Ah_sa：
问： 何谓均衡，均衡必定有效率吗？
veryhard：
这个问题也太基本了，不过就是不好回答。在此小弟谈一下自己的看法。所谓的均衡实际上是一种稳定的状态，也就是在长期的动态的调整中达到的一种结果，而该结果在没有外部条件改变的情况下会长期保持。用博弈论的观点讲就是一种博弈结果的稳定常态。而均衡并不意味着有效率，例如你能讲囚徒的纳什均衡不是一种均衡吗?不过在我看来瓦尔拉斯的一般均衡好像是有效率的！！
----------------------------------------------
潇潇洒洒过一生，人生何求?淡泊明智，宁静致远.
wbreeze：
均衡分为马歇尔的局部均衡和瓦尔拉斯的一般均衡.均衡与效率是俩马事。达到了均衡不以定事有效率的。资源配置达到有效阶段时，我们一般是用帕累托均衡来衡量，如果达到了帕累托最优，（起码时次优），我们可以说时达到了效率。
Ah_sa：
Many people think that equilibrium has to be efficient.
This is their "proof", what's wrong with it?
Suppose X is an efficient allocation without taken into account any frictions.
If we assume agents are utility maximizing, agents should have achieved X. If not, there must be some obstacle (eg. information cost, contracting cost, property right problem...) avioding them to achieve it. In this case, agents will evaluate the cost C of removing this obstacle. There are 2 cases:
(i) If this cost C is lower than the value of X, agents will pay this C and achieve X, thus efficiency is achieved in equilibrium.
(ii) If this cost C is higher than the value of X, agents will not pay this C and X is not achieved. But because the net value of achieving X is negative, not attaining X is actually an efficient choice.
Therefore, in equilibrium, it is always efficient.
An example:
Economists always say that a monopoly charging P>MC is not efficient.
However, removing a monopoly involves a cost C.
If C is higher than the value of removing monopoly, then in equilibrium there is monopoly, but charging P>MC is efficient.
If C is lower than the value of removing monopoly, then in equilibrium there is no monopoly, and the allocation is efficient.
Comment.
jack&jones：
"Total costs is smaller than Total benefits" is the necessary condition for an reallocation to be efficient.
"Individual costs is smaller than Individual benefits "provides the incentive for making improvement.
In an equilibrium where Pareto Improvement is possible for the whole society, it may not be implemented by any of the individual agent if for any agent, it takes the relatively high costs to get an average benefit.
The two costs in 1 and 2 are not the same. This is where the rub is: the logic of collective action.
Ah_sa：
The standard answer to your question is as follows:
Now, we can explicitly model the cost of enforcing collective action.
Consider the Prisoner Dilemma game below.
_____________not confess____confess
not confess_____1, 1 ________-1,2
confess_________2,-1________ 0,0
Suppose the prisoners can write binding contract to enforce (not confess, not confess) with cost $C.
If C<1, then equilibrium is (not confess,not confess). Payoff is (1-C,1-C)>(0,0) and is Pareto optimal.
If C>1, then equilibrium is (confess,confess). Payoff is (0,0)>(1-C,1-C) and is still Pareto optimal. [Note that in standard PD game, C is infinite and belongs to this case]
民自荣：
cooperative game???
but with the constraint of information,how can they contact each other?so you cannot say every equilibrium is Pareto efficient,beacsuse the standard PD is also an equilibrium.
Ah_sa：
Well, some economists (not me) will probably say:
"In the standard PD game where the prisoners cannot communicate, it''s just the case of C=infinite. In this case, the equilibrium is (confess,confess). Payoff is (0,0)>(1-C,1-C) and is still Pareto optimal."
Is this arguement correct and standard textbook wrong ?
Or is this arguement wrong?
Or...maybe both are correct !
To resolve this puzzle, try to think about the definition of "Pareto optimality".
民自荣：
OK,could you answer me:is Nash equilibrium necessarily walrasian equilibrium?is Nash equilibrium necessarily Pareto optimal?
I cannot agree with them.If only there exists a person who can act as a medium,then the cost of conract is not infinite.
Ah_sa：
Firstly I have to clarify that the above arguement that "Equilibrium is always Pareto Efficient" is what some economists (I call them "Chicago school") believe but NOT what I think. I put it here to raise a discussion.
以下是引用民自荣在2003-1-18 12:47:00的发言：
isNashequilibriumnecessarilywalrasianequilibrium?
NO.
Is Nash equilibrium necessarily Pareto optimal?
"Chicago school": Yes, because equilibium is always Pareto optimal.
Textbook: No.
Who is correct? Both!
Because they are using different definition of Pareto Optimality!
Definition of PO: an allocation such that it is not feasible to make someone better of without making others worse off.
The problem is: what is the set of "feasible" allocation?
Consider the above PD game with C>1.
Both "Chicago School" and Textbook think that (confess, confess) is the equilibrium.
But textbook would say that the (not confess, not confess) is "feasible" and better than (confess, confess). So equilibrium is not PO.
"Chicago School" would say that (not confess, not confess) is only ALLOCATIVELY "feasible" but not INCENTIVELY (because C>1). So equilibrium is PO.
Ok, what do you think?
