The Choice of Direct Dealing or Electronic Brokerage in Foreign Exchange Trading



Similar documents
Can Auto Liability Insurance Purchases Signal Risk Attitude?

Lecture 3: Force of Interest, Real Interest Rate, Annuity

Crossing Network Trading and the Liquidity of a Dealer Market: Cream-Skimming or Risk Sharing?

Addendum to: Importing Skill-Biased Technology

Chapter 7: Answers to Questions and Problems

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING

The Cox-Ross-Rubinstein Option Pricing Model

Interest Rate Futures

Management Quality, Financial and Investment Policies, and. Asymmetric Information

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts

Problem Set 3. a) We are asked how people will react, if the interest rate i on bonds is negative.

The Personalization Services Firm: What to Sell, Whom to Sell to and For How Much? *

Solution: Let i = 10% and d = 5%. By definition, the respective forces of interest on funds A and B are. i 1 + it. S A (t) = d (1 dt) 2 1. = d 1 dt.

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

Management Quality and Equity Issue Characteristics: A Comparison of SEOs and IPOs

SPEE Recommended Evaluation Practice #6 Definition of Decline Curve Parameters Background:

Credit Limit Optimization (CLO) for Credit Cards

The role of time, liquidity, volume and bid-ask spread on the volatility of the Australian equity market.

Scale Dependence of Overconfidence in Stock Market Volatility Forecasts

Criminal Justice System on Crime *

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

The Application of Fractional Brownian Motion in Option Pricing

ADVERSE SELECTION IN INSURANCE MARKETS: POLICYHOLDER EVIDENCE FROM THE U.K. ANNUITY MARKET *

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy

An Alternative Way to Measure Private Equity Performance

Applied Research Laboratory. Decision Theory and Receiver Design

Faraday's Law of Induction

Hedging Interest-Rate Risk with Duration

The DAX and the Dollar: The Economic Exchange Rate Exposure of German Corporations

Economic Models for Cloud Service Markets

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

The impact of hard discount control mechanism on the discount volatility of UK closed-end funds

PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION

The Economics of Two-sided Markets 2. Platform competition!

Estimation of Perceived Flood Damage in Tokyo Metropolitan Area

Lecture 14: Implementing CAPM

Capacity Reservation for Time-Sensitive Service Providers: An Application in Seaport Management

How To Evaluate A Dia Fund Suffcency

World currency options market efficiency

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

The literature on many-server approximations provides significant simplifications toward the optimal capacity

Why Do Traders Choose to Trade Anonymously? *

THE METHOD OF LEAST SQUARES THE METHOD OF LEAST SQUARES

Stock Profit Patterns

Forecasting Irregularly Spaced UHF Financial Data: Realized Volatility vs UHF-GARCH Models

On-Line Fault Detection in Wind Turbine Transmission System using Adaptive Filter and Robust Statistical Features

Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy

A) 3.1 B) 3.3 C) 3.5 D) 3.7 E) 3.9 Solution.

DEFINING %COMPLETE IN MICROSOFT PROJECT

5 Multiple regression analysis with qualitative information

Simple Interest Loans (Section 5.1) :

17 Capital tax competition

Binomial Link Functions. Lori Murray, Phil Munz

Risk Model of Long-Term Production Scheduling in Open Pit Gold Mining

Structural Estimation of Variety Gains from Trade Integration in a Heterogeneous Firms Framework

SPECIALIZED DAY TRADING - A NEW VIEW ON AN OLD GAME

Traditional versus Online Courses, Efforts, and Learning Performance

Evidence of Adverse Selection in Automobile Insurance Markets

STATISTICAL DATA ANALYSIS IN EXCEL

Analysis of Premium Liabilities for Australian Lines of Business

Price Risk and Bid-Ask Spreads of Currency Options

Time-Varying Liquidity in Foreign Exchange

Sprottivas And Implications For Productivity

DO LOSS FIRMS MANAGE EARNINGS AROUND SEASONED EQUITY OFFERINGS?

