• <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.

    av卡一久久| 精华霜和精华液先用哪个| 熟妇人妻久久中文字幕3abv| 联通29元200g的流量卡| 欧美最黄视频在线播放免费| 99热精品在线国产| 亚洲在线观看片| 国产在视频线在精品| 久久99热6这里只有精品| 99热6这里只有精品| 嘟嘟电影网在线观看| 国产综合懂色| 蜜桃久久精品国产亚洲av| 精品久久久久久久人妻蜜臀av| 不卡视频在线观看欧美| 久久久久久久久中文| 国产色爽女视频免费观看| 国产v大片淫在线免费观看| 午夜激情欧美在线| 十八禁国产超污无遮挡网站| 99视频精品全部免费 在线| 搡女人真爽免费视频火全软件| 日本爱情动作片www.在线观看| 好男人视频免费观看在线| 日韩成人av中文字幕在线观看| 亚洲av不卡在线观看| 能在线免费观看的黄片| 好男人视频免费观看在线| 99国产精品一区二区蜜桃av| 免费搜索国产男女视频| 亚洲四区av| 久久精品国产鲁丝片午夜精品| 国内精品一区二区在线观看| 日本色播在线视频| av卡一久久| 精品久久久久久成人av| 亚洲无线观看免费| 国产伦在线观看视频一区| 99久久精品一区二区三区| 精品少妇黑人巨大在线播放 | 亚洲精品国产av成人精品| 嫩草影院入口| 午夜爱爱视频在线播放| 久久久久久久久中文| 欧美+亚洲+日韩+国产| 国产一区二区亚洲精品在线观看| 免费在线观看成人毛片| 人人妻人人看人人澡| 免费av不卡在线播放| 久久久久久久久久成人| 午夜精品一区二区三区免费看| 日本黄色片子视频| 一级av片app| 在线播放无遮挡| 精品人妻偷拍中文字幕| 国产一区二区亚洲精品在线观看| 女人被狂操c到高潮| 亚洲无线观看免费| 村上凉子中文字幕在线| 久久亚洲精品不卡| 菩萨蛮人人尽说江南好唐韦庄 | 国产精品久久久久久久电影| 亚洲av免费在线观看| 日本一二三区视频观看| 亚洲久久久久久中文字幕| 亚洲最大成人手机在线| 久久久色成人| 老司机福利观看| 一个人看视频在线观看www免费| 国产午夜精品久久久久久一区二区三区| 欧美性感艳星| 成人二区视频| www.av在线官网国产| 欧美日本亚洲视频在线播放| 亚洲一级一片aⅴ在线观看| 深夜精品福利| 亚洲av.av天堂| 欧美性猛交╳xxx乱大交人| 在线免费观看不下载黄p国产| 午夜视频国产福利| 精华霜和精华液先用哪个| 成人午夜高清在线视频| 女人被狂操c到高潮| 久久久久久久久中文| 成年版毛片免费区| 亚洲性久久影院| 老熟妇乱子伦视频在线观看| 欧美bdsm另类| 性插视频无遮挡在线免费观看| 国产精品永久免费网站| 麻豆久久精品国产亚洲av| 床上黄色一级片| 乱系列少妇在线播放| 国产色婷婷99| 久久亚洲国产成人精品v| av黄色大香蕉| 久久久精品大字幕| 国内少妇人妻偷人精品xxx网站| 亚洲精品国产成人久久av| 日本撒尿小便嘘嘘汇集6| 伦精品一区二区三区| 五月伊人婷婷丁香| 国产精品嫩草影院av在线观看| 国产综合懂色| 一个人看视频在线观看www免费| 白带黄色成豆腐渣| 全区人妻精品视频| av免费在线看不卡| 少妇人妻精品综合一区二区 | 国产蜜桃级精品一区二区三区| 在线观看午夜福利视频| 可以在线观看毛片的网站| ponron亚洲| 国产一区亚洲一区在线观看| 99九九线精品视频在线观看视频| av卡一久久| 国产精品久久久久久精品电影小说 | 亚洲精品日韩av片在线观看| 亚洲乱码一区二区免费版| 