民自荣：
福利经济学第一定理说,在偏好局部非满足性的条件下Walrasian 均衡是Pareto
最优.这应该是经济学的公认的基本知识.
其次,对于非Walrasian均衡的Nash均衡,比如囚徒困境,是否一定是Pareto最优呢?这上我们讨论的焦点.若这仅仅是不同学派对feasible的理解问题,那么我们的讨论就完全变成
了字面上的辩论,没有实际的意义了.
Ah_sa：
以下是引用民自荣在2003-1-21 12:24:00的发言：
福利经济学第一定理说,在偏好局部非满足性的条件下Walrasian 均衡是Pareto
最优.这应该是经济学的公认的基本知识.
其次,对于非Walrasian均衡的Nash均衡,比如囚徒困境,是否一定是Pareto最优呢?这上我们讨论的焦点.若这仅仅是不同学派对feasible的理解问题,那么我们的讨论就完全变成
了字面上的辩论,没有实际的意义了.
Yes and No
(1) Yes:
- the debate relies very much on the definition of "feasibility constraint"
(2) No:
- The traditional approach only says that equilibrium can be pareto inefficient if allocation is subject only to allocative feasibility.
- "Chicago School" adds the idea that equilibrium must be pareto efficient if allocation is subject to incentive fesibility. It is more than "字面上的辩论".
wangxq2000：
有意思，我也来说几句。
“芝加哥学派”的观点有个前提：人是完全理性的。人的每一个选择都是经过了理性完全的计算，并且有能力“完全”地计算，自然所有均衡都是最优的了。
leeleel：
以下是引用wangxq2000在2003-10-29 20:09:00的发言：
有意思，我也来说几句。
“芝加哥学派”的观点有个前提：人是完全理性的。人的每一个选择都是经过了理性完全的计算，并且有能力“完全”地计算，自然所有均衡都是最优的了。
昏鸦：
呵呵，不错，按照均衡的定义，均衡本身就是有效率的，因为均衡强调这样一种状态：即任何人都没有动力去改变这种状态，除非外生变量发生了变化，既然没有改变的动力，那么毫无疑问最优并且有效率
qiulin：
我考试的时候一碰到一般均衡的题目，凡是可以选做的我一般不选，实在是觉得难。但一点基本常识还是知道的。福利经济学第一定理中间瓦尔拉斯均衡必然帕雷托有效，除了偏好局部非厌足的假设外，起码的一点是价格接受者和价格一致假设，所以外部性、垄断、信息不对称等问题都可以导致第一定理失效。至于说均衡本身就是有效率，这纯粹是瞎掰，不光“纳什均衡”，协调困难等问题，大家平常经常用局部均衡算的那个有关税扭曲，有垄断的情况下的那个价格和数量，都是均衡，不见得有效率。
luopu：
在某些条件下面，存在着瓦尔拉斯一般均衡，可以证明它一定是帕雷托最优的，
也就是帕雷托有效的。我不知道这个效率是不是你说的效率，也不知道这个
一般均衡是不是你所指的均衡。
证明挺复杂的，我也不会……你想要知道可以去看高级微观。
顺便说一下，这个好像不是哈耶克证明的。
luguihua：
均衡是所有的决策者没有动机改变其决策或行为，如消费者没有动机改变消费数量，生产者没有动机改变供应数量，此时处于一种稳定的状态。
均衡是否有效，不仅决定于均衡的性质，同时决定于效率的标准，即什么样的状态是有效配置状态。一般而言，经济学中最常采用的效率标准是pareto效率标准。
Arrow-Debreau证明了，福利经济学第一定理：完全竞争的均衡是Pareto最优的。福利经济学第二定理，对于任何一个Pareto最优配置，当合理安排所有经济主体的初始财富配置，可以通过完全竞争来达到。这两个定理的含义是：第一定理说完全竞争均衡，从pareto的意义上来廛，具有最高的效率；第二定理是说，任何一个pareto，可以通过完全竞争不定期达到，只是需要所有经济主体的初始财富的配置满足特定条件而已。
加尔布雷思：
均衡与效率是两码事
均衡是指一种稳定的状态，但并不一定是有效的，象囚徒困境。
效率是只指资源配制最优，但并不一定是均衡状态，否则一旦经济体达到最优就不要经济学家了。
buguliao：
其实回到纳什均衡是不是Pareto有效的这一问题更有意义，如果不是，为什么？是的条件又是什么？张维迎有一篇文章只是提了一下这个问题，有必要看看Holmstrom（1982）的那一篇原文以及相关文章。
exwuxun：
一点愚见！
本人认为，有效是相对于整个社会而言的，而不是一个群体，比如奥尔森在《集体行动的逻辑》一书中的论述。囚徒困境对参与人是非有效的，但对于整个社会却是有效率的。
quin001：
均衡和效率是个两难冲突。
如果经济长期处于“理想” 的均衡之中，那么，经济就不能实现增长，因为经济增长的本质就是非均衡的。但是，经济的稳定增长（宏观经济政策的真正目标）似乎也是一种均衡，在这个意义上说，二者又是不冲突的。
zhouechang：
我只想知道垄断中的均衡也是有效率的吗?
参阅：http://bbs.efnchina.com/dispbbs.asp?BoardID=92507&id=29171
