• <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
    免费看日本二区| 亚洲最大成人手机在线| 亚洲综合精品二区| 国产又黄又爽又无遮挡在线| 亚洲电影在线观看av| 免费观看a级毛片全部| 高清av免费在线| 69人妻影院| 免费大片18禁| 亚洲国产精品久久男人天堂| 欧美zozozo另类| 亚洲内射少妇av| 亚洲自偷自拍三级| 中文欧美无线码| 男人舔奶头视频| 亚洲成人精品中文字幕电影| 村上凉子中文字幕在线| 日韩大片免费观看网站 | 亚洲最大成人av| 春色校园在线视频观看| 成人美女网站在线观看视频| 两个人的视频大全免费| 国产亚洲91精品色在线| 久久国内精品自在自线图片| 中文字幕人妻熟人妻熟丝袜美| 18+在线观看网站| 男的添女的下面高潮视频| 黑人高潮一二区| 午夜福利在线观看吧| 老司机福利观看| 高清视频免费观看一区二区 | av又黄又爽大尺度在线免费看 | 欧美潮喷喷水| 亚洲av中文av极速乱| 国产国拍精品亚洲av在线观看| 久久6这里有精品| 一级av片app| 国产精品不卡视频一区二区| 亚洲成人精品中文字幕电影| 两个人视频免费观看高清| 国产69精品久久久久777片| 人人妻人人看人人澡| 欧美精品一区二区大全| 日韩强制内射视频| 美女高潮的动态| 国产精品人妻久久久久久| 午夜福利高清视频| 中文字幕熟女人妻在线| 精品无人区乱码1区二区| a级一级毛片免费在线观看| 亚洲中文字幕一区二区三区有码在线看| 你懂的网址亚洲精品在线观看 | 夜夜爽夜夜爽视频| 亚洲av日韩在线播放| 国产精品人妻久久久影院| 性色avwww在线观看| 99久久人妻综合| 国产精品国产高清国产av| av免费观看日本| 成人午夜高清在线视频| 精品人妻一区二区三区麻豆| 亚洲中文字幕一区二区三区有码在线看| www日本黄色视频网| 一卡2卡三卡四卡精品乱码亚洲| 久久精品久久精品一区二区三区| 在线免费十八禁| 国产午夜精品久久久久久一区二区三区| 日韩中字成人| 色噜噜av男人的天堂激情| 亚洲一级一片aⅴ在线观看| 青青草视频在线视频观看| 日韩高清综合在线| 亚洲国产成人一精品久久久| 成人综合一区亚洲| 亚洲精品影视一区二区三区av| 99久久精品国产国产毛片| 日本黄色片子视频| 国产免费又黄又爽又色| 亚洲色图av天堂| 久久久午夜欧美精品| 国产欧美另类精品又又久久亚洲欧美| 国产精品日韩av在线免费观看| 国产精华一区二区三区| 2021少妇久久久久久久久久久| 久久久久性生活片| 国产精品久久久久久久久免| 中文在线观看免费www的网站| 日韩欧美国产在线观看| .国产精品久久| 成人三级黄色视频| 噜噜噜噜噜久久久久久91| 国语对白做爰xxxⅹ性视频网站| 国产69精品久久久久777片| 欧美成人午夜免费资源| 精品久久久久久久人妻蜜臀av| 免费无遮挡裸体视频| 只有这里有精品99| 国产精品人妻久久久久久| 黄片wwwwww| 午夜福利高清视频| 久久亚洲精品不卡| 亚洲精品色激情综合| 免费搜索国产男女视频| 成年免费大片在线观看| 美女被艹到高潮喷水动态| 午夜免费男女啪啪视频观看| 女人久久www免费人成看片 | 国产一区亚洲一区在线观看| av又黄又爽大尺度在线免费看 | 一区二区三区四区激情视频| 午夜老司机福利剧场| 免费黄色在线免费观看| 亚洲不卡免费看| 少妇的逼水好多| 亚洲国产精品合色在线| 九九在线视频观看精品| 亚洲精品乱久久久久久| 少妇人妻精品综合一区二区| 成人高潮视频无遮挡免费网站| 一个人看的www免费观看视频| 亚洲av不卡在线观看| 女人久久www免费人成看片 | 亚洲丝袜综合中文字幕| 97人妻精品一区二区三区麻豆| 我要搜黄色片| 亚洲av电影不卡..