股权融资成本GLS模型计算

news2024/11/25 2:33:32

一、模型公式

式中：

r 为股权融资成本

P为股价

B为每股净资产

FROE为预测每股净资产收益率

目标：求解股权融资成本r

二、模型口径参考来源

PS：实际以代码为准

①FROE（预测每股净资产收益率）:

资本市场开放与股权资本成本 2020 金融研究

②股利支付率k（综合考虑股利政策不稳定与公司规模变化、公司成长机会，采用过去3年股利支付率平均值作为k）：

媒体曝光度_信息披露环境与权益资本成本 2014 金融研究

机构投资者、所有权性质与权益资本成本 2018 金融研究

终极所有权结构对权益资本成本的影响 2016 管理科学学报

贸易摩擦与权益资本成本 2021会计研究

IFRS、财务分析师、机构投资者和权益资本成本 2016会计研究

③Pt（股价）、Bt（每股净资产）：二者同为t时点

资本市场开放和股权资本成本 2020 金融研究

中美交叉上市与权益资本成本研究 2013 国际金融研究

真实盈余管理与权益资本成本 2015 金融研究

④Matlab求解：

三、代码片段

3.1导入库及读取相关文件

需要的文件 1.kFlieName：股利支付率国泰安 -> 股利政策 -> 基本数据 -> 财务指标数据注意：

1.包含当年的过去5年 2.必须先选每股股利后选每股收益！

2.FROEFlieName：预测净资产收益率锐思 -> 研究报告 -> 个股盈利预测
注意：1.截止日期12-31 2.需要字段：股票代码、预测年度、净资产收益率平均 3.统计时间间隔为180时数据才与wind基本一致！！！！！

3.indROEFlieName：行业ROE 锐思 -> 研究报告 -> 行业财务指标注意：1.截止日期12-31 2.行业划分标准：9申万 3.需要字段：行业代码_IndCd、行业净资产收益率(%)_IndROE

4.closeFlieName：12月收盘价平均值锐思 -> 股票 -> 日股票综合数据注意：1.日期范围：12.1-12.31 2.需要字段：股票代码、日期、收盘价

5.bNewFlieName：每股净资产锐思 -> 股票 > 年股票数据注意：1.日期：12.31 2.需要字段：股票代码、每股净资产

6.newStkCode.csv:目标股票代码

7.RESSET_INDCLS_1.csv：公司所属行业

import pandas as pd
import numpy as np
from numpy import *
import sympy as sp


#2008
kFlieName = "2008年数据\k2003_2008.csv"
FROEFlieName = "2008年数据\RESSET_RRSTKEAFOCA_2008_1.csv"
indROEFlieName = "2008年数据\INDFI_EARNPOWER.csv"
closeFlieName = "2008年数据\RESSET_DRESSTK_2001_2010_1.csv"
bNewFlieName = "2008年数据\RESSET_YRESSTK_1.csv"

stkCode =  pd.read_csv("newStkCode.csv",header=None,names=["stkCode"])  #目标股票代码

indClass = pd.read_csv("RESSET_INDCLS_1.csv",index_col="股票代码_StkCd")[["行业代码_IndCd"]]
indClass = pd.read_csv("RESSET_INDCLS_1_new.csv",index_col="股票代码_StkCd")[["行业代码_IndCd"]]
# for index,each in enumerate(indClass.index):
#     indClass.index[index] = 0
newK =  pd.read_csv(kFlieName,index_col="Stkcd").dropna()
FROE =  pd.read_csv(FROEFlieName,index_col="股票代码_StkCd")[["预测年度_ForYear","净资产收益率平均(%)_ROEAvg"]]
FROE.columns = ["预测年度","FROE"]
# close =  pd.read_csv("RESSET_MRESSTK_1.csv",index_col="股票代码_Stkcd")#.iloc[:,:-1]

# indROE =  pd.read_csv(indROEFlieName,index_col="行业代码_IndCd")["行业净资产收益率(%)_IndROE"]
indROE =  pd.read_csv(indROEFlieName,index_col="IndustryCode")["ROEB"]

close = pd.read_csv(closeFlieName,index_col="股票代码_Stkcd")
bNew = pd.read_csv(bNewFlieName,index_col="股票代码_Stkcd")

# indClass
# indROE

stkCodeList = set(stkCode["stkCode"])
newKIndex = set(newK.index)
FROEIndex = set(FROE.index)
indClassIndex = set(indClass.index)
indROEIndex = set(indROE.index)
closeNewIndex = set(closeNew.index)
bNewIndex = set(bNew.index)

selectIndex = stkCodeList & newKIndex & FROEIndex  & bNewIndex & indClassIndex & closeNewIndex

print(len(selectIndex))

3.2股利支付率

kMean5 = []

#股利支付率k = 每股股利 / 每股收益
for eachIndex in set(newK.index):
    newK.loc[eachIndex,"k"] = newK.loc[eachIndex,"T20497"]/newK.loc[eachIndex,"T60200"] #股利支付率 = 每股股利 / 每股收益

for eachIndex in sorted(list(selectIndex)):
    test = newK.loc[eachIndex].iloc[:5]["k"].sum()/5                #过去5年均值
    kMean5.append(test)

panel = pd.DataFrame({"k":kMean5},index=sorted(list(selectIndex)))
print(panel)

