國立虎尾科技大學 |

Essays in Learning and Experimentation.

紀錄類型:	書目-語言資料,手稿 : Monograph/item
正題名/作者:	Essays in Learning and Experimentation./
作者:	Paasivirta, Ilari.
面頁冊數:	1 online resource (122 pages)
附註:	Source: Dissertation Abstracts International, Volume: 79-01(E), Section: A.
標題:	Economics. -
電子資源:	click for full text (PQDT)
ISBN:	9780355128307

Essays in Learning and Experimentation.
Paasivirta, Ilari.

Essays in Learning and Experimentation. - 1 online resource (122 pages)

Source: Dissertation Abstracts International, Volume: 79-01(E), Section: A.

Thesis (Ph.D.)--New York University, 2017.

Includes bibliographical references

In our increasingly interconnected society, social learning has become an important element in more and more contexts that range from voting decisions in elections to consumers' product choices. In Chapter 1 we study social learning in a sequential decision-making environment where the individual actions of agents generate information externalities that vary over time. We model individual decisions as continuous time two-armed bandit problems, where players learn about two unknown states: the idiosyncratic type of an agent and the type of the risky arm, which is common across all players. Both types influence observed outcomes, which has the effect of giving rise to non-monotone posterior belief paths. Such belief dynamics are absent in one-dimensional settings, which the existing literature largely focuses upon. We provide a constructive method to derive closed-form solutions for the value function and characterise the possible sequences that can arise in this sequential learning environment. We also analyse welfare outcomes and find that, in the absence of successes, welfare has a downward trend but follows a wave-like path that is absent in the planner's solution. This pattern arises due to the fact that the non-monotone posterior beliefs induce similar non-monotone dynamics in the information externality generated by a single player's actions.

Electronic reproduction.
Ann Arbor, Mich. :
ProQuest,
2018

Mode of access: World Wide Web

ISBN: 9780355128307Subjects--Topical Terms:

555568
Economics.
Index Terms--Genre/Form:

554714
Electronic books.

Essays in Learning and Experimentation.
LDR:03375ntm a2200325K 4500 001 912329
005 20180608141652.5
006 m o u
007 cr mn||||a|a||
008 190606s2017 xx obm 000 0 eng d
020 $a 9780355128307
035 $a (MiAaPQ)AAI10261105
035 $a (MiAaPQ)nyu:12917
035 $a AAI10261105
040 $a MiAaPQ $b eng $c MiAaPQ
100 1 $a Paasivirta, Ilari. $3 1184643
245 1 0 $a Essays in Learning and Experimentation.
264 0 $c 2017
300 $a 1 online resource (122 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertation Abstracts International, Volume: 79-01(E), Section: A.
500 $a Adviser: Ennio Stacchetti.
502 $a Thesis (Ph.D.)--New York University, 2017.
504 $a Includes bibliographical references
520 $a In our increasingly interconnected society, social learning has become an important element in more and more contexts that range from voting decisions in elections to consumers' product choices. In Chapter 1 we study social learning in a sequential decision-making environment where the individual actions of agents generate information externalities that vary over time. We model individual decisions as continuous time two-armed bandit problems, where players learn about two unknown states: the idiosyncratic type of an agent and the type of the risky arm, which is common across all players. Both types influence observed outcomes, which has the effect of giving rise to non-monotone posterior belief paths. Such belief dynamics are absent in one-dimensional settings, which the existing literature largely focuses upon. We provide a constructive method to derive closed-form solutions for the value function and characterise the possible sequences that can arise in this sequential learning environment. We also analyse welfare outcomes and find that, in the absence of successes, welfare has a downward trend but follows a wave-like path that is absent in the planner's solution. This pattern arises due to the fact that the non-monotone posterior beliefs induce similar non-monotone dynamics in the information externality generated by a single player's actions.
520 $a In Chapter 2, we switch our focus to a strategic experimentation setting, where players face a two-armed bandit problem in continuous time. The information structure is such that players observe each other's actions, but the reward outcomes are private. Similarly to Chapter 1, we assume that agents can choose between a safe arm with a constant payoff, and a risky arm whose reward rate depends on two parameters---the idiosyncratic type of the agent and an arm type that is common across all players. Even though players never observe each other's reward realisations, we show that in equilibrium, exits release bursts of information about the exiting agents' reward histories. We characterise the amount of information that such exits release to the public and show that it is always possible to give upper and lower bounds for the number of rewards that an exiting player has experienced.
533 $a Electronic reproduction. $b Ann Arbor, Mich. : $c ProQuest, $d 2018
538 $a Mode of access: World Wide Web
650 4 $a Economics. $3 555568
655 7 $a Electronic books. $2 local $3 554714
690 $a 0501
710 2 $a ProQuest Information and Learning Co. $3 1178819
710 2 $a New York University. $b Economics. $3 1179154
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=10261105 $z click for full text (PQDT)