• <tr id="yyy80"></tr>
  • <sup id="yyy80"></sup>
  • <tfoot id="yyy80"><noscript id="yyy80"></noscript></tfoot>
  • 99热精品在线国产_美女午夜性视频免费_国产精品国产高清国产av_av欧美777_自拍偷自拍亚洲精品老妇_亚洲熟女精品中文字幕_www日本黄色视频网_国产精品野战在线观看 ?

    Fraud Identification of Chinese Listed Companies—an Improvement Based on M-Score

    2020-09-23 05:16:16LUWanting陸晚亭ZHAOXiaokang趙曉康

    LU Wanting(陸晚亭), ZHAO Xiaokang(趙曉康)

    Glorious Sun School of Business and Management,Donghua University,Shanghai 200051,China

    Abstract: To evaluate the applicability of the M-score model in the Chinese capital market,this research observed 190 financial fraud samples punished by the China Securities Regulatory Commission (CSRC) in the years from 2014 to 2018. The test results indicate that two types of errors are high,which means that the applicability of the M-score is unacceptable. Therefore,in this paper,a 9-index model is constructed by Wald’s backward stepwise regression method,and the optimal threshold is set by the Beneish expected cost method (ECM). The accuracy of the modified M-score is significantly improved,especially the Type I error rate of is reduced from 70.37% to 19.75%. The receiver operating characteristic (ROC) curve test also proves the superior identification effect of the modified M-score applied in the Chinese market. Finally,variables such as current ratio,fixed asset index,and equity concentration in the modified model could represent the fraud characteristics of Chinese listed companies.

    Key words: financial fraud; M-score model; threshold setting; receiver operating characteristic (ROC)curve

    Introduction

    The recent scandal of Kangmei (stock code: 600518) falsely increasing profits has brought the issue of financial fraud of listed companies to the forefront once again. The frequent occurrence of financial frauds is not only due to the profit-seeking nature of the capital market itself,but also due to the lack of effective and reliable detection tools,which helps investors avoid risks and plays a warning role in the fraud.

    The financial fraud identification system has experienced the evolution process from a single variable to multivariable,from static to dynamic. In 1932,Fitzpatrick first used the single variable analysis method to study the companies under financial pressure,which pioneered the theory of financial warning[1]. Most of the subsequent financial fraud identification studies used multivariate analysis in Refs.[2-4]. Their research proved the identification function of the company’s overall financial situation. After that,Beneish[5]proposed the idea of using financial statement data to identify fraud through a dynamic index analysis method,which has been widely used in subsequent research[6-8]. On this basis,Chinese scholars took listed companies punished by China Securities Regulatory Commission (CSRC) for financial fraud as samples to test the impact of various characteristic factors on financial fraud,such as Song and Tan[9],Kong[10],Qian and Luo[11]. However,the characteristic variables used in the existing financial fraud identification research are different. The tested samples occur in different periods of time. The methods to evaluate the accuracy of the model are not detailed enough. Moreover,few studies discuss how to set the threshold of fraud identification. Investors lack a simple and practical fraud identification model.

    The M-score model developed by Beneish[12]in 1999 is widely used in the western market and has gained a reputation for successfully predictingthe Enron case. The model calculates the M-value by extracting the financial variables of the company,thereby scoring the company’s financial rationality,and then judging the possibility of the company’s manipulation of profits. The variables of M-score are selected according to the characteristics of financial fraud in American listed companies. However,China and the United States are quite different in terms of market environment and accounting system.

    This papers elects the M-score for research and modify the model based on the data of Chinese A-share listed companies to establish a financial fraud identification model that is more suitable for the Chinese market. The modified model can effectively help investors identify fraudulent companies and avoid financial risks. At the same time,this study also theoretically complements the parts of threshold setting and model evaluation that have been lacking in previous related studies.

    1 M-Score Model

    M-score can identify financial fraud through thresholds. Beneish took the fraudulent companies announced by the U.S. Securities and Exchange Commission (SEC) between 1982 and 1992 as observation samples and estimated an 8-index model based on the financial statement data:

    M=-4.840+0.920R1+0.528R2+0.404R3+0.892R4+

    0.115R5-0.172R6-0.327R7+4.697R8,

    (1)

    whereMis the M-value calculated by Eq. (1). The definition and description of the variables are as follows.

    R1: days’ sales in receivables index (DSRI). DSRI measures whether there is any abnormal change in receivables and operating revenue in two consecutive years. The higher the DSRI is,the more likely financial fraud is.

    R2: gross margin index (GMI). The higher the GMI is,the worse the profitability is and the higher the possibility of financial fraud is.

    R3: asset quality index (AQI). The higher the AQI is,the lower the asset quality,the higher the risk of asset realization and the higher the possibility of fraud.

    R4: sales growth index (SGI). Companies with high growth will have higher capital requirements and therefore have greater incentives for financial fraud.

    R5: depreciation index (DEPI). The higher the DEPI is,the lower the depreciation rate is and the more likely the company is to manipulate the profit by prolonging the service life of the assets.

    R6: sales,general,and administrative expenses index (SGAI). High SGAI indicates an unreasonable increase in sales,general,and administrative expenses. In order to reverse this negative signal,companies are more likely to commit fraud.