How To Get A Tax Refund On A Retirement Account

Survival analysis methods in Insurance Applications in car insurance contracts

Two Faces of Intra-Industry Information Transfers: Evidence from Management Earnings and Revenue Forecasts

When Talk is Free : The Effect of Tariff Structure on Usage under Two- and Three-Part Tariffs

1 De nitions and Censoring

Oligopoly Theory Made Simple

Viscosity of Solutions of Macromolecules

Transcription:

The Choce of Drect Dealng or Electronc Brokerage n Foregn Exchange Tradng Mchael Melvn Arzona State Unversty & Ln Wen Unversty of Redlands

MARKET PARTICIPANTS: Customers End-users Multnatonal frms Central banks Hedge funds... Tradtonally trade wth dealers, not each other Trades prvate nfo to dealers 2

MARKET PARTICIPANTS: Dealers Trade wth customers Low transparency Trade wth each other nterbank market Multple of customer trades Passng hot potato postons 3

INTERBANK TRADING Snce 1930s, drect telephone tradng Snce 1960s voce brokers speaker boxes 1987, Reuters Dealng 2000-1 Untl early 1990s, trade splt almost n half between drect dealng and voce brokers 4

ELECTRONIC BROKERS 1992, Reuters Dealng 2000-2 1993, Mnex and EBS 1995, EBS/Mnex merger 5

ELECTRONIC BROKERS Market and Lmt orders prce/tme prorty Anonymous pror to trade Lower costs Greater transparency Contnuous multlateral nteracton 6

CUSTOMER INTERNET TRADING Nonbank stes Take prces from nterbank market Not elect. Brokers, ste s counterparty to trades May 1996, Deal4Free (CMC Group) March 2001, OANDA Bank stes Request quotes from several banks August 1996, FX Connect (State Street) Aprl 2000, Currenex Multple bank quotes and crossng network Increases competton and lowers costs 7

ELECTRONIC BROKERS Start from a base of zero n 1992 Aprl 2001 Aprl 1998 FRB of NY 54% 32% Bank of England 66% 30% Bank of Japan 48% 37% 8

Queston: How would a trader choose when facng two competng tradng venues? Theoretcal model Choce of tradng venue for large and small traders Emprcal Analyss Tests hypotheses Summary & Dscusson 9

Tradng Venues Drect Dealng (DD) Immedacy of transacton Electronc Brokerage (EB) Watng tme dscount factor δ Transacton cost s (dealer s bd-ask spread) Transacton cost c c<s 10

Theoretcal Model Players One large trader who trades a large amount Many small traders who trade 1 unt Strateges Go to DD Go to EB Don t trade 11

Theoretcal Model Asset (Currency) a random future value v Expectaton E ( v) = u Varance Payoff DM: CN: u s σ v δ ( u c) 12

Effectve Dscount Rate Effectve Dscount Rate: For a small trader: For a large trader: : dscount factor, δ s = Eβ t δ = Eβ l l β : number of perods t takes for a small trader to fnd a match t s : number of perods t takes for a large trader to fnd a match t l t F () t F () t l ts Et Et, Eβ Eβ l s l s t s 0 < β <1 13

Optmal Decson Rules Trade wth DD f u s > δ ( u c), and u s >0 Trade wth EB f u s< δ ( u c), δ ( u c) >0 Indfferent f u s = δ ( u c) >0 No trade f δ ( u c) <0, u s <0 14

Optmal Outcome u<c nobody would trade c<u<s exclusve EB tradng u>s two possble equlbra when DD & EB coexst The large trader trades wth DD and small traders go to the EB. ( s δ lc) /(1 δ l ) < u < ( s δ sc) /(1 δ s ), δ s > δ l; The large trader trades on EB and small traders trade wth DD. (ruled out) 15

Emprcal Analyss Data Descrpton Reuters D2000-2 electronc brokerage Mark/Dollar Oct 6-10, 1997, 130,535 orders Avalable Informaton: order type, order entry tme, removal tme, removal code, prce, quantty ordered and quantty dealt 16

Duraton tme of orders Average duraton for lmt orders s longer than that for market orders Mean watng tme s longer for unsuccessful lmt orders than flled lmt orders Tme of day effect Clusterng n the duraton data 17

Descrptve Statstcs for Duraton Flled Lmt Orders Faled Lmt Orders Flled Market All sample Orders Number of 38239 70453 21783 130475 Orders Mean (mn) 1.7886 3.4331 0.0012 2.3782 Std Devaton 10.8107 18.6327 0.0008 14.9453 Range 398.394 802.6677 0.0503 802.6702 Skewness 19.2068 14.4837 14.2692 17.4444 Kurtoss 494.4853 308.2410 689.0021 446.8854 18

