• <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 ?

    A PENALTY FUNCTION METHOD FOR THE PRINCIPAL-AGENT PROBLEM WITH AN INFINITE NUMBER OF INCENTIVE-COMPATIBILITY CONSTRAINTS UNDER MORAL HAZARD?

    2021-10-28 05:45:20JiaLIU劉佳
    關(guān)鍵詞:劉佳

    Jia LIU(劉佳)

    School of Economics and Management,Wuhan University,Wuhan 430072,China

    E-mail:liujia.06@163.com

    Xianjia WANG(王先甲)

    School of Economics and Management,Wuhan University,Wuhan 430072,China Institute of Systems Engineering,Wuhan University,Wuhan 430072,China

    E-mail:wangxj@whu.edu.cn

    Abstract In this paper,we propose an iterative algorithm to find the optimal incentive mechanism for the principal-agent problem under moral hazard where the number of agent action pro files is in finite,and where there are an in finite number of results that can be observed by the principal.This principal-agent problem has an in finite number of incentivecompatibility constraints,and we transform it into an optimization problem with an in finite number of constraints called a semi-in finite programming problem.We then propose an exterior penalty function method to find the optimal solution to this semi-in finite programming and illustrate the convergence of this algorithm.By analyzing the optimal solution obtained by the proposed penalty function method,we can obtain the optimal incentive mechanism for the principal-agent problem with an in finite number of incentive-compatibility constraints under moral hazard.

    Key words principal-agent problem;mechanism design;moral hazard;semi-in finite programming problem;penalty function method

    1 Introduction

    In the principal-agent problem under moral hazard,the agent has private information about his personal behaviors,while the principal can only observe the results caused by the agent’s actions.In this case,the principal needs to design an incentive mechanism for all agents based on the observed results to maximize his expected pro fit.Since the agent has some private information and acts with the goal of maximizing his own expected pro fit,the purpose of designing the incentive mechanism for the principal is to maximize his own expected pro fit and to motivate the agents to act in a certain way.

    The optimal incentive mechanism satis fies two kinds of constraints:the incentive-compatibility constraint and individual rationality constraint.The incentive-compatibility constraint shows that the expected pro fit obtained by the agent when he chooses the action desired by the principal is not less than that obtained when he chooses any other action.This is the motivation of the principal to design the incentive mechanism based on the agent’s actions to maximize their own pro fits.The individual rationality constraint shows that the agent’s pro fit under the optimal incentive mechanism is not less than the maximum pro fit obtained when the optimal incentive mechanism is not accepted.This is the basis for the agent to participate in the incentive mechanism.All feasible actions taken by all individuals in making decisions must satisfy these two constraints.

    In the incentive mechanism design problem under moral hazard,the principal designs different payment schemes to the agent according to the observable results.Generally,it is assumed that the number of agent action pro files is finite and there are a limited number of results that can be observed by the principal.This assumption makes the number of incentive-compatibility constraints limited when the principal seeks to maximize his expected pro fit.Given the form of the optimal incentive mechanism,the problem of finding the optimal incentive mechanism can be transformed into that of solving an optimization problem with finite constraints.In this case,the Kuhn-Tucker theorem or the Lagrange method can be used to solve the optimization problem,and the optimal incentive mechanism can be obtained according to the optimal solution to this optimization problem.The vast majority of researchers study incentive mechanism design problems based on this idea.However,in practical economic and political problems,agents may have an in finite number of action pro files and there are an in finite number of results that can be observed by a principal,which makes the number of incentive-compatibility constraints in finite.At this point,the incentive mechanism design problem with an in finite number of incentive-compatibility constraints can be transformed into an optimization problem with an in finite number of constraints;this is called a semi-in finite programming problem.

    The main aim of this paper is to design an iterative algorithm to find the optimal incentive mechanism for the principal-agent problem with an in finite number of constraints under moral hazard.When the number of variables increases and the number of constraints increases,the traditional method becomes too complex to find an analytic solution.Since the 1970s,there has been much development in terms of solving general mechanism design problems,and a feasible analytical framework for obtaining analytical solutions has been established.However,there are many inconveniences when there are too many variables and too many constraints,such as discussion of the complex optimality conditions used in analytical methods.Inspired by the algorithm of optimization theory,we hope to design a iterative algorithm which can solve some complex mechanism design problems.This is the main motivation of this paper.

    The iterative algorithm designed in this paper is a exterior penalty function method,called the M-penalty function method,where M is the ampli fication factor for the penalty function.This algorithm is designed to transform the solving of a semi-in finite programming problem into the solving of a series of unconstrained optimization problems.By establishing a set of penalty functions,the optimal solutions of the unconstrained programming problems can be approximated to the optimal solutions of the original semi-in finite programming problems.So far,few researchers have used the optimization algorithm to find the optimal incentive mechanism of the principal-agent problem with an in finite number of constraints.

    The traditional analytical method and the iterative algorithm designed in this paper have their own advantages and disadvantages.For the principal-agent problem with a simple setting,it is more convenient to use the traditional analytical method to find the optimal incentive mechanism.When the parameters become more numerous and the constraints become more complex,the traditional method is more limited and it is difficult to get the optimal solution.At this point,the iterative algorithm proposed in this paper is more conducive to solving principal-agent problems.

    Since there are no additional assumptions on individual utility functions,such as satisfying convexity or concavity,the proposed iterative algorithm can solve the mechanism design problem with complex functions that cannot be solved by traditional methods.Also,because this paper does not make additional assumptions about the form of the contract of the principal,the iterative algorithm in this paper can be used for the nonlinear contract as well as for the traditional linear contract.