    R7: leverage index (LVGI). High LVGI represents a high financial risk. The higher the risk of financial fraud is,the higher the probability of financial fraud is.

    R8: total accruals to total assets (TATA). TATA is used to evaluate the degree of management earnings. Generally speaking,high TATA indicates a high probability of profit manipulation.

    The formulas used to calculate the M-score’s variables are as follows:

    wheretrefers to the year of fraud;Rrefers to receivables;Srefers sales;Grefers to gross profit margin;Carefers to current assets;Trefers to total assets;Drefers to depreciation;Prefers to property,plant,and equipment;Serefers to sales,general and administrative expense;Tlrefers to total long term debt;Clrefers to current liabilities;Irefers to income before extraordinary item;Orefers to operating cash flow. IfM>-1.78,it means that the possibility of financial fraud is high.

    The variables selected by M-score are all in the form of a ratio,which takes account of the articulation between accounting elements,and prevents companies from “cooking books” against a single indicator[13]. According to Beneish’s test,M-score’s accuracy in identifying financial fraud is 76%.

    2 Model Testing and Selection of New Variables

    2.1 Sample test of Chinese listed companies

    In this paper,financial fraud companies announced by the CSRC during the years from 2014 to 2018 are taken as fraud samples,and the non-event years of all companies are taken as non-fraud samples. The sample information comes from the “penalties for major violations” section in the RESSET database. A total of 190 fraud samples and 9 693 non-fraud samples were obtained. This paper randomly select 40% of the total samples as the testing samples for M-score accuracy test. The results are shown in Table 1.

    Table 1 Test results of M-score

    An effective fraud identification model should minimize two kinds of error rates at the same time. Among them,Type I error refers to misjudgment of fraud samples as non-fraud samples,and Type II error refers to misjudgment of non-fraud samples as fraud samples. Although M-score has a high recognition accuracy for normal samples (84.84%),it has a poor recognition effect for fraud samples (29.63%),with Type I error rate as high as 70.37%,which represents irreparable investment losses[14].

    For the fraud samples failed in the M-score test,the fraud methods are summarized in Table 2. In addition to characterizing the fraud characteristics of Chinese listed companies,the fraud means summarized below also point out the blanks that the M-score model cannot capture,which can be used as the basis for index addition.

    2.2 Selection of new variables

    According to the construction form of the M-score model and the fraud methods summarized above,the ratio formal variables are selected as follows.

    2.2.1Financialvariables

    R9: current asset turnover (CAT) is the ratio of main business income to average current assets. CAT is used to evaluate the efficiency of assets utilization. The faster the turnover speed is,the higher the utilization efficiency of current assets is,and the stronger the operation capacity is[15]. In terms of index composition,CAT is composed of operating income and current assets which are most related to the fraudulent means of sample companies.

    R10: inventory turnover rate (ITR) is the ratio of main business cost to average inventory. ITR is mainly used to measure the sales ability and inventory management level of a company. The higher the ratio,the stronger the liquidity of the inventory,the higher the inventory operation efficiency of the purchase,production,sales,and the weaker the motivation of enterprises to commit financial fraud[16]. ITR is composed of inventory and cost of sales,which are major related projects of common fraud means of Chinese listed companies.

    Table 2 Fraud methods of M-score test failed samples

    R11: other receivables index (ORI) is the ratio of other receivables to total assets. Other receivables are used to account for the economic transactions between companies and other units. The contents are various,and there is no clear standard for this account. Therefore,it is often used for financial fraud. Generally speaking,the ORI is high,which means that companies are likely to use the account to hide profits,cover up expenses,intercept earnings,and transfer funds[17]. As shown in Table 2,a large part of fraud involved current assets is related to other receivables.

    R12: bad debt provision index (BDPI) is the ratio of bad debt provision to total profit. For the company with a large number of receivables,a slight change in bad debt provision will lead to a huge difference in net profit. By less provision for bad debts,the company can instantly “pull up” the company’s profit; by more provision for bad debts,the newly appointed managers can deliberately “do bad” the performance of the previous manager,and then create the illusion that profits increased year by year through less or even no provision for bad debts. As shown in Table 2,many sample companies committed fraud by illegal bad debt provision for account receivable and other receivable.

    R13: cash rate of sales (CRS) is the ratio of operating cash flow to main business income. CRS is used to measure the ability to obtain cash. The larger the ratio is,the higher the quality of the sale is,the better the effect of capital utilization is. The low cash rate of sales indicates that the sales quality and cash ratio are both low,and the possibility of false income is high[11]. CRS is composed of main business income and operating cash flow,both of which are related to the most commonly used operating income frauds by Chinese listed companies.

    R14: fixed asset index (FAI) is the ratio of fixed assets to total assets. As a non-current asset account,fixed assets are often used for financial fraud in a concealed way. If the early inflated fixed assets are not discovered,the value of this part can be gradually resolved by depreciation and impairment in the later stage,which is not easy to cause the sudden disconnection between profit and cash flow and won’t cause investors to suspect. By deliberately underestimating the investment cost of fixed assets,enterprises can reduce the cost or expense and artificially increase profits[18]. As shown in Table 2,fixed asset fraud is a common type of asset fraud.

