The decision maker is likely to fav

The decision maker is likely to favor the piograms with reliable rather than laigely speculative estimates of antic¬
ipated accomplishments per dollar. Second, if a new piogram requires a
substantial initial commitment of funds but the year-to-year increment of
fresh funds for an agency may never be that large, the tendency will be
to expand existing programs rather than to innovate. Third, additional
expenditure on a project under way (e.g., a dam or building) whose costs
are now seen as likely to exceed its benefits may be preferred to a pro¬
posed new project with a very favorable benefit-cost ratio. The decisional
question is whether die additional expenditure for completion (not the
total project costs) will be less than total project benefits, to which the
answer is usually affirmative. The already "sunk costs" reinforce the case,
since they will usually be a total loss if the project is abandoned.
The Basic Model Appraised
The basic rational-analysis model presents a number of problems.
Note first that of the four steps, only two—the identification of alterna¬
tives and the calculation and comparison of benefits and costs—call for
rational analysis. The first step, the clear, operational statement of the
goal or objective, is a/prerequisite of rational analysis. The statement must
come from someone outside the process itself (e.g., Congress, the Presi¬
dent, an agency head). The last step, decision in favor of the best
alternative, is essentially a "programmed" decision in our sketch of the
process. It follows automatically from the analysts' identification of
alternatives and indication of which has the best ratio of benefits to costs.
Only a much watered-down version of the model leaves the decision
maker free to consider other factors than those furnished by rational
analysis.
Goals. At least three problems beset the goal-statement issue: multiple
goals, redistributional goals, and sources of goal definitions. In our basic
model we postulated a single operationally stated goal or objective. Yet
many governmental undertakings have multiple goals. This complicates
development of an operational statement, for a set of goals has sufficient
explicitness only if weights are assigned to the goals. Essentially, the
analyst needs to be told by what degree each goal may be sacrificed in
order to achieve more of each other goal. Nice reckoning of such trade¬
offs is not within the capacity of most goal-setting olficials.
If that is uncongenial to policy-level officials, redistributional goals are
uncongenial to rational analysts. Yet many governmental programs seek
to redistribute benefits among regions, classes, and other social groups.
Because the rational-analysis approach was developed principally by economists, it tends to measure the effectiveness of projects and programs
by their contribution to increasmg the national income. This goal of
national efficiency, defined in economic terms, displaces or downgrades
redistributional and other noneconomic objectives. As a less parochial
economist puts it, "It is very likely that education of the hard-core
unemployed may make less of a contribution to national income than
education of professionals at universities. Yet we are willing to devote
resources to training the hard-core unemployed because of the need to
achieve political stability."®
hideed, rational economic analysis tends to a bias against the redistri¬
bution objective. In the ordinary free market, the value of a commodity
or service is what people are willing to pay for it; this expression of
consumer preference incorporates their measure of its benefit to them.
Much rational analysis in government is a search for an equivalent
pricing indicator of the benefit of public projects and programs to
citizens. The best guide, of course, is furnished when the government
sells its services at prices that cover the costs. Hence, some economic
analysts urge that the government fix an economically sound charge for
vacationers' use of campgrounds in the national parks and forests.� The
number of campers and their length of stay would then reveal the dollar
value people attach to the service. Yet such a user charge, however
convenient for economists, would subvert the redistributional purpose of
enabling less affluent citizens to enjoy our national parks and forests.
Because goal setting in the pure model is outside the analytical process
proper, the classic literature on rational decision making has litde to say
about who sets goals. Instead, a decision maker is hypothesized who is
presumed able to define an objective in the precise, operational form
needed for analysts to proceed with their tasks. Yet we know that
Congress's statements of objectives often lack clarity and consistency, that
the President's assertion of comprehensive goal-setting responsibility
would be contested by Congress, and that an agency head rarely has
such complete jurisdiction over a major policy area as to be conceded the
goal-setting role by heads of other agencies active in the area.