    In order to construct an iterative algorithm to find the optimal incentive mechanism for the principal-agent problem under moral hazard,this paper transforms this principal-agent problem into a semi-in finite programming problem.Then,by designing a penalty function method for solving semi-in finite programming problems,we obtain an iterative algorithm to find the optimal incentive mechanism for this principal-agent problem.

    The structure of this paper is as follows:in Section 3,the principal-agent problem with an in finite number of agent action pro files and an in finite number of observable results for the principal under moral hazard is established.In the Section 4,the principal-agent problem is transformed into a standard semi-in finite programming problem(SIP),and we propose a kind of exterior penalty function method,called the M-penalty function method,to solve this semi-in finite programming problem.In Section 5,we study the convergence of the M-penalty function method.In Section 6,a numerical example is used to illustrate the method of finding the optimal incentive mechanism of a two-person principal-agent problem under moral hazard where the agent’s action pro file is in finite and the principal can observe an in finite number of results.Finally,Section 7 offers the conclusions of this paper.

    2 Related Literature

    2.1 Mechanism Design Theory

    The research and development of mechanism design theory includes two categories:the mechanism design problem under adverse selection and the mechanism design problem under moral hazard.

    The main feature of the adverse selection model is the ex ante asymmetric information.Nature selects the type of agent,and the agent knows his own type while the principal does not.In the signaling model,in order to show his type,the agent chooses a signal and the principal signs a contract with the agent after observing the signal.In the screening model,the principal provides multiple contracts for the agent to choose,and the agent selects a suitable contract according to his or her own type and acts according to the contract(see,e.g.,[1–6]).

    The main feature of the moral hazard model is the ex post information.The information is symmetrical before signing the contract.In the moral hazard with hidden action model,after signing the contract,the principal can only observe the results caused by the agent’s choice of actions and the state of the world.At this time,the principal should design incentive contracts to induce the agent to choose the most bene ficial action for the principal from the perspective of maximizing the agent’s own pro fits.In the moral hazard with hidden information model,nature chooses the state of the world,the agent observes the nature’s selection and acts,while the principal observes the agent’s actions but cannot observe the state of nature’s selection.At this point,the principal should design an incentive contract to encourage the agent to choose the best action for the principal under a given state of the world(see,e.g.,[7–10]).

    There are several papers in the literature on the incentive mechanism design problem in which agents have continuous action spaces that are closely related to the present work.The main methods for dealing with such a mechanism design problem is based on the analytic method for solving the optimization problem.The basic treatment includes:(1)A discrete agent’s action space to make the incentive-compatibility constraint a finite number.In this case,the Kuhn-Tucker theorem can be used to find the optimal incentive mechanism;(2)Replacing the in finite number of incentive-compatibility constraints with relatively simple local incentive-compatibility constraints.The classic single-agent moral hazard problem already has“in finitely many actions”and thus an“in finite number of incentive-compatibility constraints”(see,e.g.,[11,12]).For example,Laffont and Martmort studied the incentive mechanism design problem in which the agents have a continuous level of effort([12]).Since the optimal solution does not necessarily exist([13]),they used this method to deal with the in finite number of incentive-compatibility constraints,and found the sufficient conditions of the optimal solution to the original principle-agent problem.

    In this paper,we study the incentive mechanism design problem under moral hazard with hidden action.We use an iterative algorithm to find the optimal incentive mechanism for the principle-agent problem under moral hazard where the number of agent action pro files is in finite,and where there are an in finite number of results that can be observed by the principal.The form of the optimal payment mechanism is obtained by an iterative method,which avoids the complex optimization discussion process emerged in using analytical methods.

    2.2 Semi-in finite Programming Problem

    In this paper,it is found that if the number of agent action pro files is in finite and there are an in finite number of results that can be observed by the principal,the optimal incentive mechanism for this incentive mechanism design problem is the optimal solution to an optimization problem with an in finite number of constraints;this called a semi-in finite programming problem.

    The semi-in finite programming problem mainly studies a constrained programming problem with an in finite number of constraints.It is applied in many fields([14]),such as in the data envelopment analysis of an in finite number of decision-making units([15]),robot trajectory planning([16]),and economic,fi nancial and engineering models([17]).

    There are many algorithms for solving semi-in finite programming problems,such as the discretization method([18,19]),the homotopy interior point method([20]),the Newton method([21]),the semismooth Newton method([22]),the trust region method([23]),the augmented Lagrangian method([24]),the exact penalty method([25]),the quadratic programming-type method([26]),and the branch-and-bound method([27]).The idea of designing these methods is basically based on the idea of constructing the methods for finding the optimal solution to the nonlinear programming problem.In this paper,we also design an algorithm for solving semi-in finite programming problems based on this idea.

    In this paper,a type of exterior penalty function method is proposed to solve the semiin finite programming problem;this is called M-penalty function method.The penalty function method is concise in terms of solving semi-in finite programming problems,and the exterior penalty function method is even more concise than the internal penalty function method.The reason why the internal penalty function method is not adopted here is mainly due to the selection of the initial point.The internal penalty function method needs to select the initial point that satis fies all constraints.Compared with the exterior penalty function method,it is more difficult to select the initial point of the internal penalty function method.By discussing the optimal solution obtained by the M-penalty function method,we can get the optimal incentive mechanism for incentive mechanism design problem with an in finite number of incentive-compatibility constraints under moral hazard.