    R15: current ratio (CR) is the ratio of current assets to current liabilities. CR is used to measure the solvency of a company. The lower CR indicates the worse short-term liquidity of the company and the higher possibility of fraud. The index is composed of current assets and current liabilities,which are the most common types of asset / liability fraud in Chinese listed companies.

    2.2.2Non-financialvariables

    R16: equity concentration (EC) is the sum of squares of the shareholding ratio of the top 5 shareholders. Moderate concentration of equity can create a convergence effect among shareholders,which is beneficial to corporate governance and performance improvement. However,highly concentrated equity will make the majority shareholders with absolute control rights infringe on the interests of other shareholders. As shown in Table 2,in some sample companies,there are illegal guarantees and illegal occupation of funds by major shareholders. However,another view is that the higher the equity concentration,the more convenient it is for the majority shareholders to control and manage the company,thus reducing the opportunistic behavior of the management and the occurrence of financial fraud[11].

    R17: independent directors ratio (IDR) is the ratio of the number of independent directors to the number of board members. Independent directors need to make objective and professional judgments on the company’s affairs. To a certain extent,independent directors can stand on the position of small and medium shareholders and play a role in supervision and warning for the management. The higher the proportion of independent directors is,the lower the possibility of financial fraud is[19].

    R18: audit opinion (AO) takes specific values according to different results. Take 1 for unqualified opinions; take 2 for unqualified opinions with explanatory notes; take 3 for qualified opinions; take 4 for rejected or unable opinions; take 5 for negative opinions; take 6 for unaudited opinions; take 7 for reserved explanatory notes.

    The type of AO can reflect the authenticity and accuracy of the audited company’s finances,while the western fraud identification model does not include this factor. According to Huang,the higher the AO type,the worse the authenticity and accuracy of the company’s financial status,so the higher the possibility of fraud[20]. Almost all of the 57 fraud samples counted in Table 2 were issued with non-standard unqualified opinions.

    The added variables themselves or the variable composition are closely related to the accounting items involved in fraud means,and can represent the financial fraud characteristics of Chinese listed companies.

    3 Logistic Regression Model Construction

    3.1 Model modification

    Whether or not fraud is a binary classification result. The logistic regression function is constructed as

    (2)

    whereYis binary variable of financial fraud. When the financial fraud occurs,Ytakes 1,otherwise,Ytakes 0;α0is constant;R1,R2,...,R8are the original model variables;R9,R10,...,R18are the added variables;αiis the pre-index coefficient,i=1,2,...,18.

    In Eq. (2),the 18 variables are modeled using the remaining 60% random sample (modeling samples) to avoid endogenous problems. Because of the large sample size and the unknown correlation between indicators,the logistic stepwise regression method based on Wald was used to build the model.

    Logistic regression results show that after ten stepwise backward regressions,nine variables finally enter into the model:R2(GMI),R4(SGI),R5(DEPI),R7(LVGI),R8(TATA),R14(FAI),R15(CR),R16(EC) andR18(AO). The significant negative correlation ofR8indicator is contrary to the results of the original model research,but consistent with the research results of Qian and Luo[11],which indicates that the proportion of accrued items of fraudulent companies in China is lower than that of normal companies. And due to the deterioration of short-term payment ability of fraudulent companies,the increment of business payables is higher than that of receivables. The negative correlation ofR7is consistent with the research conclusion of Qiao and He[21]. The fraudulent company disguised financing ability and solvency by falsely increasing assets and reducing liabilities. At this time,the asset-liability ratio drops,and leverage index decreased.

    The formula of the modified M-score model can be obtained in Table 3. The modified M-score can be calculated a probability value by substituting relevant indicators to judge the possibility of financial fraud in a company.

    M′=-1.645+0.009R2+0.051R4+0.061R5-

    0.586R7-2.049R8-0.240R14-0.148R15-

    2.001R16+0.405R18,

    (3)

    whereM′ is the modified M-score value calculated by Eq. (3). The nine variables of the modified M-score model can explain the financial fraud from three perspectives.

    (1) Motivation perspective. The deterioration of gross profit margin (GMI rises) indicates that the company’s profit has decreased and its profitability has been weak. The decline of TATA and current ratio is a signal of poor short-term liquidity. At this time,to maintain sales growth and continue to obtain capital,the management tends to manipulate profits. The loose shareholding structure also makes management unable to be effectively restrained,so there is more implementation space for fraud.

    Table 3 Logistic regression results

    (2) Behavior perspective. In order to alleviate the performance pressure,the manager often adopts the method of reducing the depreciation rate (DEPI rises) to extend the service life of the assets,“dilute” the depreciation expenses of each period,and reduce the deduction of profits; or adopt the method of reducing liabilities and increasing assets to create the illusion of good asset conditions and strong financing solvency; or use more concealed methods to falsely increase fixed assets in the early stage,and then gradually depreciate this part of the value through depreciation and impairment.

    (3) Consequence perspective. The type of audit opinion can reflect the authenticity and accuracy of the company’s finances. A listed company that has been issued a non-standard unreserved audit opinion has a noteworthy financial position.