在线观看| 欧美三级亚洲精品| 成人漫画全彩无遮挡| 国产精品久久久久久久久免| 国产成人精品久久久久久| 99在线人妻在线中文字幕| 最近的中文字幕免费完整| 国产精品人妻久久久久久| av国产免费在线观看| 日日干狠狠操夜夜爽| 午夜精品在线福利| 欧美日韩国产亚洲二区| 99热网站在线观看| 水蜜桃什么品种好| 两性午夜刺激爽爽歪歪视频在线观看| 美女大奶头视频| 日日摸夜夜添夜夜爱| 精品人妻一区二区三区麻豆| 国产极品天堂在线| 成人无遮挡网站| 国产色婷婷99| 天天躁夜夜躁狠狠久久av| 久久精品夜夜夜夜夜久久蜜豆| 中国国产av一级| 国产精品久久久久久av不卡| av黄色大香蕉| 国产乱人视频| 丝袜美腿在线中文| 日韩一区二区视频免费看| 亚洲精品aⅴ在线观看| 最近视频中文字幕2019在线8| 桃色一区二区三区在线观看| 在线a可以看的网站| 亚洲av成人av| 久久久国产成人精品二区| 国产精品.久久久| 日日撸夜夜添| 亚洲欧美一区二区三区国产| 久久这里有精品视频免费| 深夜a级毛片| 99久久无色码亚洲精品果冻| 日韩视频在线欧美| 亚洲欧洲国产日韩| 国产精品美女特级片免费视频播放器| 欧美高清成人免费视频www| 亚洲成人中文字幕在线播放| 中文乱码字字幕精品一区二区三区 | 久久精品综合一区二区三区| 亚洲,欧美,日韩| 成人高潮视频无遮挡免费网站| 久久久久久久久久成人| 久久久久国产网址| 国产精品一区二区性色av| 国产伦精品一区二区三区视频9| 国国产精品蜜臀av免费| 美女cb高潮喷水在线观看| 国产乱来视频区| 亚洲最大成人中文| 精品久久久久久久久av| 麻豆av噜噜一区二区三区| 亚洲色图av天堂| 亚洲内射少妇av| 女的被弄到高潮叫床怎么办| 22中文网久久字幕| 插阴视频在线观看视频| 精品少妇黑人巨大在线播放 | 国产一级毛片在线| 免费观看在线日韩| 国产免费视频播放在线视频 | 大话2 男鬼变身卡| av视频在线观看入口| 欧美一级a爱片免费观看看| 国产亚洲5aaaaa淫片| 亚洲高清免费不卡视频| 尾随美女入室| 高清av免费在线| 晚上一个人看的免费电影| 亚洲电影在线观看av| 偷拍熟女少妇极品色| 国产乱人偷精品视频| 天天躁夜夜躁狠狠久久av| 国产精品久久久久久av不卡| 久久久久久久久久成人| 丰满乱子伦码专区| 级片在线观看| 亚洲欧美日韩无卡精品| av女优亚洲男人天堂| 日韩,欧美,国产一区二区三区 | 久久久久久久久中文| 少妇丰满av| 久久久久网色| 看免费成人av毛片| av在线天堂中文字幕| 高清在线视频一区二区三区 | 亚洲va在线va天堂va国产| 亚洲国产日韩欧美精品在线观看| 亚洲欧洲国产日韩| 国产一级毛片在线| 国产极品天堂在线| 乱系列少妇在线播放| www.av在线官网国产| 久久精品影院6| 国产v大片淫在线免费观看| 国产免费福利视频在线观看| 日日干狠狠操夜夜爽| av专区在线播放| 在线观看一区二区三区| 国产精品一二三区在线看| 熟女电影av网| 日韩av在线大香蕉| 一个人看视频在线观看www免费| 99热网站在线观看| 成年av动漫网址| 精品人妻偷拍中文字幕| 麻豆国产97在线/欧美| 亚洲av电影不卡..