    3 Model

    In a multi-player principal-agent game,a principal(player without private information)wants the agents(players with private information)to choose an appropriate action pro file according to the goal of maximizing his expected pro fit.The principal cannot directly observe the action pro file taken by the agents,but can only observe a certain result which is determined by the agents’action pro file and some exogenous random factors.The principal’s problem is how to design a reward or punishment contract for the agents based on the observed results to encourage them to take actions that maximize his expected pro fit.

    In the principal-agent problem under moral hazard,the order of the game is:the principal proposes a payment contract,then the agent decides whether to participate or not;if he does,he makes a certain effort to act and receives a bene fit under this contract.If he does not participate,the payoffs are all zero.Each individual has the goal of maximizing his or her own return.Therefore,the level of effort taken by the agent and the payment contract proposed by the principal are both obtained by solving a maximization problem.

    The principal’s goal is to design an incentive contract t(s)=(t(s),···,t(s)),which is a payment vector function proposed by the principal to the agents based on the observation of results which is denoted by s.Generally,it is assumed that the observed result s is a function of the agents’action pro file a and the exogenous random variable ξ∈Ξ,which is exogenously given.For example,the observable result s is related to the agents’actions a and the exogenous factor ξ in the following separate form:s(a,ξ)=k(a)+k(ξ)(see,e.g.,([12])).The set of all possible observed results is denoted by S.The probability distribution function and probability density function of the exogenous random variable ξ are P(ξ)and p(ξ),respectively.After the agents take action pro file a∈A,they need to pay a certain cost vector c(a),and the principal can get π(s(a,ξ))based on the observed result s(a,ξ).Suppose that the utility function π(s(a,ξ))is a strictly increasing concave function with respect to a.At this point,that means that the better the agents take the action pro file,the more the principal can obtain,and the marginal revenue will decrease.

    Assume that both the principal and the agents are risk neutral.The expected pro fit of the principal is

    The expected pro fit of agent i∈N is

    The first kind of constraint for designing the optimal incentive mechanism is the incentivecompatibility constraint.The principal cannot observe the agents’action pro file a∈A,and in any incentive mechanism,the agent always chooses the action that maximizes his expected pro fit,so ifˉa is the agents’action pro file that maximizes their expected payoff,and a is the other action pro file that the agents can take,then the agents will only chooseˉa if they can get more expected pro fit when they takeˉa as opposed to when they take a.Therefore,the incentive-compatibility constraint is denoted by IC as follows:

    For the principal,his aim is to design a payment t(s)for all observed results s∈S and to maximize his expected pro fit.Thus,the optimal payment contract is the solution to the following optimization problem:

    In this optimization problem,since the action pro file set A is an in finite point set,the numbers of constraints(IC)on the optimization problem(P)are in finite.In this paper,we assume that the principal knows the form of the payment function when designing the payment vector function,but the payment vector function contains some unknown parameters.For example,the principal designs a linear contract,i.e.,t(s)=α+βs.At this point,fi nding the optimal contract function t(s)is converted to a matter of finding its optimal parameters.In addition,ˉa is determined by the agent by maximizing his utility,and it also depends on the agent’s payment function.

    4 The Optimal Incentive Mechanism

    4.1 Semi-in finite Programming Model

    We transform the incentive mechanism design problem(P)into the following optimization problem(P’):

    There are in finite number of constraints in the optimization problem(P),and we cannot use the Kuhn-Tucker theorem to solve this optimization problem.

    It is assumed that the form of the incentive contract t(s)is known,but contains some unknown parameters.We denote all the variables that need to be decided as x=(x,···,x).As can be seen from the optimization problem(P),its general form is

    where A is an in finite-point measurable set,and N is a finite-point set.This optimization problem is called a semi-in finite programming problem(SIP).

    In the semi-in finite programming problem(SIP),the variable x∈Rrepresent the parameters in the payment vector function t,and there may be more than one unknown parameter.Therefore,after we use the algorithm of the semi-in finite programming problem to get the optimal solution to the optimization problem(P),we can find the optimal incentive mechanism for the incentive mechanism design problem(P).Thus,we need to design an algorithm for solving the semi-in finite programming problem(SIP) first.

    4.2 M-penalty Function Method

    Here,a kind of exterior penalty function method is proposed to solve the semi-in finite programming problem(SIP);this is called the M-penalty function method.The idea of designing this algorithm is derived from transforming the constrained programming problem into an unconstrained programming problem,and then obtaining the optimal solution to the constrained programming problem by solving this unconstrained programming problem.When designing an unconstrained programming problem,if the point is in the feasible domain of the original optimization problem,the objective function value will not be changed.If the point is not in the feasible domain,a penalty term is designed to punish the original objective function.This is an exterior penalty function method.

    First,a constraint violation function for the semi-in finite programming problem(SIP)is constructed and denoted by G(x)as follows:

    Obviously,G(x)=0 if and only if the constraints of the semi-in finite programming problem(SIP)are satis fied.G(x)>0 if and only if the point x∈Rdoes not satisfy the constraints of the semi-in finite programming problem(SIP).

    We de fine the penalty function as

    where σ>0 is the penalty factor.

    We denote x(σ)as a solution to the following optimization problem:

    The relationship between the optimal solution to the optimization problem(PG)and the optimal solution to the original semi-in finite programming problem(SIP)is illustrated in the following Lemma:

    Lemma 4.1

    If x(σ)is the optimal solution to an optimization problem(PG)and G(x(σ))=0,that is,if x(σ)satis fies the constraints of the semi-in finite programming problem(SIP),then x(σ)is the optimal solution to the semi-in finite programming problem(SIP).