Some think that a nation should formulate goals for five, ten, or
twenty years ahead and restrict debate thereafter to the best means for
attaining the goals. But our elaborately pluralistic political system is designed to be constandy formulating and reformulating goals as political
majorities shift, the effectiveness of various interest groups waxes and
w�anes, population changes, and new technologies and new information
alter the shape of problems and our capacity to deal with them. The art
of assembling majorities, both electoral and legislative, includes vague
identificadon of goals and flexibility in accommodation of others" ideas
and preferred language. The system is not tidy. It violates the belief that
a ship, even a ship of state, may go in circles if it has no port of
destination. But the constitutional separation of powers, political realities,
bureaucratic dynamics, and rapidity of change in world and domestic
conditions afford little prospect of success in the search for an authori¬
tative fixer of stable, operationally stated goals.
The goal-definition problem can be transformed by proposing, as
some writers do, that the analyst formulate the goal or at least contribute
vigorously to its clarification.® The first proposed role simply revives the
old issue of whether major governmental choices of direction should be
made by "experts" or by responsible policy-level officials. The second
role—offering of bright ideas that may change a decision makers
thinking about an objective—is open to all who have access to the decision
maker; it is neither a monopoly nor a special skill of systems analysts.
Identification of Alternatives. In our outline of the basic model we have
referred ambiguously to "identification of alternative ways of achieving
the goal." The pure model of rationality requires that a// such alternatives
be identified. This requirement needs to be greatiy relaxed, both because
our wit is unequal to the task of so complete a search of the possibilities
(and, indeed, is unable to know when the search has discovered every
possibility), and because the "cost" (in time and energy, as well as money)
of a comprehensive search is too high. In real life people tend to follow
two search behaviors: (1) they simplify by screening out all silly alterna¬
tives and by restricting themselves to a few major alternatives (instead of
multiplying them by giving separate status to each minor variation); and
(2) they stop searching when they come upon a satisfactory alternative,
even though further search might turn up a better one. The latter
behavior has been termed "satisficing" by James March and Herbert
Simon.® This welcome realism does not carry us very far, for the critical question then becomes: What standards will be used to determine
satisfactoriness? If this, or the process for defining standards, is left
vague, little science remains in "the science of i-ational decision making."
Costs-and-benefits Calculation. Rational-analysis proponents may treat
goals as not their responsibility, and identification of alternatives, though
their job, as a matter of imagination and judgment in the search proc¬
ess. It is in the calculation of costs and benefits of the several alternatives
that their technical competence is most substantially brought into play.
And it is here that the approach has made the greatest contribution.
Yet there are problems of moment even in this area. Some of them
are implicit in our earlier discussion. First, given a goal or a set of closely
related goals, should only those costs and benefits be calculated that bear
on that goal or set? Or should spillover effects on other goals (including
those of other agencies' programs) be included in the reckoning? Even
if the answer supports the second procedure, how far should such
analysis go in view of the "all-togetherness of everything,"' which makes
any significant decision have almost infinite repercussions?
Second, granted that costs and benefits may not all be expressed in
dollars in the real world, can qualitative deprivations and enhancements
be converted into dollar terms or some other quantitative form that
permits benefit-cost ratios to combine the real-world qualitative and
quantitative factors? How much worth should be accorded a life, a child's
happiness, a scenic view from the highway, or a pleasing architectural
design? Somehow these and the lesser concerns do "count," perhaps
especially in America. Lewis M. Branscomb writes,
Despite the well-publicized conflict between economic and ecological inter¬
ests, our appreciation for environmental impact and technology assessment
is unique in the world. Where else would (i) a $600-million hydroelectric
dam be held up to protect an endangered 3-inch freshwater fish called the
snail darter, (ii) a unique butterfly enjoy priority at the end of the main
runway of the Los Angeles International Airport, and (iii) the sexual