在线观看| 日本与韩国留学比较| 毛片女人毛片| 欧美区成人在线视频| 乱人视频在线观看| 精品久久久噜噜| 哪个播放器可以免费观看大片| 国产高清有码在线观看视频| 亚洲精华国产精华液的使用体验| 午夜日本视频在线| 99久久精品国产国产毛片| 久久久久国产网址| 中国国产av一级| 午夜精品一区二区三区免费看| 久久99蜜桃精品久久| 99在线视频只有这里精品首页| 日韩视频在线欧美| 久久久久久久久久黄片| 亚洲精品乱码久久久v下载方式| 身体一侧抽搐| 好男人视频免费观看在线| 91精品一卡2卡3卡4卡| 丰满人妻一区二区三区视频av| 韩国av在线不卡| 99视频精品全部免费 在线| or卡值多少钱| 99热精品在线国产| 亚洲国产欧洲综合997久久,| 天堂影院成人在线观看| 搡女人真爽免费视频火全软件| 国产精品美女特级片免费视频播放器| 国产精品乱码一区二三区的特点| 国产av不卡久久| 久久精品国产鲁丝片午夜精品| 久久久亚洲精品成人影院| 男插女下体视频免费在线播放| 高清视频免费观看一区二区 | 国产片特级美女逼逼视频| 九草在线视频观看| 18禁在线播放成人免费| 欧美日本视频| 国产黄a三级三级三级人| av线在线观看网站| 少妇被粗大猛烈的视频| 免费看av在线观看网站| 日韩人妻高清精品专区| 18禁裸乳无遮挡免费网站照片| 中文字幕人妻熟人妻熟丝袜美| 边亲边吃奶的免费视频| 99久久无色码亚洲精品果冻| 天天一区二区日本电影三级| av在线天堂中文字幕| av专区在线播放| 国产精品一及| 简卡轻食公司| 久久久a久久爽久久v久久| 国产精品人妻久久久久久| 午夜久久久久精精品| 亚洲欧美精品自产自拍| 婷婷色综合大香蕉| 视频中文字幕在线观看| 男人和女人高潮做爰伦理| 十八禁国产超污无遮挡网站| 久久精品91蜜桃| 亚洲国产欧洲综合997久久,| 国产淫片久久久久久久久| 大又大粗又爽又黄少妇毛片口| 久久精品熟女亚洲av麻豆精品 | 亚洲最大成人av| 精品一区二区三区视频在线| 成人欧美大片| 在线观看av片永久免费下载| 国产成人免费观看mmmm| 日本免费a在线| 蜜桃久久精品国产亚洲av| 一个人看的www免费观看视频| 最新中文字幕久久久久| 精品免费久久久久久久清纯| 成人av在线播放网站| 欧美潮喷喷水| 神马国产精品三级电影在线观看| 亚洲美女视频黄频| 亚洲va在线va天堂va国产| 国产免费又黄又爽又色| 亚洲av中文av极速乱| 欧美zozozo另类| 成人综合一区亚洲| 国产极品天堂在线| 在线免费十八禁| 夫妻性生交免费视频一级片| 天天躁日日操中文字幕| 三级经典国产精品| 久久韩国三级中文字幕| 高清视频免费观看一区二区 | 亚洲av电影在线观看一区二区三区 | 国产精品三级大全| 国产淫语在线视频| 久久精品影院6| 亚洲人成网站高清观看| 欧美精品国产亚洲| 久久这里有精品视频免费| 国产av不卡久久| 国产成人精品一,二区| 国产精品国产三级国产av玫瑰| 久久久久久久国产电影| 亚洲美女视频黄频| 天堂影院成人在线观看| 美女黄网站色视频| 国产色爽女视频免费观看| 日本wwww免费看| 亚洲国产成人一精品久久久| 亚洲欧美成人综合另类久久久 | 亚洲伊人久久精品综合 | 久久热精品热| 精品无人区乱码1区二区| 久久久亚洲精品成人影院| 能在线免费观看的黄片| 中文字幕av成人在线电影| 国产精品嫩草影院av在线观看| 国产精品一区二区三区四区免费观看| 乱人视频在线观看| 内射极品少妇av片p| 少妇裸体淫交视频免费看高清| av专区在线播放| 有码 亚洲区| 国产伦精品一区二区三区四那| 国产精品人妻久久久影院| 99热这里只有是精品50| 国产精品一区二区三区四区久久| 