    Proof

    Since x(σ)is the optimal solution to the optimization problem(PG),then?x∈R,

    Since x(σ)satis fies the constraints of the semi-in finite programming problem(SIP)such that G(x(σ))=0,for any point x that satis fies the constraints of the semi-in finite programming problem(SIP),we have

    Then x(σ)is also the optimal solution for the semi-in finite programming problem(SIP).Therefore,the conclusion is established.

    Lemma 4.1 shows that finding the optimal solution to the semi-in finite programming problem(SIP)can be transformed into solving the optimization problem(PG).Based on this idea,we design the M-penalty function method to solve the semi-in finite programming problem(SIP)as follows:

    M-penalty function method

    (1)Step 1:Given an initial point x,take σ>0,M>1,ε>0.k:=1.

    (2)Step 2:Using the initial point value xto solve the optimization problem(PG)when σ=σ,obtain x(σ).Let x=x(σ).

    (3)Step 3:If

    then stop,and the optimal solution is x.Otherwise,let σ:=Mσ,k:=k+1,and go to the Step 2.

    Here ε is called the error factor and M is called the ampli fication factor.

    5 Convergence Analysis

    Since M>1 and σ>0,it can be seen from the equation(5.5)that G(x)≤G(x)holds.

    According to equation(5.3)and equation(5.1),we have

    Then,based on equation(5.6),we know that f(x)≥f(x)holds.Therefore,the conclusion is established.

    As can be seen from Lemma 5.1,the value of the objective function corresponding to the sequence of points generated by the M-penalty function method decreases gradually.

    Lemma 5.2

    Let x(σ)be the optimal solution to optimization problem(PG),and let δ=G(x(σ)).Then x(σ)is also the optimal solution to the constrained optimization problem

    Proof

    For any x that satis fies equation(5.8),we have

    According to equation(5.9),for any x that satis fies equation(5.8),we have

    Therefore,x(σ)is also the optimal solution to the constrained optimization problem(5.7)–(5.8).The conclusion is established.

    Through Lemma 5.1 and Lemma 5.2,we can get the convergence theorem of the M-penalty function method.

    Theorem 5.3

    Assume that the error factor ε in the M-penalty function method satis fies the condition that

    Then,the algorithm must be terminated within finite iterations.

    As can be seen from equation(5.11),there isˉx such that

    As shown by equation(5.2),we have

    Let σ→+∞.Then

    This conclusion contradicts equation(5.14).Therefore,Theorem 5.3 is true,and the conclusion is established.

    According to Theorem 5.3,if the semi-in finite programming problem(SIP)has a feasible solution,for any given ε>0,the M-penalty function method will be terminated at the optimal solution to the optimization problem(5.7)–(5.8)and δ≤ε.

    If the algorithm cannot be terminated within finite iterations,we have the following conclusions:

    Theorem 5.4

    If the M-penalty function method will not be terminated within finite iterations,then for any error factor ε,we have

    and

    Therefore,the equation(5.18)is established.

    and the constraint condition(5.20)is satis fied.Therefore,we have

    It can be seen from Lemma 5.1 that f(x)monotonically approaches to f(x).Therefore,it can be known from equation(5.22)that,for any sufficiently large k,

    It can be known from equation(5.23)and equation(5.24)that

    From the above two theorems,the following corollary can be directly obtained:

    Then,we have

    6 Numerical Example

    The expected pro fit of the agent is

    Assuming that the agent’s disagreement pro fit uis 0,the optimal incentive mechanism for this principal-agent problem is the solution to the following semi-in finite programming problem:

    Next,we use the M-penalty function method in Section 4 to solve the semi-in finite programming problem(6.3)–(6.6).

    The constraint violation function is

    The optimization problem(PG)is

    Since this is a standard linear contract,we can use traditional method(e.g.,in[11,12])to verify this result.

    First,we modify the equation(6.1),and we get

    7 Conclusions

    This paper mainly designs an iterative algorithm to find the optimal incentive mechanism for the principal-agent problem under moral hazard where the number of agent action profi les is in finite,and where there are an in finite number of results that can be observed by the principal.The main characteristic of this principal-agent problem is that the number of constraints is in finite.Given the form of payment vector function designed by the principal,the principal-agent problem can be transformed into a semi-in finite programming problem.A exterior penalty function method(called the M-penalty function method)for solving semi-in finite programming problems is proposed,and its convergence is illustrated.By using the M-penalty function method designed in this paper,we can obtain the optimal incentive mechanism for the principal-agent problem with an in finite number of incentive-compatibility constraints under moral hazard.

    In this paper,the in finite number of constraints are mainly caused by the in finite number of agent action pro files.In addition,if the set of agents is assumed to be continuous,the number of individual rational constraints will be in finite,which will also make the incentive mechanism design problem a semi-in finite programming problem.At this point,the M-penalty function method designed in this paper for solving semi-in finite programming problems is also applicable for solving the incentive mechanism design problem.

    It should be noted that in this paper,we assume that the form of the principal’s payment function to the agent is given,but it contains some uncertain parameters.The optimal solution to the semi-in finite programming problem are the optimal parameters of the payofffunction to the incentive mechanism design problem.In a more general case,the form of the payment function may not be known.Therefore,the incentive mechanism design problem is a variational problem with an in finite number of constraints.Compared with the algorithm for solving semiin finite programming problems,the method for solving this problem is more complex.

