**Introduction**

The post provides a very basic introduction to the idea of "second best." The term "second best" originated in a famous 1956 article by Lipsey and Lancaster (see bibliography), and it was originally used as a technical economic concept. Despite its technical origins, the idea behind the second best is very general: sometimes the ideal solution to a problem (or "optimal policy option") is infeasible. The best should not be the enemy of the good; so, when the first-best policy option is unavailable, then normative legal theorists should consider second-best solutions. In this post, we will take a hard look at the idea of the second best, beginning with a statement of the intuitive idea and then looking at the more formal idea of the second best in its original economic context.

As always, the *Lexicon* is aimed at law students, especially first-year law students, with an interest in legal theory.

**The Intuitive Idea**

The intuition behind the idea of the second best is simple. We would
like to have the best possible legal system. But sometimes the best
legal policies are not in the cards; that is, the best policy may be
impractical. *Why?* In legal theory, one common reason that we
cannot adopt the best policy is politics. Given the political forces
that operate, the best campaign finance system may be pie in the sky.
So we ask the question, of those systems that might be politically
feasible, which is the "second best"?

Although I've introduced the intuitive idea by talking about
"political feasibility," the idea of the second best is more general
than that. First-best solutions may be unavailable because of a variety
of constraints, of which politics is only one. The intuitive idea of
the second best is a bit broader and less technical than the way
economists define "second best," so let's turn to the technical
economic idea of second best now.

**The Second Best in Economics**

The very general idea of the economic theory of the second best can be expressed as follows:

Assume a system with multiple variables. Take the most desirable state the whole system could assume and the associated values that all of the variables must assume to produce this state: call this condition, the first-best state of the system and call the associated values of the variables, the first-best values. Now assume that one variable will not (or cannot) assume the value necessary for the first-best state of the whole system: call this the constrained variable. Holding the constrained variable constant, consider the most desirable state the whole system could then assume and the associated values that all the nonconstrained variables must assume to produce this state: call this the second-best state of the system. There are systems in which achieving the second-best state will require that at least one variable other than the constrained variable must assume a value other than the first-best value: call these value(s) the second-best value(s).

And here is the way that Lipsey and Lancaster formulated the idea:

[I]f there is introduced into a general equilibrium system a constraint which prevents the attainment of one of the Paretian conditions, the other Paretian conditions, though still attainable, are in general, not desirable.

(If "Paretian" is unfamiliar to you, there is an explanation in Legal Theory Lexicon 060: Efficiency, Pareto, and Kaldor-Hicks.) Lipsey and Lancaster are making a normative (but technical) argument. They assert that if one variable is constrained and cannot assume its first-best value, then "in general" other variables should not assume their first-best values. The "in general" qualification is important. Lipsey and Steiner didn't and couldn't show that it is always (or necessarily) the case that constraint of one variable affects the most desirable value for other variables. Rather, their proof shows that this is possible. In the real world, whether nonconstrained variable should depart from their first-best variables depends entirely on the facts. In fact, if a policymaker lacks certain information about the second-best variables, it may turn out that the real world policy that will produce the best result is to try to move the constrained variable as close as possible to its optimal state, leaving the second-best variables in their first-best states. The possibility was called the "third best" by Ng (see bibliography below).

One or two additional points are necessary to complete the technical story.

*First*, the definition that I just
gave assumes that only one variable is constrained. But there is no
reason to limit the theory of the second best in this way, more than
one variable may be constrained. In fact, in theory every variable
could be constrained: in this limiting case, the second-best state
would be the only possible state of the system.

*Second*, the second best is usually
understood as relative to a constrained variable. We could use the
phrase "second best" to refer to the second-best state the system could
assume if all the variables were unconstrained, but this is not the way
that Lipsey and Lancaster used that phrase.

*Third*, there is an important
difference between the way economists understand "second best" and the
way the same phrase is understood by noneconomists. What was
interesting and powerful about Lipsey and Lancaster's proof is that it
produced the counterintuitive result that sometimes when one variable
is constrained, the best policy choice will involve moving other
variables away from their first-best values.

Although technically, the definition of second best need not be limited to that special situation, that is the *interesting*
result, and the use of the theory of the second best in economics may
be limited to the special case. Outside of economics, however, the
phrase "second best" tends to be used in a much looser sense. The
important thing is not the terminology, but the ideas. To be clear,
however, it is useful to explain what you mean by second best!