老女人水多毛片| 久久久久久伊人网av| 久久精品国产亚洲av天美| 日本wwww免费看| 亚洲国产精品久久男人天堂| 免费播放大片免费观看视频在线观看 | 日韩在线高清观看一区二区三区| 成人特级av手机在线观看| 国产午夜精品一二区理论片| 亚洲最大成人av| 中文资源天堂在线| 超碰av人人做人人爽久久| 久久亚洲国产成人精品v| 18禁在线播放成人免费| 九草在线视频观看| 一边摸一边抽搐一进一小说| 国产一级毛片七仙女欲春2| 国产精华一区二区三区| 亚洲精品aⅴ在线观看| 精品人妻熟女av久视频| 神马国产精品三级电影在线观看| 国语对白做爰xxxⅹ性视频网站| 亚洲人成网站在线观看播放| 国产老妇女一区| 久久精品91蜜桃| 淫秽高清视频在线观看| 国产91av在线免费观看| 国产极品天堂在线| 久久久色成人| 精品少妇黑人巨大在线播放 | 成人特级av手机在线观看| 中文字幕亚洲精品专区| 欧美最新免费一区二区三区| 国产高清国产精品国产三级 | 国产伦一二天堂av在线观看| 久久久久久久久久黄片| 成人二区视频| 久久欧美精品欧美久久欧美| 国产一级毛片在线| 菩萨蛮人人尽说江南好唐韦庄 | 亚洲精品日韩在线中文字幕| 久久久国产成人精品二区| av免费观看日本| 内地一区二区视频在线| 99久久中文字幕三级久久日本| 日韩欧美国产在线观看| 久久精品影院6| 国产午夜精品一二区理论片| 观看美女的网站| 国产成人freesex在线| 久久6这里有精品| 在线观看66精品国产| 日韩欧美精品免费久久| 欧美一区二区精品小视频在线| 国产人妻一区二区三区在| 欧美又色又爽又黄视频| 婷婷色av中文字幕| 国产伦精品一区二区三区视频9| 色尼玛亚洲综合影院| 一本—道久久a久久精品蜜桃钙片 精品乱码久久久久久99久播 | 91久久精品国产一区二区成人| kizo精华| 欧美激情在线99| 国产老妇女一区| 欧美色视频一区免费| 欧美一区二区精品小视频在线| 国产美女午夜福利| 日本五十路高清| 18禁动态无遮挡网站| 精品国产三级普通话版| 国产成人精品一,二区| 久久精品综合一区二区三区| 国产探花极品一区二区| 亚洲五月天丁香| 美女脱内裤让男人舔精品视频| 欧美成人一区二区免费高清观看| 亚洲最大成人中文| 一区二区三区乱码不卡18| 国产久久久一区二区三区| av国产免费在线观看| 国产av一区在线观看免费| 国产精品一区二区三区四区久久| 成人午夜精彩视频在线观看| 在线观看美女被高潮喷水网站| 日本熟妇午夜| 精品久久久久久久末码| 三级国产精品欧美在线观看| 日本黄色片子视频| 久久久成人免费电影| 国产高清三级在线| 嫩草影院精品99| 日本熟妇午夜| 国产亚洲av嫩草精品影院| av在线蜜桃| 青春草国产在线视频| 18禁在线无遮挡免费观看视频| 在线a可以看的网站| 亚洲av免费高清在线观看| 波多野结衣高清无吗| 国产精品一二三区在线看| 久久精品夜夜夜夜夜久久蜜豆| 尤物成人国产欧美一区二区三区| 久久亚洲精品不卡| 精品国内亚洲2022精品成人| a级毛片免费高清观看在线播放| 人人妻人人看人人澡| 日日干狠狠操夜夜爽| 你懂的网址亚洲精品在线观看 | 自拍偷自拍亚洲精品老妇| 蜜桃久久精品国产亚洲av| 国产高清视频在线观看网站| 亚洲欧美成人精品一区二区| 丝袜美腿在线中文| 永久免费av网站大全| 人妻系列 视频| 久久精品国产自在天天线| 国产精品麻豆人妻色哟哟久久 | 精品不卡国产一区二区三区| 日本av手机在线免费观看| 国产精品乱码一区二三区的特点| 黑人高潮一二区| 一级毛片aaaaaa免费看小| 一边亲一边摸免费视频| 在线播放国产精品三级| 日韩欧美三级三区| 联通29元200g的流量卡| 九色成人免费人妻av| 99热精品在线国产| 国产亚洲最大av| 亚洲国产日韩欧美精品在线观看| 淫秽高清视频在线观看| 99久久精品一区二区三区| 麻豆国产97在线/欧美| 人人妻人人看人人澡| 麻豆国产97在线/欧美| 在线播放无遮挡| 国产精品女同一区二区软件| 91狼人影院| 久久6这里有精品| .