    Acknowledgements

    We would like to express our gratitude to all those who helped us during the writing and revising of this paper.In particular,we are very grateful to the two anonymous reviewers for their comments,which were of great signi ficance.

    猜你喜歡
    劉佳
    ON THE GRAPHS OF PRODUCTS OF CONTINUOUS FUNCTIONS AND FRACTAL DIMENSIONS?
    太陽的歌
    Relativistic Hartree–Fock model and its recent progress on the description of nuclear structure*
    江蘇省宜興市陶城實(shí)驗(yàn)小學(xué)劉佳
    遇見,又如何
    照相機(jī)(2022年12期)2022-02-09 09:13:54
    Interaction induced non-reciprocal three-level quantum transport?
    劉佳美術(shù)作品
    Principles and Teaching Application of Suggestopedia’s 6 Technical Characteristics
    He’s just been to the zoo.
    On the Translation of English Film Titles from the Perspective of Reception Aesthetics
    久久午夜综合久久蜜桃| 欧美精品一区二区免费开放| 久久精品人人爽人人爽视色| 久久久国产精品麻豆| 美女脱内裤让男人舔精品视频| av.在线天堂| 精品一区二区三卡| 99热这里只有精品一区| 中文字幕免费在线视频6| 国产一级毛片在线| 久久99热6这里只有精品| 一级黄片播放器| 久热久热在线精品观看| 亚洲图色成人| 亚洲av成人精品一区久久| 国产欧美日韩一区二区三区在线 | 精品久久久噜噜| 99久久人妻综合| 少妇精品久久久久久久| 亚洲第一av免费看| 日韩精品免费视频一区二区三区 | 国产精品一区www在线观看| 99九九在线精品视频| 久久久久久久精品精品| 亚洲欧美日韩卡通动漫| 国产成人a∨麻豆精品| 亚洲av免费高清在线观看| 一本久久精品| 我要看黄色一级片免费的| 最近手机中文字幕大全| 在线免费观看不下载黄p国产| 国内精品宾馆在线| 夜夜骑夜夜射夜夜干| 男人爽女人下面视频在线观看| 国产乱来视频区| 超碰97精品在线观看| 久久精品久久久久久久性| 晚上一个人看的免费电影| 欧美日韩av久久| a级毛色黄片| 性色av一级| 日日啪夜夜爽| 日日啪夜夜爽| 亚洲第一av免费看| 国产乱人偷精品视频| 久久久久久久亚洲中文字幕| 日韩电影二区| 成人午夜精彩视频在线观看| 成年女人在线观看亚洲视频| 亚洲性久久影院| 亚洲经典国产精华液单| 日本黄大片高清| 亚洲精品自拍成人| 一区二区三区精品91| 最新的欧美精品一区二区| 26uuu在线亚洲综合色| 色吧在线观看| 九九久久精品国产亚洲av麻豆| 日韩三级伦理在线观看| 大片电影免费在线观看免费| 91午夜精品亚洲一区二区三区| 母亲3免费完整高清在线观看 | 多毛熟女@视频| 狂野欧美激情性xxxx在线观看| 黑人欧美特级aaaaaa片| 最近中文字幕高清免费大全6| 最近手机中文字幕大全| 一边摸一边做爽爽视频免费| 自线自在国产av| 新久久久久国产一级毛片| 少妇被粗大的猛进出69影院 | 亚洲精品国产av蜜桃| av在线app专区| 午夜视频国产福利| 久久av网站| 国产男女超爽视频在线观看| 91久久精品电影网| 如日韩欧美国产精品一区二区三区 | 国产乱人偷精品视频| 午夜影院在线不卡| 久热这里只有精品99| 国产日韩欧美在线精品| 亚洲欧美色中文字幕在线| 18+在线观看网站| 少妇高潮的动态图| 人成视频在线观看免费观看| av免费观看日本| 日韩中文字幕视频在线看片| 黄片播放在线免费| 精品久久久精品久久久| 最近的中文字幕免费完整| 欧美 日韩 精品 国产| 欧美一级a爱片免费观看看| 热99久久久久精品小说推荐| www.av在线官网国产| 国产永久视频网站| 亚洲精品日韩av片在线观看| 日韩三级伦理在线观看| 亚洲av电影在线观看一区二区三区| 18禁在线无遮挡免费观看视频| 99热6这里只有精品| 夜夜爽夜夜爽视频| 久久久久久久久久成人| 夜夜看夜夜爽夜夜摸| 