    Table 4 Three perspectives of explanatory variables

    Therefore,when the gross profit margin of listed companies deteriorates,the short-term solvency becomes worse,the management under theperformance pressure is prone to financial fraud. The loose shareholding structure makes room for the growth of fraudulent motives. To whitewash the report,the management usually adopts methods such as reducing liabilities,increasing fixed assets,and lowering depreciation rate. Companies with financial problems are more likely to be issued non-standard unqualified audit opinions,which is also a signal of financial warning.

    3.2 Threshold setting

    Any threshold can make two types of errors and produce error costs. In the case of the Type I error,investors can’t identify the fraudulent companies,which results in the loss of investment;In the case of the Type II error,investors mistakenly judge the normal operation companies as fraudulent companies,which limits the investment opportunities. There is no perfect threshold in theory,but the better the value is,the more accurate the model classification is,and the lower the error cost is. Three threshold determination methods are used to find the optimal threshold.

    3.2.1Beneishexpectedcostmethod(ECM)

    ECM is calculated according to two types of errors and the expected cost caused by misjudgment. The formula is as follows:

    CECM=PMP1C1+(1-PM)P2C2,

    (4)

    whereCis the expected cost;PMis the ratio of fraud sample to total sample;P1andC1are the Type I error and its misjudgment cost;P2andC2are the Type II error and its misjudgment cost. The threshold is optimal when ECM takes the minimum value. Beneish’s assumption isC1/C2=N,C2=1,C1=N=40[12].

    The modeling samples are used here. It is calculated thatPM= 0.018 38 andCECM= 0.735 4P1+ 0.981 6P2. When the Type I error rate is 12.84% and the Type II error rate is 9.81%,the ECM value is the smallest (0.190 76),and the threshold value is -2.90.

    3.2.2Dechowmethod

    Dechow threshold method can be directly calculated by[22]:

    D=ln[PM/(1-PM)]=-3.98,

    (5)

    whereDis the threshold calculated by Dechow method.

    3.2.3Youdenindexmethod

    Youden index is a method to evaluate and screen the authenticity of a test. The larger the Youden value is,the better the screening effect is[23].

    Yd=Ssen+Sspe-1,

    (6)

    whereYdis the threshold calculated by Youden index method; sen is the sensitivity,which is the correct detection ratio of fraud samples;Sis the value of authenticity; spe is the specificity,which is the correct detection ratio of non-fraud samples. By calculating the Youden value of 5 929 groups through Visual Basic for Applications (VBA),it is found that when the threshold value is 2.07,the Youden value reaches the maximum (0.982 1).

    In order to select the optimal threshold,testing samples were used to test the three thresholds respectively. From the comparison results in Table 5,Type II error of the Youden index threshold (2.07) is zero,but the Type I error rate is as high as 97.25%,which violates the original intention of the fraud identification tool to help investors avoid risks. The Type I error of the Dechow threshold (-3.98) is the lowest,but the Type II error rate reaches 29.81%,which represents an overly cautious investment method and misses many possible returns. Taken together,the two types of error of the ECM threshold (-2.90) are both low,indicating that while helping investors identify fraudulent companies,and it will not miss too many investment opportunities.

    Table 5 Comparison of threshold detection results

    4 Results and Discussion

    For the modified model constructed in Eq. (3) and the new threshold value of -2.90,the testing samples are used for effect test and the results are shown in Table 6. The comparison of test results in Table 7 shows that the detection effect of the modified M-score model has been significantly improved compared to the original model. The Type I error rate is reduced to 19.75% and the Type II error rate is only 9.81%,which proves that the modified M-score model is more suitable for the Chinese market.

    Table 6 Test results of modified M-score

    Table 7 Comparison of test results between M-score

    The Receiver Operating Characterist (ROC) curve is used to further verify the overall prediction ability of the modified model. Figure 1 is a comparison of the ROC curves of the M-score and the modified M-score for the full sample. The horizontal and vertical axes represent the specificity and sensitivity of the model indicators respectively[24]. The area below the ROC line measures the discrimination ability of the model. The area between 0.7 and 0.8 indicates that the model’s prediction ability is average. In the range of 0.8 to 0.9,the predictive ability of the model is good. The modified M-score model has an area of 0.920,and its prediction ability is much higher than that of the M-score.

    Fig. 1 ROC curves of M-score and modified M-score

    5 Conclusions

    The modified M-score includesR2(GMI),R4(SGI),R5(DEPI),R7(LVGI),R8(TATA),R14(FAI),R15(CR),R16(EC) andR18(AO). This 9-index model indicates that when the gross profit margin of a listed company with a loose equity structure gradually deteriorates,the asset-liability ratio,current ratio,fixed assets ratio,and depreciation rate all show a downward trend,and short-term payments are in dilemma. At this time,the executives under sales pressure tend to take financial fraud. And fraudulent companies are likely to be issued with non-standard unqualified opinions.

    Beneish ECM sets an optimal threshold to minimize the cost of classification errors. Compared with M-score,the overall discrimination accuracy of the modified model is improved to 89.98%. Meanwhile,the area of 0.920 under ROC curve also proves that the modified model has significant discrimination and prediction effect on financial fraud. In general,the modified M-score model characterizes the fraud characteristics of Chinese listed companies better,and is more suitable for the Chinese market.

