner at a particular business.

a) natural selection. Many small firms enter the market and make decisions, and one of them has the skill and luck to make the fewest mistakes, becoming the dominant firm.

b) bureaucratic filtering. A single firm with a large bureaucracy faces many of the same choice points, and it uses its resource-intensive planning processes to sort out the decisions. These processes minimize mistakes, enabling the firm to reach the same point that would be reached in the natural selection process.

My guess is that process (a) will increase in importance in the future, and that process (b) will be less productive. The challenge with defending this guess is the fact that large companies with bureaucratic management are so successful at present.

The rationalization I would offer is that the ability to suppress wages has enabled large firms to survive. The result of this process should be that it will become easier for smaller firms to compate for labor. If so, then smaller firms will start to draw away highly-skilled workers. This will leave large firms with fewer skilled workers and a relatively larger share of bureaucrats. That might lead to a sort of vicious cycle in which smaller firms become ever more fast-moving and larger firms become ever more bureacratic. Eventually, that would have an adverse impact on the profits of the larger firms.

That is one scenario, and it is only a guess.