少妇的逼水好多| 国产午夜精品一二区理论片| 大陆偷拍与自拍| 中文欧美无线码| 午夜激情久久久久久久| 啦啦啦中文免费视频观看日本| 亚洲精品国产av成人精品| 下体分泌物呈黄色| 久久久久久人妻| 国产免费又黄又爽又色| 九草在线视频观看| 午夜福利网站1000一区二区三区| 搡女人真爽免费视频火全软件| 夫妻午夜视频| 午夜福利影视在线免费观看| 国产日韩欧美视频二区| 亚洲国产毛片av蜜桃av| 亚洲精品一二三| 日韩熟女老妇一区二区性免费视频| 母亲3免费完整高清在线观看 | 亚洲精品成人av观看孕妇| 国产在线一区二区三区精| 国产又色又爽无遮挡免| 国产在线视频一区二区| 91午夜精品亚洲一区二区三区| 国产午夜精品久久久久久一区二区三区| 亚洲精品乱久久久久久| 久久久久久久精品精品| 下体分泌物呈黄色| 久久精品国产a三级三级三级| 赤兔流量卡办理| 五月开心婷婷网| 狠狠婷婷综合久久久久久88av| 亚洲av中文av极速乱| 日韩精品有码人妻一区| 精品久久国产蜜桃| 好男人视频免费观看在线| 狂野欧美白嫩少妇大欣赏| 狠狠精品人妻久久久久久综合| 国产av一区二区精品久久| 九九在线视频观看精品| 国产午夜精品久久久久久一区二区三区| 国产精品国产三级国产av玫瑰| 欧美性感艳星| 老司机亚洲免费影院| 18禁动态无遮挡网站| 亚洲精品一二三| 两个人的视频大全免费| 看非洲黑人一级黄片| 欧美少妇被猛烈插入视频| 精品久久久精品久久久| 国产69精品久久久久777片| 国产一区有黄有色的免费视频| 国产在线视频一区二区| 日韩一本色道免费dvd| 少妇人妻 视频| 人人澡人人妻人| 国产精品熟女久久久久浪| 乱人伦中国视频| 十八禁网站网址无遮挡| 亚洲精品亚洲一区二区| 成年美女黄网站色视频大全免费 | 好男人视频免费观看在线| 男人爽女人下面视频在线观看| 久久精品国产鲁丝片午夜精品| 赤兔流量卡办理| 夫妻性生交免费视频一级片| 黄片无遮挡物在线观看| 国产男人的电影天堂91| 在线观看免费高清a一片| 国产精品国产三级国产专区5o| 久久久久久久亚洲中文字幕| 不卡视频在线观看欧美| 亚洲av国产av综合av卡| 国产男人的电影天堂91| 精品少妇内射三级| 国产一级毛片在线| 18+在线观看网站| 99视频精品全部免费 在线| 一区在线观看完整版| 久久久久人妻精品一区果冻| 卡戴珊不雅视频在线播放| 一级黄片播放器| 亚洲丝袜综合中文字幕| 久久久久久久久久人人人人人人| 国产午夜精品久久久久久一区二区三区| 日本猛色少妇xxxxx猛交久久| 精品国产国语对白av| 婷婷成人精品国产| 80岁老熟妇乱子伦牲交| 国产精品一区二区在线观看99| 亚洲美女黄色视频免费看| 97在线视频观看| 日本欧美国产在线视频| 嘟嘟电影网在线观看| 色5月婷婷丁香| 波野结衣二区三区在线| 欧美日韩视频精品一区| 丰满乱子伦码专区| 日本欧美国产在线视频| 特大巨黑吊av在线直播| 一本一本久久a久久精品综合妖精 国产伦在线观看视频一区 | 一级毛片 在线播放| 天堂8中文在线网| 三级国产精品片| 如何舔出高潮| av专区在线播放| 熟女电影av网| 久久精品久久久久久噜噜老黄| 99九九线精品视频在线观看视频| 国产成人aa在线观看| 日本91视频免费播放| .国产精品久久| 久久久亚洲精品成人影院| 最新中文字幕久久久久| av一本久久久久| 一本久久精品| 久久精品国产a三级三级三级| 午夜视频国产福利| 男的添女的下面高潮视频| 各种免费的搞黄视频| 午夜老司机福利剧场| 亚洲精品乱码久久久v下载方式| 人妻制服诱惑在线中文字幕| 街头女战士在线观看网站| av一本久久久久| 少妇人妻精品综合一区二区| 美女中出高潮动态图| 你懂的网址亚洲精品在线观看| 亚洲综合色惰| 欧美日韩国产mv在线观看视频| 亚洲av综合色区一区| 欧美人与善性xxx| 69精品国产乱码久久久| 国产日韩一区二区三区精品不卡 | 免费黄频网站在线观看国产| 色网站视频免费| 亚洲国产av新网站| 精品卡一卡二卡四卡免费| 亚洲不卡免费看| 伊人久久精品亚洲午夜| 亚洲美女黄色视频免费看| 免费大片18禁| 在线亚洲精品国产二区图片欧美 | 亚洲内射少妇av| 在线 av 中文字幕| 在线观看免费日韩欧美大片 | 免费黄色在线免费观看| 午夜免费男女啪啪视频观看| 纵有疾风起免费观看全集完整版| 久久精品国产自在天天线| 校园人妻丝袜中文字幕| 嘟嘟电影网在线观看| 看免费成人av毛片| 国产极品粉嫩免费观看在线 | 97在线人人人人妻| 国产免费现黄频在线看| 3wmmmm亚洲av在线观看| 制服人妻中文乱码| 天堂俺去俺来也www色官网| 免费看不卡的av| 国产一区二区三区av在线| 校园人妻丝袜中文字幕| 特大巨黑吊av在线直播| 成年人免费黄色播放视频| 国产精品一国产av| 欧美另类一区| 久久久久国产精品人妻一区二区| 