Table 3 Intradaly Pattern of Duraton Tme of Day Average Duraton Number of Orders Percentage 0 10.2146 477 0.37% 1 6.7147 692 0.53% 2 10.8359 317 0.24% 3 35.8062 64 0.05% 4 9.0134 200 0.15% 5 5.5536 891 0.68% 6 2.6893 5595 4.29% 7 2.3079 14491 11.10% 8 2.3178 15097 11.57% 9 3.0254 9696 7.43% 10 3.2417 7360 5.64% 11 2.1809 13006 9.96% 12 1.5406 16790 12.86% 13 1.4885 18976 14.54% 14 1.5596 14518 11.12% 15 2.391 6416 4.92% 16 4.7557 2139 1.64% 17 4.8778 1570 1.20% 18 3.1406 1510 1.16% 19 4.6772 446 0.34% 20 6.0416 143 0.11% 21 36.8439 43 0.03% 22 43.329 29 0.02% 23 44.7129 69 0.05% 19

Estmaton of duraton model Three Hypotheses Sze Effect Prce Impact Lqudty Effect 20

ACD model ACD Model Duraton Condtonal duraton ε x ψ s an IID error sequence EACD (flat hazard functon) Webull ACD (monotone hazard functon) x =ψ ε 21

ACD model Burr ACD model Inverted U-shaped Hazard functon Hazard functon ncreasng for small duraton and decreasng for long duraton Nests EACD and WACD model as specal cases 22

Burr-ACD Burr-Dstrbuton: Densty Functon: Hazard Functon EACD WACD ) 1 1 ( ) 1 (1 1) 1 ( ) ( ) ( 2 2 ) 1 (1 2 κ σ κ σ σ ψ ψ κ Γ + Γ + Γ = + f κ κ κ κ ξ σ ξ κ θ x x x x x h + = 2 1 1 1 1 ) ;,..., ( γ γ 1 1 1 ),..., ( = x x x x h x x x h ψ 1 ),..., ( 1 1 = 23

Representatve Hazard Functons 2.5 Hazard Functon 2 1.5 1 0.5 Webull 0.5, 0 Burr 2, 0.5 0 0 2 4 6 8 10 Duraton 24

ACD model Concerns: Dependent Varable: Condtonal duraton Rght hand sde of estmaton equaton needs to be postve Non-negatvty constrants on the coeffcents of exogenous varables 25

Log ACD Model Log-ACD Model Duraton x = exp(ψ ) ε ψ : Logarthm of condtonal duraton ε s an IID sequence as n ACD model. Log-ACD(1,1) specfcaton ψ x 1) βψ = ω + α ln( + 1 x ε 26

Censorng Potental bas from gnorng unflled orders or partal flls Estmate jont lkelhood n c 1 c f(x ;X ) g(x ;X ) = f(x ;X ) g(x ;X ) = 1 F C 27

Model Estmaton Over peak European busness hours 8:00am 5:00 pm GMT Varables SIZE: Quantty submtted n mllons of dollars PRICEDIF: submsson prce - last transacton prce DEPTH: depth of order book 28

Model Estmaton Dummy varables DummyBP 1 for buy orders wth prcedf>0; 0 otherwse DummyBN 1 for buy orders wth prcedf<0; 0 otherwse DummySP 1 for sell orders wth prcedf>0; 0 otherwse DummySN 1 for sell orders wth prcedf<0; 0 otherwse 29

Model Estmaton Burr Log-ACD (1,1) model ψ + + δ δ x = ω + α ln( + βψ 1 2 3 DummyBP DummySP 1) + + δ δ 3 4 + δ 1 DummyBN DummySN SIZE + δ 5 DEPTH 30

Model Estmates (flled orders) Coeffcent Std. Error T-Stat Prob SIZE 0.0197 0.0085 2.33 0.0197 DummyBP -1.2171 0.0402-30.31 0.0000 DummyBN 1.7880 0.0368 48.59 0.0000 DummySP 1.7140 0.0355 48.31 0.0000 DummySN -1.2902 0.0389-33.17 0.0000 LDEPTH -0.0084 0.0003-30.20 0.0000 MDEPTH -0.0101 0.0004-25.22 0.0000 31

Model Estmates (censored orders) Coeffcent Std. Error T-Stat Prob SIZE -0.0484 0.0007-64.78 0.0000 DummyBP -0.2920 0.0295-9.89 0.0000 DummyBN 1.5238 0.0308 49.47 0.0000 DummySP 1.4583 0.0313 46.60 0.0000 DummySN -0.3193 0.0304-10.49 0.0000 LDEPTH 0.0155 0.0060 2.60 0.0000 MDEPTH -0.0052 0.0004-14.19 0.0000 32

Estmated Hazard Functon Hazard Functon 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 0 2 4 6 8 10 Duraton Burr, 0.6544,0.5135 33

Conclusons Explan choce of tradng venues Large traders prefer drect dealng whle small traders utlze the electronc brokerage Emprcal results consstent wth hypotheses from theory. Large orders wat longer on EB gven the depth of the market and prce compettveness. 34