[求助]关于回归方程的估计方法

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

(1) (2) (3) (4) (5) (6) INTERCEPT -1.4625*** -1.9294*** -1.6670*** -1.5238*** -2.2189*** -1.5603*** (255.8372) (185.2898) (298.2622) (263.9823) (209.0065) (255.9571) MIDCONCEN 0.5295*** 0.5611*** (15.8160) (15.2407) HICONCEN 1.0289*** 1.1427*** (31.2782) (34.5082) PREFINANCE 0.4075*** 0.3393*** 0.4693** (11.8439) (7.0013) (4.5049) PREFINANCE*LOWCONCEN -0.3224 (1.3632) PREFINANCE*MIDCONCEN -0.0763 (0.1137) FCF 0.2037 0.1086 0.2941 0.3063 0.2819 -0.0510 (0.2638) (0.0737) (0.5031) (0.5858) (0.4469) (0.0372) ROA 0.8320*** 0.7732*** 1.0011*** 0.8235*** 0.9295*** 3.1580*** (13.2685) (11.1145) (18.0699) (12.9215) (14.9799) (8.2609) DIVPt-1 0.8293*** 0.8101*** 0.7975*** 0.7889*** 0.7419*** 1.1370*** (67.3945) (63.2022) (56.0763) (59.9157) (46.8519) (142.7375) Chi-square 82.1767*** 115.0872*** 296.1234*** 93.7303*** 337.9886*** 156.1420*** 这是我在一篇文章中看到的结果,一个方程怎么能出现这么多的估计结果呢?

用什么方法进行估计的呢? 我们常见的估计方法有那些? 什么软件做的呢请大家指教? 我将文章传了上来，请指教

19169.rar (248.42 KB) 本附件包括：

• 3-2-3.pdf

2005-7-10 18:13:00 上传

[求助]关于回归方程的估计方法

[此贴子已经被作者于2005-7-10 20:57:11编辑过]

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏1 回帖

关键词：回归方程 Refinance Intercept Finance Financ 方程

Hello, there are six different variable to be estimated.

You should learn to look its answer vertically, for example, the first model inclusides three variables>

FCF, ROA,DIVPt-1

Its intercept and Beta coefficient you should know, i guess the data in bracket is T value or standard diviation, it should be explained in original text. when signed with star, means significant in 1%,5% or 10% confidence level

Second model adds two more variable, midconcen, hiconcen.

The others are are same.

Estimation method> ordinary least square is the most usful.

For your reference

如果是这样的话 就简单了 那么这样逐步加入变量的目的是什么？有什么优点？直接全部加入不是一样吗？

，我怀疑他使用了不同的估计方法， 不同的方法有不同的结果。是这样吗？

难道是我理解的复杂了？

你应该把文章附上来，否则只看结果没有什么用处。可能文章里面已经讲了，你没有看到。
你的结果中有两个方程一摸一样，但结果不同。所以可能是：（1）样本变化；（2）不是单方程，可能是2个或几个估计
（3）工具变量的法，用不同的工具变量。
只有看了原文才可能了解。

大家帮着看看阿，谢谢了

the goal is to compare the end results of different models, and to draw a conclusion, which variable is critical, which variable can just do a marginal improvement, for another words, this variable is not so important.

by the way, you should check the R square, if the new model is improved, or not.

为什么　最后用的　Chi-square　呢？我们一般的估计用的Ｆ值，这儿怎么会是　Chi-square？

我猜想是这样的：

方法应该是一样的，但是从（1）到（6）分别加入了不同的变量

从而看出加入的变量对原来变量的系数的影响

My explaination:

chi square =n*obs

it test the significance of the coefficient of different variables. for example: r1=0,r2=0,till r6=0, since here chi-square is so big, absolute exceeding the critical value, that means, the probability of all values jointly as zero is very low.

With another words, the author want to say, all these variables (x) have significant impact on the dependent variable(y), but his model is not so ideal, all R-square is very low. He should try other estimation methods, i think.

For your reference