国产精品久久| 国产亚洲av嫩草精品影院| 两个人的视频大全免费| 午夜激情福利司机影院| 日产精品乱码卡一卡2卡三| 人妻系列 视频| 2021天堂中文幕一二区在线观| 亚洲欧美一区二区三区国产| 午夜福利在线在线| 久久99热6这里只有精品| 一区二区三区高清视频在线| 久久久色成人| 91久久精品电影网| 最近的中文字幕免费完整| 亚洲精品自拍成人| 秋霞在线观看毛片| 国产三级中文精品| 国产黄色视频一区二区在线观看 | 国产麻豆成人av免费视频| 亚洲精品亚洲一区二区| 青春草国产在线视频| 日韩欧美精品v在线| 欧美成人免费av一区二区三区| 国产真实伦视频高清在线观看| 午夜激情福利司机影院| 日本五十路高清| 免费看日本二区| 午夜a级毛片| 老女人水多毛片| 长腿黑丝高跟| 午夜免费男女啪啪视频观看| 欧美性猛交╳xxx乱大交人| 国产精品99久久久久久久久| 亚洲av熟女| 久久久久久久久久久免费av| 一级毛片我不卡| 小说图片视频综合网站| 国产免费视频播放在线视频 | 永久免费av网站大全| 国产中年淑女户外野战色| 神马国产精品三级电影在线观看| 国产精品蜜桃在线观看| 国产欧美另类精品又又久久亚洲欧美| 欧美日本视频| 国产精华一区二区三区| 国产 一区精品| 日韩欧美国产在线观看| 在线观看美女被高潮喷水网站| 最近中文字幕2019免费版| 三级国产精品片| 久久久久久久久大av| 色5月婷婷丁香| 亚洲国产最新在线播放| av播播在线观看一区| 久久久精品大字幕| 91在线精品国自产拍蜜月| 欧美不卡视频在线免费观看| 久久久久九九精品影院| 看免费成人av毛片| 国产乱人视频| 精品无人区乱码1区二区| 村上凉子中文字幕在线| 免费看av在线观看网站| 男人舔女人下体高潮全视频| 免费看日本二区| 少妇被粗大猛烈的视频| 老师上课跳d突然被开到最大视频| 日韩精品青青久久久久久| 久久99热6这里只有精品| 久久99热这里只有精品18| av线在线观看网站| 亚洲国产欧美在线一区| 亚洲国产精品久久男人天堂| 亚洲精品色激情综合| 久久久久久久久中文| 日本爱情动作片www.在线观看| 久久鲁丝午夜福利片| av卡一久久| 欧美精品一区二区大全| 国产69精品久久久久777片| 国产精品av视频在线免费观看| 观看免费一级毛片| av在线观看视频网站免费| 欧美另类亚洲清纯唯美| 国产男人的电影天堂91| 欧美日韩国产亚洲二区| videossex国产| 两个人的视频大全免费| 国产精品一二三区在线看| 少妇人妻一区二区三区视频| 晚上一个人看的免费电影| 国产午夜精品久久久久久一区二区三区| 亚洲av电影不卡..在线观看| 美女黄网站色视频| 尾随美女入室| 国模一区二区三区四区视频| 午夜福利网站1000一区二区三区| 国产黄色视频一区二区在线观看 | 最后的刺客免费高清国语| 久久欧美精品欧美久久欧美| 欧美成人午夜免费资源| 一级毛片电影观看 | 免费av毛片视频| 看黄色毛片网站| 99久久精品热视频| 国产成人a区在线观看| 乱人视频在线观看|