**The Second Best and Nonideal Theory**

The idea of the second best that is used by economists is analogous
to a distinction made famous by the political philosopher, John Rawls.
Rawls distinguished between two ways of approaching political
philosophy, *ideal* and *nonideal* theory. *In ideal theory*,
we assume compliance with the normative requirements of our theory.
Rawls used the phrase "well-ordered society" to refer to the situation
that obtains in ideal theory. In a society that is well ordered by
Rawls's principles of justice, citizens actually would be guaranteed a
fully adequate scheme of basic liberties and the basic structure would
actually work to the advantage of the least well off group in society. *In nonideal theory*, we relax the assumption that the society is well ordered by the principles of justice. *Can you make that very abstract description more concrete?*
Yes, here is a really good example. In a society that is well-ordered
by Rawls's principles of justice, we might assume that if there are
local governmental units, they will comply with the restraints imposed
by the freedom of speech. But in the real world, local governments
might be more susceptible to political pressure to suppress unpopular
speech than would be the central government (i.e. the national
government in Washington, D.C., in the case of the United States). So,
in the real world of nonideal theory, we might be very considered with
constraining the jurisdiction and powers of local governments; whereas,
this issue may not even arise in the case of ideal theory.

**Pinpointing the Constrained Variable**

The notion of the second best and the related idea of nonideal
theory get tossed around quite a lot in legal theory, but sometimes
these terms are used carelessly or without precision. Whenever you hear
or read the term "second best," ask yourself the question, "Which
variable is constrained, and why is it constrained?" Because the
"second best" is *second best* relative to a constrained
variable, use of the concept of the second best doesn't mean anything
unless and until the constrained variable is specified. Moreover, it is
sometimes very important to know *why* the constrained variable
is constrained. This is because it is easy to construct an argument for
a second-best policy option that uses a double standard with respect to
whether variables should be considered to be constrained. Here is a
simple example:

Suppose our problem is racial justice with respect to the distribution of income and resources. Someone might make the case for reparations (a one time payment of a compensatory amount to descendents of the former slaves) on the ground that reparations are the second-best solution. The first-best solution would be a just economic order in which market mechanisms operate in a nondiscriminatory fashion to allocate income and wealth according to just criteria. (For this purpose, we don't need to specify what the just criteria are.) But the first-best solution is unavailable, because a just economic order is politically infeasible. Therefore, we ought to support reparations, which is the second-best policy.

So far, so good. But notice that there is a hidden assumption in
this argument. The argument assumes that reparations are politically
feasible. If this assumption is incorrect (which it may well be as an
empirical matter), then it follows that the argument for reparations as
the preferred second-best solution is fallacious. Of course, one can
deploy double standards with respect to which variables are constrained
(or which options are infeasible) so long as the double standard is
made clear. But when the double standard is concealed and the argument
is made in the context of policy evaluation, then we have either an
innocent mistake or an attempt at manipulation.

**The Feasible Choice Set**

Another way of approaching the general problem revealed by the theory of the second best is *via*
the notion of the feasible choice set. Take all of the possible legal
policy options with respect to a particular legal problem. Then lay out
a set of well-defined criteria for feasibility. Apply the criteria to
the set, sorting the options into the feasible choice set and the
infeasible choice set. Practical policy discussion will usually be
limited to the options within the feasible choice set, but legal theory
is not limited to the practical. Frequently we can learn something
important by considering options that are outside the feasible choice
set. For example, a rule of strict liability might turn out to be the
optimal rule of tort law. It could also turn out that strict liability
regimes are politically infeasible--perhaps because the fault-based
social norms are very strongly held. But that fact should not preclude
legal theorists from examining the merits of strict liability regimes.
Not only may such an examination be of intrinsic interest, but the
insights gleaned from such an examination may well assist in the
evaluation of the options that are within the feasible choice set.

**The Bottom Line**

The notion of the second best, the distinction between ideal and
nonideal theory, and the idea of the feasible choice set, are all
essential tools for a legal theorist. As a first year student, you are
likely to encounter these ideas in classroom discussion or in law
review articles assigned as ancillary reading. The trick to mastering
these concepts and using them effectively is to identify the
constrained variable (or the nonideal conditions). Once you've done
that, you can move to the next step, which is the question, "What
criteria are used to identify the constrained variables?" And if you
can answer that question, you are now in a position to respond in an
intelligent and sophisticated way to applications of the theory of the
second best!

**Bibliography & Links**

- R.G. Lipsey & Kelvin Lancaster, The General Theory of the Second Best, 24 REV. ECON. STUD. 11 (1956).
- Yew-Kwang, Ng, Welfare Economics (London: Macmillan, 1983).
- Karla Hoff, Second and Third Best Theories (pdf file).
- Steven Suranovic, The Theory of the Second Best
- Thomas McCarthy, Political Philosophy and Racial Injustice: From Normative to Critical Theory (explores Rawls's use of ideal and nonideal theory).
- Lawrence B. Solum, Constitutional Possibilities (2007).

**Related Lexicon Entries**

(This entry was last updated on January 20, 2008.)