    精品国产乱码久久久久久小说| 免费日韩欧美在线观看| 精品久久久久久电影网| 亚洲内射少妇av| 日本av手机在线免费观看| videossex国产| 老司机影院成人| 亚洲精品一区蜜桃| 色94色欧美一区二区| 街头女战士在线观看网站| 亚洲一级一片aⅴ在线观看| 在线观看免费高清a一片| 亚洲精品中文字幕在线视频| 18+在线观看网站| 最近最新中文字幕大全免费视频 | 男女午夜视频在线观看 | 热re99久久国产66热| 蜜桃国产av成人99| 久久精品国产亚洲av涩爱| freevideosex欧美| 午夜久久久在线观看| 一区二区三区乱码不卡18| 精品99又大又爽又粗少妇毛片| 伦精品一区二区三区| 亚洲成色77777| 少妇被粗大猛烈的视频| av在线播放精品| 99热这里只有是精品在线观看| 久久国产精品男人的天堂亚洲 | 久久国内精品自在自线图片| 国产免费又黄又爽又色| 免费观看在线日韩| 国产精品欧美亚洲77777| 我要看黄色一级片免费的| 久久久久久久国产电影| 亚洲成人一二三区av| 美女国产高潮福利片在线看| www.av在线官网国产| 90打野战视频偷拍视频| 精品一区二区三卡| 男人舔女人的私密视频| 啦啦啦啦在线视频资源| 啦啦啦啦在线视频资源| 高清在线视频一区二区三区| 国产精品麻豆人妻色哟哟久久| 九九爱精品视频在线观看| 精品亚洲成a人片在线观看| 免费少妇av软件| 亚洲欧美成人精品一区二区| 精品酒店卫生间| 久久国产精品大桥未久av| freevideosex欧美| 美女国产视频在线观看| 国产免费一区二区三区四区乱码| 99热网站在线观看| 国产精品欧美亚洲77777| 高清欧美精品videossex| 狂野欧美激情性bbbbbb| tube8黄色片| 在线精品无人区一区二区三| 22中文网久久字幕| 哪个播放器可以免费观看大片| 丝袜人妻中文字幕| av又黄又爽大尺度在线免费看| 夫妻午夜视频| 丝袜脚勾引网站| 亚洲国产精品国产精品| 曰老女人黄片| 18禁动态无遮挡网站| 精品99又大又爽又粗少妇毛片| 内地一区二区视频在线| 精品亚洲成国产av| 一区二区日韩欧美中文字幕 | 考比视频在线观看| 巨乳人妻的诱惑在线观看| 天天影视国产精品| 捣出白浆h1v1| 女人久久www免费人成看片| 少妇高潮的动态图| 99视频精品全部免费 在线| 久久久久精品人妻al黑| 国产成人一区二区在线| 嫩草影院入口| 国产男女内射视频| 成人国语在线视频| 久久99热这里只频精品6学生| 亚洲精品国产av成人精品| 国产精品嫩草影院av在线观看| 黄网站色视频无遮挡免费观看| 一二三四在线观看免费中文在 | 亚洲av电影在线进入| 国产色爽女视频免费观看| 侵犯人妻中文字幕一二三四区| 国产精品嫩草影院av在线观看| 日日爽夜夜爽网站| 国产在视频线精品| 日本猛色少妇xxxxx猛交久久| 国产精品三级大全| 久久影院123| 天堂8中文在线网| 国产成人av激情在线播放| 亚洲国产欧美在线一区| 九色成人免费人妻av| 亚洲av电影在线观看一区二区三区| 亚洲精品av麻豆狂野| 亚洲国产av影院在线观看| 成人亚洲精品一区在线观看| 精品国产露脸久久av麻豆| 黄色配什么色好看| 最近中文字幕2019免费版| 伦理电影大哥的女人| 国产成人午夜福利电影在线观看| 99re6热这里在线精品视频| 国产在线视频一区二区| 尾随美女入室| 国产高清不卡午夜福利| 夫妻性生交免费视频一级片| 久久久久久久久久久久大奶| 久久精品国产自在天天线| 有码 亚洲区| 久久影院123| 国产精品国产三级国产专区5o| 成人综合一区亚洲| 国产高清不卡午夜福利| 最新中文字幕久久久久| 国产精品女同一区二区软件| 下体分泌物呈黄色| 乱码一卡2卡4卡精品| 亚洲国产日韩一区二区| 亚洲av中文av极速乱| 亚洲综合色网址| 咕卡用的链子| 狠狠精品人妻久久久久久综合| 尾随美女入室| 欧美日韩成人在线一区二区| 国产 一区精品| 欧美人与性动交α欧美精品济南到 | 视频中文字幕在线观看| 老司机影院成人| 91精品国产国语对白视频| 亚洲欧美中文字幕日韩二区| 伊人久久国产一区二区| 久久久国产精品麻豆| 91成人精品电影| 国产精品久久久久久av不卡| 国产欧美另类精品又又久久亚洲欧美| 午夜福利在线观看免费完整高清在| 亚洲精品自拍成人| 九九在线视频观看精品| 99香蕉大伊视频| av免费观看日本| 