中文字幕久久专区| 欧美三级亚洲精品| 精品午夜福利在线看| 国产片特级美女逼逼视频| 久久精品熟女亚洲av麻豆精品| 久久免费观看电影| 久久午夜综合久久蜜桃| 国产乱来视频区| 熟女av电影| av在线老鸭窝| 欧美日本中文国产一区发布| 亚洲国产av影院在线观看| 中文字幕精品免费在线观看视频 | 欧美97在线视频| 肉色欧美久久久久久久蜜桃| 欧美精品国产亚洲| 在线 av 中文字幕| 考比视频在线观看| 日韩av在线免费看完整版不卡| 美女xxoo啪啪120秒动态图| 美女视频免费永久观看网站| 欧美日本中文国产一区发布| 国产国语露脸激情在线看| 欧美日韩国产mv在线观看视频| 亚洲欧美色中文字幕在线| av网站免费在线观看视频| 有码 亚洲区| 少妇人妻精品综合一区二区| 亚洲性久久影院| 欧美精品一区二区免费开放| 啦啦啦中文免费视频观看日本| 亚洲不卡免费看| 精品熟女少妇av免费看| 国产国拍精品亚洲av在线观看| 国产 精品1| 午夜福利,免费看| 国产精品秋霞免费鲁丝片| 高清欧美精品videossex| 国产日韩欧美在线精品| 国产免费一级a男人的天堂| 国产亚洲最大av| 婷婷成人精品国产| 成人亚洲精品一区在线观看| 草草在线视频免费看| 我要看黄色一级片免费的| 另类亚洲欧美激情| 亚洲天堂av无毛| 国产免费一区二区三区四区乱码| 日本91视频免费播放| 日本-黄色视频高清免费观看| 在线观看三级黄色| 国产 精品1| 欧美+日韩+精品| 亚洲欧洲国产日韩| 日韩免费高清中文字幕av| 日韩电影二区| av播播在线观看一区| 91精品伊人久久大香线蕉| 乱码一卡2卡4卡精品| 男女免费视频国产| 日韩强制内射视频| 中文字幕制服av| 国产欧美亚洲国产| 我的女老师完整版在线观看| 国语对白做爰xxxⅹ性视频网站| 赤兔流量卡办理| 精品国产一区二区久久| 日本欧美国产在线视频| 久久av网站| 亚洲av.av天堂| 亚洲欧美一区二区三区国产| 国产精品.久久久| 免费播放大片免费观看视频在线观看| 狂野欧美激情性xxxx在线观看| 国产欧美日韩一区二区三区在线 | 一本一本综合久久| 欧美激情 高清一区二区三区| www.色视频.com| 97精品久久久久久久久久精品| 黄色配什么色好看| 一本色道久久久久久精品综合| 国产精品嫩草影院av在线观看| 91精品一卡2卡3卡4卡| 亚洲国产av新网站| 日韩 亚洲 欧美在线| 黄色一级大片看看| 国产亚洲欧美精品永久| 另类精品久久| 一级毛片黄色毛片免费观看视频| 97在线视频观看| 亚洲人成网站在线播| 中文天堂在线官网| av在线播放精品| 日韩视频在线欧美| 观看av在线不卡| 简卡轻食公司| 亚洲中文av在线| 多毛熟女@视频| 久久久久久久久久久丰满| 母亲3免费完整高清在线观看 | 国产毛片在线视频| 欧美日韩成人在线一区二区| 成人无遮挡网站| 日韩在线高清观看一区二区三区| 极品人妻少妇av视频| 精品一区二区三区视频在线| 国产欧美亚洲国产| 97在线视频观看| 老熟女久久久| 男女啪啪激烈高潮av片| 国产免费一区二区三区四区乱码| 热99久久久久精品小说推荐| 极品人妻少妇av视频| 亚洲精品成人av观看孕妇| 99视频精品全部免费 在线| 日日撸夜夜添| 国产69精品久久久久777片| 亚洲国产精品一区二区三区在线| 久久影院123| 乱码一卡2卡4卡精品| 91aial.com中文字幕在线观看| 丝袜美足系列| 热99久久久久精品小说推荐| 看非洲黑人一级黄片| 狂野欧美白嫩少妇大欣赏| 人人妻人人澡人人爽人人夜夜| 久久ye,这里只有精品| 国产亚洲精品第一综合不卡 | 色婷婷av一区二区三区视频| 秋霞在线观看毛片| 高清午夜精品一区二区三区| 在线观看国产h片| 日本欧美视频一区| 制服诱惑二区| 久久久久精品性色| 亚洲精品美女久久av网站| 有码 亚洲区| 亚洲精品国产av蜜桃| 亚洲精品久久久久久婷婷小说| 一级毛片我不卡| 一个人看视频在线观看www免费| 在线观看三级黄色| 99热这里只有精品一区| 蜜臀久久99精品久久宅男| 久久午夜福利片| 欧美激情国产日韩精品一区| 欧美精品国产亚洲| 18禁观看日本| 18在线观看网站| 一本一本综合久久| 成人免费观看视频高清| 久久ye,这里只有精品| 久久久a久久爽久久v久久| 欧美日韩精品成人综合77777| 国产片特级美女逼逼视频| 国产在线视频一区二区| 亚洲综合色惰| 18禁动态无遮挡网站| 精品人妻偷拍中文字幕| av不卡在线播放| 国产在线视频一区二区| 伦理电影免费视频| 久久久久久久亚洲中文字幕| 亚洲av中文av极速乱| 国产亚洲精品久久久com| 大香蕉久久网| 搡女人真爽免费视频火全软件| 黄片播放在线免费| 亚洲精品日本国产第一区| 久久精品国产亚洲网站| 99热这里只有精品一区| 日韩精品免费视频一区二区三区 | 综合色丁香网| 国产精品久久久久成人av| 