Personal satisfaction is captured by the concept of utility. But it is hard to transfer this idea to a group of people. An action may improve utility of one person and, at the same time, reduce utility of another.
Let us say, we build a gondola to SFU as has been suggested a couple of years ago. A student who likes to get to the campus fast and see great scenery from above will be better off. A house owner whose house happens to be under a new gondola line and who likes to suntan naked on his roof will be worse off (there are now 50 eyes looking at his roof from above!).
How would we be ever able to reconcile the conflicting desires of many people toward same things? We do it by taking a step back and thinking about the best possible outcome. Here's a proposition, "A situation is the best possible if it cannot possibly be improved. If it may be improved, it is not the best possible." May look like a useless wordplay but it helps us to think about and identify those best situations. This is the general idea of efficiency:
A situation is efficient if it is impossible to improve that situation.
The second step in this line of thinking is to figure out what we mean by "improving" the situation. There are many possible ways to define an improvement. So there are many possible ways to think about efficiency.
We could define an improvement as a change that makes at least one person better off without making anybody else worse off. Such an improvement is called a Pareto improvement.
A Pareto efficient outcome is such that a change to any other feasible outcome is not a Pareto improvement. In other words, if it is impossible to make a Pareto improvement, the status quo is Pareto efficient.
How would Pareto efficiency perform as a criterion for government policies? Turns out, not very well. Its advantage is that a Pareto improvement is likely to gain a consensus. There will be some people who support it, and there is no reason anybody would oppose it. The two big problems are that (1) it is likely that any status quo is Pareto efficient – so there is neither need nor potential for some policy to improve the society's wellbeing, and (2) there are usually many Pareto efficient outcomes possible and we still would have to decide which one to choose (we would need another criterion to do so).
We could define an improvement as a change that makes some (at least one) people better off and it is (at least hypothetically) possible to compensate everybody else who is made worse off by the change. Such an improvement is called a Kaldor-Hicks improvement.
A Kaldor-Hicks efficient outcome is such that a change to any other feasible outcome is not a Kaldor-Hicks improvement. In other words, if it is impossible to make a Kaldor-Hicks improvement, the status quo is Kaldor-Hicks efficient.
There are several theoretical (the criterion lacks some very desirable formal properties) and practical (it is very hard to evaluate hypothetical compensations) difficulties with Kaldor-Hicks efficiency. It is widely used in welfare economics and managerial economics (when people attempt the cost-benefit analysis). Yet for our class, the technical side of this criterion is a bit too advanced, so we will just shrug our shoulders and ignore this criterion
We could define an improvement as a change that makes some (at least one) people better off (call them the Winners) and the combined gains in value of the Winners are larger than combined losses of the Losers (those who are made worse off by the change). Such an improvement is called a Marshall improvement.
A Marshall efficient outcome is such that a change to any other feasible outcome is not a Marshall improvement. In other words, if it is impossible to make a Marshall improvement, the status quo is Marshall efficient.
Unlike Kaldor-Hicks, Marshall efficiency refers explicitly to the values rather than utilities. It also does not require that it is possible (even hypothetically) to compensate the losers. As long as combined gains from a change are greater than combined losses, the change is a Marshall improvement. Since we may be able to observe revealed values, Marshall efficiency is more feasible as a practical guide to policies. I suspect that most economists use this criterion when they think about efficiency, whether they admit it or not. Notice that the criterion implies that a Marshall efficient outcome maximizes the aggregate value in a society.
Chapter 3 of the course reader
Economic Efficiency (by David Friedman)
Competitive Market Efficiency (1 of 2)
Competitive Market Efficiency (2 of 2)