国产精品人妻久久久影院| 欧美国产精品一级二级三级| 黑人欧美特级aaaaaa片| 国产1区2区3区精品| 日韩电影二区| 亚洲高清免费不卡视频| 中文字幕人妻熟女乱码| 国产有黄有色有爽视频| 日韩熟女老妇一区二区性免费视频| 久久久欧美国产精品| 亚洲欧美成人精品一区二区| 性色av一级| 国内精品宾馆在线| 麻豆精品久久久久久蜜桃| 国产成人aa在线观看| 免费黄网站久久成人精品| 欧美精品亚洲一区二区| www日本在线高清视频| 久久久久久久精品精品| 色婷婷久久久亚洲欧美| 亚洲图色成人| 男女午夜视频在线观看 | 久久久亚洲精品成人影院| av有码第一页| 久久久精品免费免费高清| 国产成人精品一,二区| 免费在线观看完整版高清| 精品熟女少妇av免费看| 2022亚洲国产成人精品| 色婷婷av一区二区三区视频| 2018国产大陆天天弄谢| 免费观看性生交大片5| 热99国产精品久久久久久7| 少妇 在线观看| 久久久久久久国产电影| 欧美激情 高清一区二区三区| 欧美xxⅹ黑人| 亚洲色图 男人天堂 中文字幕 | 热99国产精品久久久久久7| 亚洲图色成人| 在线精品无人区一区二区三| 国产精品.久久久| 美女主播在线视频| 91aial.com中文字幕在线观看| 22中文网久久字幕| 欧美精品一区二区大全| 成年美女黄网站色视频大全免费| 亚洲精品美女久久久久99蜜臀 | 人体艺术视频欧美日本| 最新的欧美精品一区二区| 一级,二级,三级黄色视频| 欧美日韩精品成人综合77777| 91久久精品国产一区二区三区| 九色亚洲精品在线播放| 男人添女人高潮全过程视频| 国产免费一级a男人的天堂| 亚洲av福利一区| 亚洲精品国产av成人精品| 一个人免费看片子| 国产av码专区亚洲av| 日本vs欧美在线观看视频| 丝袜美足系列| av黄色大香蕉| 99视频精品全部免费 在线| 国产精品人妻久久久影院| 在现免费观看毛片| 蜜桃在线观看..| 日本wwww免费看| 最近2019中文字幕mv第一页| 99国产综合亚洲精品| 国产视频首页在线观看| 男女午夜视频在线观看 | 九九爱精品视频在线观看| 天天躁夜夜躁狠狠久久av| 91精品伊人久久大香线蕉| 成年女人在线观看亚洲视频| 亚洲国产精品专区欧美| 黄色 视频免费看| 纵有疾风起免费观看全集完整版| 99久久综合免费| 亚洲欧美成人精品一区二区| 在线观看三级黄色| av在线app专区| 欧美日韩成人在线一区二区| 国产精品熟女久久久久浪| 日本黄色日本黄色录像| 丰满乱子伦码专区| 欧美精品高潮呻吟av久久| 观看av在线不卡| 精品一区二区三区视频在线| 全区人妻精品视频| av在线播放精品| 成人午夜精彩视频在线观看| 中国三级夫妇交换| 蜜桃国产av成人99| 亚洲激情五月婷婷啪啪| 老司机亚洲免费影院| 国产一区二区三区综合在线观看 | 成年人免费黄色播放视频| 国产精品国产三级国产av玫瑰| 在线观看www视频免费| 九九在线视频观看精品| 亚洲成人手机| 亚洲精品美女久久av网站| 国产男女内射视频| 亚洲欧美色中文字幕在线| 亚洲熟女精品中文字幕| 国产精品一二三区在线看| 你懂的网址亚洲精品在线观看| 亚洲色图综合在线观看| 宅男免费午夜| 亚洲欧洲日产国产| 国产精品久久久av美女十八| 国产精品秋霞免费鲁丝片| 久久精品久久久久久久性| 午夜福利网站1000一区二区三区| 欧美 亚洲 国产 日韩一| 欧美激情极品国产一区二区三区 | 国产成人精品在线电影| 久久综合国产亚洲精品| 精品国产一区二区三区四区第35| 色吧在线观看| 18禁观看日本| 哪个播放器可以免费观看大片| 97精品久久久久久久久久精品| 成人亚洲欧美一区二区av| 亚洲情色 制服丝袜| 国产亚洲av片在线观看秒播厂| 波多野结衣一区麻豆| 最近2019中文字幕mv第一页| 中文天堂在线官网| av不卡在线播放| 欧美 亚洲 国产 日韩一| 国产午夜精品一二区理论片| 国产国拍精品亚洲av在线观看| 欧美成人午夜免费资源| 五月伊人婷婷丁香| 飞空精品影院首页| 黄片无遮挡物在线观看| 亚洲精品成人av观看孕妇| 久久久久久久久久成人| 九色成人免费人妻av| 丝袜在线中文字幕| 国产在线视频一区二区| 亚洲国产成人一精品久久久| 国产日韩欧美在线精品| 欧美xxⅹ黑人| 久久热在线av| 欧美丝袜亚洲另类| 九草在线视频观看| xxx大片免费视频| 久久精品夜色国产| 