3.3FROE

FROE = FROE.loc[selectIndex]
FROE = FROE.sort_index()

for eachIndex in sorted(set(selectIndex)):
    test = FROE.loc[eachIndex]
    if len(test) != 3:
        FROE.drop(index=[eachIndex],inplace=True,axis=0)
        
FROEIndex = set(FROE.index)
selectIndex = stkCodeList & newKIndex & FROEIndex  & bNewIndex & indClassIndex & closeNewIndex
print(len(selectIndex))
print(len(panel))

for eachIndex in list(selectIndex):
#     print(close.loc[eachIndex]["b"])
    panel.loc[eachIndex,"t+1"] = FROE.loc[eachIndex].sort_values(by="预测年度")["FROE"].iloc[1]*0.01
    panel.loc[eachIndex,"t+2"] = FROE.loc[eachIndex].sort_values(by="预测年度")["FROE"].iloc[2]*0.01
#     close.loc[eachIndex,"t+3"] = FROE.loc[eachIndex]["FROE"].iloc[2]*0.01
    panel.loc[eachIndex,"t+3"] = (panel.loc[eachIndex,"t+1"] + panel.loc[eachIndex,"t+2"])/2
    panel.loc[eachIndex,"行业代码"] = indClass.loc[eachIndex][0]
print(len(panel))

for eachIndex in selectIndex:
#     print(int(close.loc[eachIndex]["行业代码"]))
    if int(panel.loc[eachIndex]["行业代码"]) in indROE.index:
        panel.loc[eachIndex,"IndROE"] = indROE.loc[int(panel.loc[eachIndex]["行业代码"])]*0.01 
print(len(panel))

for eachIndex in selectIndex:
    begin = panel.loc[eachIndex,"t+3"]
    end = panel.loc[eachIndex,"IndROE"]
    dif = abs((begin-end)/9)#9
    bFront = panel.loc[eachIndex,"b"] #b0
    for i in list(range(4,13)):#13
        begin = begin-dif
        panel.loc[eachIndex,"t+"+str(i)] = begin
    for i in list(range(1,13)):
        bFront = bFront + panel.loc[eachIndex,"t+"+str(i)] * bFront * (1-panel.loc[eachIndex,"k"])
        panel.loc[eachIndex,"b"+str(i)] = bFront
print(len(panel))

3.4收盘价

close = close.loc[selectIndex]
close = close.sort_index()

for eachIndex in list(selectIndex):
    each = close.loc[eachIndex,"收盘价_Clpr"]
    panel.loc[eachIndex,"close"] = each.mean()
print(len(panel))

3.5每股净资产

for eachIndex in list(selectIndex):
    each = bNew.loc[eachIndex,"每股净资产(元/股)_NAPS"]
    panel.loc[eachIndex,"b"] = each
print(len(panel))

3.6保存各年面板数据

newData = pd.read_csv("result.csv",header=None)[0].dropna() + 0.5
print(len(newData))
newData = newData[newData!=1.5]
print(len(newData))
# print(newData.sort_values()[-50:])
print(newData.sort_values()[7:-7].describe())

各年面板数据如下图：

3.7Matlab求解方程（Matlab）

clear all;clc;
data=csvread('C:\Users\zsllsz2022\Desktop\10月实证\GLS模型\core\2-原始口径合并版本-重新计算2008-2020\panel2008.csv',1,0);
P=data(:,3);
B(:,1)=data(:,4);
B(:,2:13)=data(:,19:30);
FROE(:,1:3)=data(:,5:7);
FROE(:,4:12)=data(:,10:18);


%% [-0.49995,0.49995]每隔0.0001计算函数值
for stock = 1:length(data)
    for j = 1 : 10000
        r(j) = 0.0001 * j; 
        yy(stock,j) = B(stock,1);
        for i = 1 : 11
            yy(stock,j) = yy(stock,j) + (FROE(stock,i)-r(j))*B(stock,i)/((1+r(j))^i);
        end
        yy(stock,j) = yy(stock,j) + (FROE(stock,12)-r(j))*B(stock,12)/(r(j)*(1+r(j))^12);
        yy(stock,j) = yy(stock,j) - P(stock);
    end
end

%% 寻找零点
for stock = 1:length(data)
    count = 1;
    for j = 1 : 9999
        if yy(stock,j+1) * yy(stock,j) < 0
            resultTemp(stock, count) = (j - 5000) * 0.0001;
            count = count + 1;
        end
    end
end

%%
for stock = 1:length(data)
    temp = 1;
    for j = 1 : size(resultTemp,2)
        if abs(resultTemp(stock, j)) < abs(temp) && abs(resultTemp(stock, j)) > 0.000000001
            temp = resultTemp(stock, j);
        end
    end
    result(stock,1) = temp;
end


%%
%对结果排序
xSort=sort(result);

%剔除头尾百分之一的数据
delNum=ceil(length(xSort) * 0.01);
for i=1:length(xSort)-2*delNum
    x2(i,1)=xSort(i+delNum);
end

%求平均值以及标准差
avg=mean(x2);
stdev=std(x2);

% function y=myfun(r,P,B,FROE)
%     F=B(1);
%     for i=1:11
%         F=F+(FROE(i)-r)*B(i)/((1+r)^i);
%     end
%     F=F+(FROE(12)-r)*B(12)/(r*(1+r)^12);
%     y=P-F;
% end