久久精品久久久久久噜噜老黄 | 国内揄拍国产精品人妻在线| 国产av不卡久久| 狂野欧美激情性xxxx在线观看| 一区二区三区四区激情视频 | 亚洲精品自拍成人| 我的女老师完整版在线观看| 中国美女看黄片| 午夜精品一区二区三区免费看| 大型黄色视频在线免费观看| 一个人观看的视频www高清免费观看| 蜜桃久久精品国产亚洲av| 中文资源天堂在线| 老熟妇乱子伦视频在线观看| 你懂的网址亚洲精品在线观看 | 搡女人真爽免费视频火全软件| 亚洲婷婷狠狠爱综合网| 人人妻人人看人人澡| 日本色播在线视频| 精品一区二区三区视频在线| 看免费成人av毛片| 99热只有精品国产| 97热精品久久久久久| 国产黄色视频一区二区在线观看 | 国产在线精品亚洲第一网站| 日韩人妻高清精品专区| 日本免费a在线| 嘟嘟电影网在线观看| 国产成人一区二区在线| 亚洲最大成人中文| 插逼视频在线观看| 色噜噜av男人的天堂激情| 嫩草影院入口| 日韩一区二区视频免费看| 国产精品无大码| 亚洲精品影视一区二区三区av| 国产色爽女视频免费观看| 男女啪啪激烈高潮av片| 一区二区三区免费毛片| 久久精品国产亚洲av天美| 亚洲在线观看片| 亚洲欧美日韩东京热| 成人毛片a级毛片在线播放| 久久精品影院6| 麻豆av噜噜一区二区三区| 免费av不卡在线播放| 亚洲内射少妇av| 国产精品伦人一区二区| 国产极品精品免费视频能看的| 久久久久久国产a免费观看| 久久久久性生活片| 久久久精品大字幕| 欧美激情久久久久久爽电影| 欧美又色又爽又黄视频| 听说在线观看完整版免费高清| 国产精品伦人一区二区| 日韩欧美 国产精品| av在线天堂中文字幕| 亚洲精品乱码久久久v下载方式| 国产免费一级a男人的天堂| 免费大片18禁| 直男gayav资源| 久久精品国产亚洲av涩爱 | 欧美极品一区二区三区四区| av在线老鸭窝| 亚洲精品日韩在线中文字幕 | av.在线天堂| 色哟哟·www| 欧美成人免费av一区二区三区| 中文亚洲av片在线观看爽| 淫秽高清视频在线观看| 国产日本99.免费观看| 久久久久久久亚洲中文字幕| 成人永久免费在线观看视频| 免费观看在线日韩| 波多野结衣高清作品| 韩国av在线不卡| 一级黄色大片毛片| 欧美性猛交╳xxx乱大交人| 日韩欧美国产在线观看| 精华霜和精华液先用哪个| 国产精品嫩草影院av在线观看| 18+在线观看网站| 欧美日韩在线观看h| 黑人高潮一二区| 嫩草影院入口| 欧美高清性xxxxhd video| 国产极品精品免费视频能看的| 免费大片18禁| 国产色婷婷99| 午夜a级毛片| 简卡轻食公司| 97超碰精品成人国产| 国产高潮美女av| 久久精品国产清高在天天线| 一级毛片电影观看 | 婷婷色av中文字幕| 亚洲中文字幕日韩| 日日摸夜夜添夜夜爱| 欧美潮喷喷水| 久久婷婷人人爽人人干人人爱| 久久精品夜色国产| 伦理电影大哥的女人| 国产不卡一卡二| 噜噜噜噜噜久久久久久91| 精品久久久久久久末码| 精品久久久久久久久久久久久| 午夜福利高清视频| 给我免费播放毛片高清在线观看| 少妇高潮的动态图| 我要搜黄色片| 国产精品野战在线观看| 久久久久久久久久黄片| 高清毛片免费看| 日韩人妻高清精品专区| 91在线精品国自产拍蜜月| 三级毛片av免费| 日韩视频在线欧美| 九九在线视频观看精品| 国产成年人精品一区二区| 如何舔出高潮| 日韩精品青青久久久久久| 午夜福利在线观看吧| 亚洲无线观看免费| 亚洲18禁久久av| 国产精品蜜桃在线观看 | 久久6这里有精品| 蜜桃久久精品国产亚洲av| 