Dissecting arbitrage costs
This paper systematically examines the impact of nine popular arbitrage costs measures on cross-sectional mispricing based on ten well-known and robust anomalies. We show that binding arbitrage barriers slowly change over time. In early years with few publications documenting return anomalies, arbit...
Saved in:
Main Authors: | , , |
---|---|
Format: | text |
Language: | English |
Published: |
Institutional Knowledge at Singapore Management University
2017
|
Subjects: | |
Online Access: | https://ink.library.smu.edu.sg/lkcsb_research/6226 https://ink.library.smu.edu.sg/context/lkcsb_research/article/7225/viewcontent/Dissecting_arbitrage_costs_2017_sv.pdf |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Singapore Management University |
Language: | English |
Summary: | This paper systematically examines the impact of nine popular arbitrage costs measures on cross-sectional mispricing based on ten well-known and robust anomalies. We show that binding arbitrage barriers slowly change over time. In early years with few publications documenting return anomalies, arbitrage costs have tiny impact even though mispricing is present. As anomalies become more widely known, arbitrage costs impact mispricing substantially. Arbitrage risk, ambiguity of fundamental value, round-trip broker’s commission plus bid-ask spreads, and stock loan supply are binding on arbitrageurs. Only arbitrage risk is binding if larger cap stocks are emphasized. In recent years when market quality improves and some arbitrageurs become more creative, only round-trip broker’s commission plus bid-ask spreads and stock loan supply remain binding on arbitrageurs. If larger cap stocks are emphasized, arbitrage costs do not matter at all because there is no longer mispricing. An empirical arbitrage costs model based on these simple dynamics subsumes annually-varying principal components of arbitrage costs in affecting mispricing. Incorporating our findings into future capital market efficiency research would mitigate type I and II errors in empirical tests applying the limits-to-arbitrage argument. |
---|