久久国产精品男人的天堂亚洲 | 欧美日韩视频精品一区| 欧美日韩av久久| 少妇被粗大的猛进出69影院 | 飞空精品影院首页| 欧美三级亚洲精品| 午夜福利视频精品| 久久亚洲国产成人精品v| 国产国拍精品亚洲av在线观看| 精品亚洲乱码少妇综合久久| 国产精品一国产av| 最近2019中文字幕mv第一页| 欧美日本中文国产一区发布| 国国产精品蜜臀av免费| 久久久a久久爽久久v久久| 免费播放大片免费观看视频在线观看| 三级国产精品片| 欧美精品一区二区大全| 精品国产国语对白av| 久久鲁丝午夜福利片| 18禁在线播放成人免费| 一级毛片aaaaaa免费看小| 女人精品久久久久毛片| 狠狠精品人妻久久久久久综合| 大香蕉久久成人网| 免费黄色在线免费观看| 中文天堂在线官网| 美女cb高潮喷水在线观看| 一个人看视频在线观看www免费| 亚洲av男天堂| 天堂俺去俺来也www色官网| 亚洲精品aⅴ在线观看| 制服丝袜香蕉在线| 国产成人精品久久久久久| 在线观看免费高清a一片| 欧美+日韩+精品| 男女高潮啪啪啪动态图| 伊人久久国产一区二区| 在线观看免费日韩欧美大片 | 亚洲av二区三区四区| 欧美日韩视频高清一区二区三区二| 午夜免费观看性视频| 啦啦啦在线观看免费高清www| 久久久国产欧美日韩av| 亚洲怡红院男人天堂| 成人二区视频| 视频在线观看一区二区三区| 在现免费观看毛片| 国产av精品麻豆| 热re99久久国产66热| 看免费成人av毛片| 夫妻午夜视频| 一级毛片aaaaaa免费看小| 一本色道久久久久久精品综合| 国产乱人偷精品视频| 欧美bdsm另类| 另类精品久久| av在线观看视频网站免费| 国产成人aa在线观看| 欧美三级亚洲精品| 精品国产国语对白av| 亚洲精华国产精华液的使用体验| 丰满乱子伦码专区| 97在线人人人人妻| 色婷婷av一区二区三区视频| av线在线观看网站| 欧美人与性动交α欧美精品济南到 | 亚洲情色 制服丝袜| 蜜桃在线观看..| 夫妻午夜视频| 久热这里只有精品99| 2018国产大陆天天弄谢| 又黄又爽又刺激的免费视频.| 亚洲av男天堂| 国产乱人偷精品视频| 看非洲黑人一级黄片| 精品99又大又爽又粗少妇毛片| 3wmmmm亚洲av在线观看| 亚洲av不卡在线观看| 欧美bdsm另类| 欧美 日韩 精品 国产| 制服诱惑二区| 欧美丝袜亚洲另类| 免费人妻精品一区二区三区视频| 麻豆乱淫一区二区| 美女大奶头黄色视频| 精品一区二区三区视频在线| 成人毛片60女人毛片免费| 高清欧美精品videossex| 夜夜看夜夜爽夜夜摸| 亚洲欧美清纯卡通| 狠狠婷婷综合久久久久久88av| 一级毛片我不卡| 精品熟女少妇av免费看| 亚洲精品国产av成人精品| 99国产综合亚洲精品| 亚洲国产最新在线播放| 九九在线视频观看精品| 一本大道久久a久久精品| 永久网站在线| a级毛片黄视频| 男的添女的下面高潮视频| 丝袜脚勾引网站| 男女边摸边吃奶| 男人爽女人下面视频在线观看| 欧美精品一区二区大全| 婷婷色综合www| 熟女av电影| 精品一区在线观看国产| 97超碰精品成人国产| 日韩强制内射视频| 18+在线观看网站| 国产精品一区二区在线不卡| 国产不卡av网站在线观看| 亚洲欧洲精品一区二区精品久久久 | av国产精品久久久久影院| a级片在线免费高清观看视频| √禁漫天堂资源中文www| 午夜视频国产福利| 一级毛片电影观看| 少妇被粗大的猛进出69影院 | 欧美bdsm另类| 国产深夜福利视频在线观看| 国产成人91sexporn| 午夜福利视频在线观看免费| 爱豆传媒免费全集在线观看| 日韩视频在线欧美| 欧美少妇被猛烈插入视频| 国产不卡av网站在线观看| 久久ye,这里只有精品| 99re6热这里在线精品视频| 欧美日韩视频高清一区二区三区二| a级片在线免费高清观看视频| 久久久久久久精品精品| 国产成人精品福利久久| 大片免费播放器 马上看| 欧美日韩一区二区视频在线观看视频在线| 人妻人人澡人人爽人人| 日本-黄色视频高清免费观看| 欧美日韩av久久| av在线观看视频网站免费| 纯流量卡能插随身wifi吗| 国产一区二区在线观看av| 国产一区有黄有色的免费视频| 色哟哟·www| 久久亚洲国产成人精品v| 久久97久久精品| 国产成人精品在线电影| 美女cb高潮喷水在线观看| 91久久精品电影网| 久久99热6这里只有精品| 久久久久精品久久久久真实原创| 久久精品国产a三级三级三级| 日韩大片免费观看网站| 成人午夜精彩视频在线观看| 天堂8中文在线网| 欧美最新免费一区二区三区| 国产精品嫩草影院av在线观看| 国产永久视频网站| 亚洲av综合色区一区| 观看av在线不卡| 在线观看人妻少妇| 有码 亚洲区| 伊人久久精品亚洲午夜| 王馨瑶露胸无遮挡在线观看| 交换朋友夫妻互换小说| 99热全是精品| 观看av在线不卡| 国产高清三级在线|