【劳动经济学】The Returns to Education in China

davidbeckham23 发表于 2012-7-3 09:08
还没有看这篇文章，但是20%的每年return是不是太高了？

我也有这样的感觉，有点儿离谱。不过当年Heckman估计的更高，没记错的话，40%每年，我的神啊。

学习学习，多谢分享

学习了

没进行不同政策背景下的因果分析？比如较早开放的东部沿海与保守的中、西部的差异。

The instrument is questionable because the implementation can be related to / affect some unobserved variables such as economics development. Those unobserved variables are not exogenous to return on education.  BTW: it's weird that they have 5 authors for such a paper

conly007 发表于 2012-7-13 07:19
The instrument is questionable because the implementation can be related to / affect some unobserved ...

The regional unobserved characteristics, such as economic development is partial out when province dummies and year dummies are included. BWT, the compulsory law, in principle, is not correlated with individual unobservables which might be the major omitted variables in a wage equation (or "returns to education" model).

rudi 发表于 2012-7-4 17:34
我也有这样的感觉，有点儿离谱。不过当年Heckman估计的更高，没记错的话，40%每年，我的神啊。

可能是CHNS的农村样本数量太大，而教育回报在这些人群中可能比我们通常认为的要高。当然，这是我个人的猜测，不知道是否有其他原因。

夸克之一 发表于 2012-7-13 07:49
The regional unobserved characteristics, such as economic development is partial out when province ...

The compulsory law is arguably exogenous because any law is still decision. However, we can buy that because otherwise we have nothing. However, the main variation in the paper is the different implementation period in different region this is definitely related to regional structure and output which is related to individual unobserbables such as their social relations, their virtues, etc. For example, in a region where people think education is totally useless, that will affect the implementation of the law, the attitude in taking the education (number of years) and the output.  Regional dummies help however it's not clear how much it helps because it only captures something constant.  Anyway, I am not saying this instrument is useless at all. It is questionable.

conly007 发表于 2012-7-13 10:15
The compulsory law is arguably exogenous because any law is still decision. However, we can buy th ...

I think you might misunderstand the main purpose of this IV. It's not the regional differential of effective date of compulsory law.

What they do is define a cohort as treatment group if an individual is 13-14 in a particular wave. For control group, they define all individuals that are 15-16 in the survey year. The "true" variation, thus,  is generated by the difference between treated group and control group given their location and survey year.  

夸克之一 发表于 2012-7-13 07:54
可能是CHNS的农村样本数量太大，而教育回报在这些人群中可能比我们通常认为的要高。当然，这是我个人的猜 ...

1. CHNS中的农村样本量的确很大，不过报告了个人工资收入的比例实际上非常低，只有10-20%。

2. 不是很确定什么样的农村居民获得工资收入，干部、教师或者在乡镇企业打工的工人？

3. 我之前做过一些简单的试验：暂时不考虑内生性等问题，分别对城市(urban)和农村(village)的工资收入做标准的Mincerian 回归。结果很有意思，在农村，在1997/2000之前，RTS都非常之低，与0无差异；即使在2000年以后，RTS快速上升，农村的RTS仍然远低于城市的RTS以及10%这个大家习以为常的标准。

根据读美国文献的感觉，似乎考虑了内生性问题的对RTS的估计，如IV，有时会高于OLS，但是比较这篇文章的结果和我之前做的简单试验的结果，我觉得高出的是在是太多了，所以我个人认为不是很可信。