亚洲国产日韩欧美精品在线观看| 日本黄大片高清| 国产高清激情床上av| 欧美成人一区二区免费高清观看| 伊人久久精品亚洲午夜| 床上黄色一级片| 99久久成人亚洲精品观看| 黄片无遮挡物在线观看| 禁无遮挡网站| 少妇高潮的动态图| 嫩草影院精品99| 蜜桃久久精品国产亚洲av| h日本视频在线播放| 最好的美女福利视频网| 国产一区二区在线av高清观看| 一本久久精品| 国产精品一区二区三区四区免费观看| av在线天堂中文字幕| 中文字幕久久专区| 国产精品人妻久久久影院| 精品99又大又爽又粗少妇毛片| 日日啪夜夜撸| 久久午夜亚洲精品久久| 成人特级黄色片久久久久久久| 亚洲熟妇中文字幕五十中出| 欧美色视频一区免费| 国产成人精品一,二区 | 日本爱情动作片www.在线观看| 精品久久久噜噜| 最新中文字幕久久久久| 中出人妻视频一区二区| 最好的美女福利视频网| 秋霞在线观看毛片| 中文字幕av成人在线电影| 高清毛片免费观看视频网站| 免费av不卡在线播放| 最近视频中文字幕2019在线8| 欧美极品一区二区三区四区| 亚洲国产精品合色在线| 欧美bdsm另类| 成人特级av手机在线观看| 欧美色视频一区免费| 久久久久久久亚洲中文字幕| 国产精品一及| 亚洲第一电影网av| 热99在线观看视频| 高清毛片免费看| 91av网一区二区| 久久久久性生活片| 国产极品天堂在线| 国产精品美女特级片免费视频播放器| 伊人久久精品亚洲午夜| 国内揄拍国产精品人妻在线| 国产高清有码在线观看视频| 色视频www国产| 日韩国内少妇激情av| 成人特级av手机在线观看| 黄色欧美视频在线观看| 黑人高潮一二区| 国产乱人偷精品视频| 91aial.com中文字幕在线观看| 国产精品一区www在线观看| 插逼视频在线观看| 欧美成人a在线观看| 色5月婷婷丁香| 久久精品夜夜夜夜夜久久蜜豆| 少妇猛男粗大的猛烈进出视频 | 国产精品1区2区在线观看.| 校园人妻丝袜中文字幕| 国产成人freesex在线| 乱人视频在线观看| 啦啦啦韩国在线观看视频| 国产精品,欧美在线| 欧美高清性xxxxhd video| 婷婷色综合大香蕉| 1024手机看黄色片| 国产精品人妻久久久久久| 自拍偷自拍亚洲精品老妇| 中文字幕熟女人妻在线| 十八禁国产超污无遮挡网站| 99国产极品粉嫩在线观看| 能在线免费观看的黄片| 大型黄色视频在线免费观看| 亚洲四区av| 日韩,欧美,国产一区二区三区 | 久久精品久久久久久噜噜老黄 | 午夜激情福利司机影院| 精品人妻偷拍中文字幕| 午夜爱爱视频在线播放| 久久精品人妻少妇| 少妇熟女欧美另类| 国产免费一级a男人的天堂| 男插女下体视频免费在线播放| 黄色日韩在线| 国产一区二区亚洲精品在线观看| 久久草成人影院| 日韩欧美 国产精品| 午夜视频国产福利| 久久久a久久爽久久v久久| 久久人人精品亚洲av| 国产精品美女特级片免费视频播放器| 精品不卡国产一区二区三区| 亚洲内射少妇av| av黄色大香蕉| 男人舔女人下体高潮全视频| a级毛色黄片| 精品一区二区三区人妻视频| 狠狠狠狠99中文字幕| 国产成人精品一,二区 | 精品欧美国产一区二区三| 欧美+日韩+精品| 丰满乱子伦码专区| 国产综合懂色| 日韩制服骚丝袜av| 成人亚洲精品av一区二区| 久久欧美精品欧美久久欧美| 亚洲av免费高清在线观看| 一级黄色大片毛片| 亚洲av第一区精品v没综合| 中文字幕久久专区| 亚洲欧美成人综合另类久久久 | 特大巨黑吊av在线直播| 欧美三级亚洲精品| 欧美性猛交黑人性爽| 色视频www国产| 欧美最新免费一区二区三区| 国国产精品蜜臀av免费| 亚洲一区高清亚洲精品| 日本黄色视频三级网站网址| 青春草亚洲视频在线观看| 又黄又爽又刺激的免费视频.