亚洲性久久影院| 视频区图区小说| 久久久久久久精品精品| 极品少妇高潮喷水抽搐| 男的添女的下面高潮视频| 午夜免费鲁丝| 丝袜喷水一区| 视频在线观看一区二区三区| 亚洲精品久久午夜乱码| 伊人久久国产一区二区| av在线观看视频网站免费| 大片免费播放器 马上看| 日韩电影二区| 婷婷色综合大香蕉| 五月天丁香电影| 国产在线视频一区二区| 国产一区亚洲一区在线观看| 久久久久网色| 涩涩av久久男人的天堂| 午夜日本视频在线| 日本猛色少妇xxxxx猛交久久| 日韩 亚洲 欧美在线| 男女无遮挡免费网站观看| 国产精品一国产av| 国产69精品久久久久777片| av网站免费在线观看视频| 一区二区三区四区激情视频| 欧美精品高潮呻吟av久久| 最黄视频免费看| 亚洲人与动物交配视频| 午夜免费男女啪啪视频观看| av在线观看视频网站免费| 国精品久久久久久国模美| 自线自在国产av| 欧美xxⅹ黑人| 观看美女的网站| 丁香六月天网| 精品人妻偷拍中文字幕| 亚洲成人手机| 国产又色又爽无遮挡免| 精品国产一区二区久久| 亚洲国产色片| 亚洲中文av在线| 精品久久国产蜜桃| 飞空精品影院首页| 纵有疾风起免费观看全集完整版| 久久久久久久精品精品| 成人综合一区亚洲| 欧美bdsm另类| 国产精品久久久久久精品电影小说| 国产成人精品福利久久| 一级片'在线观看视频| 晚上一个人看的免费电影| 国产 一区精品| 欧美丝袜亚洲另类| 亚洲熟女精品中文字幕| 18禁动态无遮挡网站| 日本色播在线视频| 久久久久久久国产电影| 国产精品秋霞免费鲁丝片| 中文字幕免费在线视频6| 婷婷色综合大香蕉| 内地一区二区视频在线| 人人妻人人澡人人看| 久久久久久久国产电影| 草草在线视频免费看| 国产爽快片一区二区三区| 久久精品国产自在天天线| 国产精品久久久久久久久免| 亚洲精品,欧美精品| 日韩 亚洲 欧美在线| 美女大奶头黄色视频| 精品国产国语对白av| 久久国产精品大桥未久av| 亚洲婷婷狠狠爱综合网| 女人久久www免费人成看片| 日日爽夜夜爽网站| 波野结衣二区三区在线| 美国免费a级毛片| 精品酒店卫生间| 久久 成人 亚洲| 国产在线视频一区二区| 在线观看免费高清a一片| 国产白丝娇喘喷水9色精品| 男女边吃奶边做爰视频| 日韩在线高清观看一区二区三区| 男女高潮啪啪啪动态图| 色5月婷婷丁香| 男女边吃奶边做爰视频| 国产精品国产三级国产专区5o| 久久午夜福利片| 亚洲图色成人| 免费女性裸体啪啪无遮挡网站| 久久人人爽人人爽人人片va| 91精品伊人久久大香线蕉| 免费看不卡的av| 母亲3免费完整高清在线观看 | 免费少妇av软件| 亚洲色图 男人天堂 中文字幕 | 麻豆乱淫一区二区| 久久久欧美国产精品| 久久久久视频综合| 国产 精品1| 色94色欧美一区二区| 观看美女的网站| 又黄又爽又刺激的免费视频.| 亚洲av成人精品一二三区| 亚洲精品456在线播放app| 狠狠婷婷综合久久久久久88av| 久久青草综合色| 美女中出高潮动态图| 色婷婷av一区二区三区视频| 亚洲av综合色区一区| 亚洲第一区二区三区不卡| 精品第一国产精品| 成人午夜精彩视频在线观看| 美女国产视频在线观看| 精品99又大又爽又粗少妇毛片| freevideosex欧美| 高清欧美精品videossex| 90打野战视频偷拍视频| 国产成人精品婷婷| 午夜福利视频在线观看免费| 一区二区三区精品91| 日本av手机在线免费观看| 久久久久久久亚洲中文字幕| 91aial.com中文字幕在线观看| 99久国产av精品国产电影| 丝瓜视频免费看黄片| 欧美人与性动交α欧美精品济南到 | 天天操日日干夜夜撸| 有码 亚洲区| 亚洲国产毛片av蜜桃av| 菩萨蛮人人尽说江南好唐韦庄| 欧美日韩视频高清一区二区三区二| 亚洲精品久久成人aⅴ小说| 精品国产一区二区久久| 久久精品国产综合久久久 | 伦精品一区二区三区| 亚洲精品乱久久久久久| 一级黄片播放器| 国产探花极品一区二区| 男女高潮啪啪啪动态图| 亚洲丝袜综合中文字幕| 两性夫妻黄色片 | 国产成人欧美| 日韩一本色道免费dvd| 亚洲国产精品一区三区| 久久精品aⅴ一区二区三区四区 | 亚洲av中文av极速乱| 下体分泌物呈黄色| 亚洲欧美一区二区三区黑人 | 日本欧美国产在线视频| 国产男女超爽视频在线观看| 激情五月婷婷亚洲| 精品亚洲成a人片在线观看| 大话2 