| 美女黄网站色视频| 国产色婷婷99| 国产黄a三级三级三级人| 青春草视频在线免费观看| 少妇的逼好多水| 久久这里有精品视频免费| 欧美一区二区国产精品久久精品| 啦啦啦韩国在线观看视频| 欧美日韩精品成人综合77777| 久久九九热精品免费| 欧美日本视频| 一区福利在线观看| 亚洲国产精品成人久久小说 | 久久精品夜色国产| 日本与韩国留学比较| 亚洲一区高清亚洲精品| 最新中文字幕久久久久| 国产大屁股一区二区在线视频| 爱豆传媒免费全集在线观看| 日韩一区二区视频免费看| 久久久久久大精品| 欧美激情国产日韩精品一区| 天美传媒精品一区二区| 成人美女网站在线观看视频| 精品人妻偷拍中文字幕| 色综合站精品国产| 国产av一区在线观看免费| 日本色播在线视频| 日本三级黄在线观看| 亚洲av.av天堂| 99热这里只有是精品50| 亚洲国产欧美人成| 欧美人与善性xxx| 日日干狠狠操夜夜爽| 国语自产精品视频在线第100页| 一边摸一边抽搐一进一小说| 我的老师免费观看完整版| 可以在线观看的亚洲视频| 12—13女人毛片做爰片一| 性插视频无遮挡在线免费观看| 国产精品电影一区二区三区| 亚洲成人中文字幕在线播放| 99久国产av精品国产电影| 少妇高潮的动态图| 国产黄片视频在线免费观看| 亚洲经典国产精华液单| 男人和女人高潮做爰伦理| 久久久久久九九精品二区国产| 少妇人妻精品综合一区二区 | 亚洲国产色片| 国产精品久久久久久亚洲av鲁大| 男人狂女人下面高潮的视频| 亚洲欧美日韩无卡精品| 夜夜爽天天搞| 国产欧美日韩精品一区二区| 久久亚洲精品不卡| 晚上一个人看的免费电影| 亚洲欧美成人综合另类久久久 | 亚洲电影在线观看av| 国产真实乱freesex| 一区二区三区高清视频在线| 能在线免费看毛片的网站| 人妻制服诱惑在线中文字幕| 亚洲一级一片aⅴ在线观看| 日本黄色片子视频| 12—13女人毛片做爰片一| 成人无遮挡网站| 亚洲电影在线观看av| 精品人妻熟女av久视频| 亚洲自拍偷在线| 亚洲不卡免费看| 国产精品,欧美在线| 国产欧美日韩精品一区二区| 日韩成人av中文字幕在线观看| 一个人看的www免费观看视频| 亚洲欧美清纯卡通| 亚洲激情五月婷婷啪啪| 免费在线观看成人毛片| 99久久久亚洲精品蜜臀av| 国产精品久久久久久av不卡| 国产激情偷乱视频一区二区| 国产蜜桃级精品一区二区三区| 又爽又黄无遮挡网站| 99久久久亚洲精品蜜臀av| av黄色大香蕉| 久久久午夜欧美精品| 偷拍熟女少妇极品色| 国产精品久久电影中文字幕| 蜜臀久久99精品久久宅男| 在线免费观看不下载黄p国产| 日韩人妻高清精品专区| 免费看日本二区| 久久久久久久久大av| 国产精品野战在线观看| 成人午夜高清在线视频| 国产一级毛片七仙女欲春2| 狂野欧美白嫩少妇大欣赏| av免费观看日本| 真实男女啪啪啪动态图| 青春草亚洲视频在线观看| 国产白丝娇喘喷水9色精品| 亚洲婷婷狠狠爱综合网| 免费av不卡在线播放| 色播亚洲综合网| 午夜视频国产福利| 国产精品人妻久久久影院| 精品99又大又爽又粗少妇毛片| 国产真实伦视频高清在线观看| 亚洲av电影不卡..在线观看| 亚洲精品色激情综合| 国产午夜精品一二区理论片| 毛片一级片免费看久久久久| 最近中文字幕高清免费大全6| 亚洲国产精品合色在线| 18禁在线无遮挡免费观看视频| 国产白丝娇喘喷水9色精品| а√天堂www在线а√下载| 欧美日韩国产亚洲二区| 色综合站精品国产| 国产高清三级在线| 国产成人91sexporn| 色尼玛亚洲综合影院| 中文精品一卡2卡3卡4更新| 亚洲性久久影院| 99riav亚洲国产免费| 大型黄色视频在线免费观看| 九九爱精品视频在线观看| 变态另类丝袜制服| 久久久久国产网址| av女优亚洲男人天堂| 亚洲成人中文字幕在线播放| 欧美一级a爱片免费观看看| 黄色视频,在线免费观看| 在线观看66精品国产| 亚洲精品久久国产高清桃花| h日本视频在线播放| 女人十人毛片免费观看3o分钟| 伦精品一区二区三区| 听说在线观看完整版免费高清| 日韩在线高清观看一区二区三区| 免费搜索国产男女视频| 大型黄色视频在线免费观看| 亚洲国产精品久久男人天堂| 欧美性猛交黑人性爽| 一个人看的www免费观看视频| 乱系列少妇在线播放| 最新中文字幕久久久久| 免费一级毛片在线播放高清视频| 日日摸夜夜添夜夜爱| 一区二区三区高清视频在线| 校园春色视频在线观看| 色哟哟·www| 国产白丝娇喘喷水9色精品| 高清毛片免费观看视频网站| 亚洲中文字幕一区二区三区有码在线看| 亚洲最大成人手机在线| 欧美日韩一区二区视频在线观看视频在线 | 亚洲自偷自拍三级| 日韩国内少妇激情av| 久久午夜亚洲精品久久| www日本黄色视频网| 天堂网av新在线| 国产黄色小视频在线观看| 国内精品美女久久久久久| 国产精品久久久久久久久免| 哪里可以看免费的av片| 成年版毛片免费区| 在线免费观看不下载黄p国产| 免费看a级黄色片| 校园春色视频在线观看| 男女做爰动态图高潮gif福利片| 国产视频内射| 亚洲乱码一区二区免费版| 色视频www国产| 免费观看a级毛片全部| 日本三级黄在线观看| 综合色av麻豆| 亚洲中文字幕日韩| 久久草成人影院| 国产黄a三级三级三级人| 国产黄片美女视频| 国产日本99.免费观看| 在线免费观看的www视频| 两性午夜刺激爽爽歪歪视频在线观看| 99久国产av精品国产电影| av国产免费在线观看| 日韩一区二区视频免费看| 亚洲五月天丁香| av天堂在线播放| 午夜福利成人在线免费观看| 日日撸夜夜添| 日韩大尺度精品在线看网址| 长腿黑丝高跟| 美女被艹到高潮喷水动态| 欧美最新免费一区二区三区| 午夜亚洲福利在线播放| 成人毛片60女人毛片免费| 成年av动漫网址| 嫩草影院精品99| 精品不卡国产一区二区三区| 在线免费观看不下载黄p国产| 国产精品一及| 国产熟女欧美一区二区| 午夜老司机福利剧场| 精品人妻熟女av久视频| 大香蕉久久网| 久久久精品欧美日韩精品| 女的被弄到高潮叫床怎么办| 国产精品久久久久久久电影| 69av精品久久久久久| 久久精品国产亚洲av涩爱 | 又爽又黄a免费视频| 26uuu在线亚洲综合色| 国产精品伦人一区二区| 日韩一本色道免费dvd| 99在线视频只有这里精品首页| 午夜福利在线观看免费完整高清在 | 午夜a级毛片| 一卡2卡三卡四卡精品乱码亚洲| 精品人妻视频免费看| 免费看日本二区| 少妇人妻一区二区三区视频| 性欧美人与动物交配| 亚洲自拍偷在线| 国产中年淑女户外野战色| 两个人的视频大全免费| 久久久久网色| 国内少妇人妻偷人精品xxx网站| 99久国产av精品国产电影| 亚洲中文字幕一区二区三区有码在线看| 亚洲婷婷狠狠爱综合网| 熟女电影av网| 禁无遮挡网站| 我的老师免费观看完整版| 人妻少妇偷人精品九色| 麻豆久久精品国产亚洲av| 国产真实乱freesex|