男鬼变身卡| 男男h啪啪无遮挡| 欧美激情极品国产一区二区三区 | 亚洲精品美女久久av网站| 日韩不卡一区二区三区视频在线| 亚洲国产成人一精品久久久| 亚洲色图 男人天堂 中文字幕 | 婷婷色av中文字幕| 欧美人与善性xxx| 日本av免费视频播放| 大片免费播放器 马上看| 久久精品国产亚洲av涩爱| 欧美性感艳星| 91成人精品电影| 国产一区二区在线观看日韩| 黑人欧美特级aaaaaa片| 热99久久久久精品小说推荐| 极品人妻少妇av视频| 自线自在国产av| 成人国产麻豆网| 欧美 日韩 精品 国产| 日产精品乱码卡一卡2卡三| 少妇的逼好多水| 亚洲丝袜综合中文字幕| 2018国产大陆天天弄谢| 男女无遮挡免费网站观看| 国产成人免费观看mmmm| 婷婷成人精品国产| 大香蕉久久成人网| 最近中文字幕高清免费大全6| 伦精品一区二区三区| 观看美女的网站| 十八禁网站网址无遮挡| 久久久久久久精品精品| 亚洲精品456在线播放app| 亚洲色图 男人天堂 中文字幕 | 天天影视国产精品| 精品午夜福利在线看| 日本91视频免费播放| 各种免费的搞黄视频| 美国免费a级毛片| 香蕉丝袜av| 国内精品宾馆在线| 亚洲性久久影院| 亚洲精品久久午夜乱码| 欧美日本中文国产一区发布| 七月丁香在线播放| 久久久a久久爽久久v久久| 哪个播放器可以免费观看大片| 超碰97精品在线观看| 亚洲熟女精品中文字幕| 午夜福利影视在线免费观看| 国产成人午夜福利电影在线观看| 国产日韩一区二区三区精品不卡| 亚洲精华国产精华液的使用体验| 国产国拍精品亚洲av在线观看| 另类精品久久| 91aial.com中文字幕在线观看| 丝袜在线中文字幕| 中文天堂在线官网| 日韩 亚洲 欧美在线| 交换朋友夫妻互换小说| 99久久精品国产国产毛片| 亚洲第一区二区三区不卡| 亚洲精品久久午夜乱码| videosex国产| 亚洲精品国产av成人精品| 性高湖久久久久久久久免费观看| 啦啦啦视频在线资源免费观看| 曰老女人黄片| 亚洲精品日韩在线中文字幕| 亚洲五月色婷婷综合| 男女边摸边吃奶| 下体分泌物呈黄色| 美女大奶头黄色视频| 亚洲五月色婷婷综合| 91aial.com中文字幕在线观看| 亚洲成人一二三区av| 成人毛片60女人毛片免费| 国产 精品1| 免费人妻精品一区二区三区视频| 免费黄色在线免费观看| 欧美日韩视频精品一区| 高清不卡的av网站| 欧美人与善性xxx| 黄色毛片三级朝国网站| 夫妻性生交免费视频一级片| 中文字幕精品免费在线观看视频 | 夫妻午夜视频| 久久亚洲国产成人精品v| 国产精品三级大全| 欧美+日韩+精品| 卡戴珊不雅视频在线播放| 在线精品无人区一区二区三| 午夜福利视频精品| 亚洲,欧美,日韩| 18禁在线无遮挡免费观看视频| 久久女婷五月综合色啪小说| 亚洲av免费高清在线观看| 成人毛片a级毛片在线播放| 制服人妻中文乱码| 成年人免费黄色播放视频| 国产免费一区二区三区四区乱码| 免费看不卡的av| 色哟哟·www| 国产探花极品一区二区| 精品卡一卡二卡四卡免费| 免费女性裸体啪啪无遮挡网站| 欧美日韩av久久| 久久婷婷青草| 在线观看美女被高潮喷水网站| videosex国产| 亚洲精品日本国产第一区| 久久免费观看电影| 水蜜桃什么品种好| 人成视频在线观看免费观看| 三级国产精品片| 视频区图区小说| a级毛片黄视频| 高清视频免费观看一区二区| 综合色丁香网| 欧美日本中文国产一区发布| 一级毛片我不卡| 午夜福利乱码中文字幕| 午夜激情久久久久久久| 国产成人一区二区在线| 成人漫画全彩无遮挡| 伦理电影大哥的女人| 男女国产视频网站| 午夜免费男女啪啪视频观看| 黄色毛片三级朝国网站| 大话2 男鬼变身卡| 麻豆精品久久久久久蜜桃| 国产欧美日韩综合在线一区二区| 少妇人妻精品综合一区二区| 午夜av观看不卡| 在线观看www视频免费| 又大又黄又爽视频免费| 久久午夜福利片| 波野结衣二区三区在线| 9热在线视频观看99| 观看av在线不卡| 纯流量卡能插随身wifi吗| 一级毛片 在线播放| a 毛片基地| 日本午夜av视频| 一级爰片在线观看| 精品一区在线观看国产| 天天躁夜夜躁狠狠躁躁| 国产亚洲欧美精品永久| 精品少妇黑人巨大在线播放| 久久久精品94久久精品| 亚洲精品美女久久av网站| 日本欧美国产在线视频| 久久女婷五月综合色啪小说| 9191精品国产免费久久|