New Micro Exchange Rate Economics

1
New Micro Exchange Rate Economics

• Keynote Address
• March 2003
• Richard K. Lyons
• U.C. Berkeley and NBER
• These slides and other resources in New Micro
  (e.g., working paper clearinghouse) are available
  at
• faculty.haas.berkeley.edu/lyons

2
New Micro Introduction

• Information Focus
• (1) What is the nature of the information the FX
  market is aggregating?
• (2) How does it achieve this aggregation?
• Tools
• Largely from microstructure finance
• Provides theoretical and empirical guidance

3
Information Focus

• (1) Public domain current macro variables
• But inflation, Y, Ms, and NX are micro data
  aggregated months later by official inst. Any
  impounding in price in the meantime?
• Also, what about disagreement on implications,
  even if observed simultaneously?
• (2) Public domain future macro variables
• How are variations in individuals expectations
  impounded in price if not common knowledge?
• (3) Non-public domain asset pricing variables
• What about shocks to money demands, shocks to
  hedging demands, or shocks to risk prefs? How are
  these impounded in price?
• (4) Non-public domain other variables
• What about information regarding the strategic
  choices of others, or information about
  behaviorally-motivated actions of others, how are
  these impounded in price?

4
Modern Exch. Rate Economics

Modern ER Econ.
Micro-founded
Non-rational
New Micro
New Macro

• Focus info econ. of fin. markets
• Info structure dispersed info
• Disconnect Q why macro so little ER impact?

• Focus sub-opt. behavior
• Approach trending from noise, feedback,
  chartism to behavioral econ.

• Focus supply side of real econ.
• Info structure CK (com. knowl.)
• Disconnect Q why ER so little macro impact?

5
Order Flow An Information Vehicle

• (1) Order flow is sum of signed trades (not
  volume)
• Signed according to which side initiates
• Quoting marketmaker is non-initiating side
• Auction structure limit order is non-initiating
• (2) Order flow is not the same as net demand
• OF measures transactions (i.e., demands after
  price has adjusted)
• Think of overshooting model ? demand ? ? in
  s price, but no executed transactions initiated
  by buyers along the way, nor any imbalance in
  dollar buying at the new price.
• Unlike net demand, cum. OF may ? 0
• In some models cum. OF follows RW
• Price impact differs depending on trader identity
  
• Link to info econ whose trades info rich?
• (3) Another perspective information is in the
  shifts
• How ID demand and supply curves?
• Think of scatter in P-Q space
• Exclusion restrict. shifts (vs moving along)
• Microstructure theory IDs shifting

6
Order Flows Role Graphically
Macro Approach

Microstructure Approach
Private info
Price
Order flow
Hybrid
Information
Price
Order flow
7
New Micro Frequently Asked Questions

• Isnt order flow just demand?
• No, as addressed earlier (transactions?demand)
• With two sides to every trade, what can we learn?
• True, but they need not be symmetric one side
  may be a demand curve shift, the other a
  price-induced movement along a stable demand
  curve
• Price impounds info in the shift
• But isnt order flow an endogenous variable?
• Yes, but so are all the traditional macro
  variables we have been using in empirical
  exchange rate models (e.g., output, money,
  interest rates, and inflation).
• The key is determining the underlying shocks
  (information) that drive order flow.
• Strategies for doing so are outlined below
• Might causality also go in reverse, from price to
  order flow?
• Surely the answer is yes. For example, price
  changes induce movements along demand curves,
  which can in turn produce initiated trades, i.e.,
  order flow. Price changes can also cause demand
  curves to shift, e.g., at times of market stress
  (due to capital constraints, etc.).
• On average, price-induced order flow in FX data
  appears to be negative (Evans and Lyons 2002 JME,
  2003 NBER Tien 2002)
• So reverse causality does not account for the
  strong positive correlation between OF and ER
  changes.
• Dont lose perspective that causality can run
  both ways does not mean that order flow is not
  the cause of most ER variation.

8
New Micro Frequently Asked Questions

• Can the info conveyed by flow be fundamental?
• Yes, even if fundamental is defined narrowly to
  mean money supply and income (Ms and Y).
• 2 examples above (1) micro data not yet
  officially aggregated and (2) individuals
  changing expectations of future M and Y
• There is evidence that flows forecast future U.S.
  money growth and real output (Evans and Lyons NSF
  2003)
• Even if not reflecting narrow fundamentals, OF
  can convey info about mkt-clearing risk premia (a
  la Portfolio Balance models)
• Examples above shocks to hedging demands and to
  risk preferences
• PB effects not traditionally called fundamental
• As empirical question, remains open
• Do order flow effects on ERs persist?
• Theory persistence depends on info type
• Distinguish nominal ER effects from real
• Data much evidence that nominal ER effects do
  persist (Evans and Lyons 2002 JPE, Payne JIE
  forthcoming, Killeen et al. 2001 NBER, Froot and
  Ramadorai 2002 NBER).
• Multi-year plots are not consistent with impact
  that fades in months or less levels would not
  track over years unless impact persists over
  years
• Profits Rapid mean reversion of OF effects would
  imply trading strategies so profitable that
  theyre unrealistic

9
What Info Drives Order Flow?

• 4 Lines of Empirical Attack (so far)
• (1) Macro news OF less important then?
• Empirical may be more importanthelps market
  aggregate differential interpretations
• Example on next slide
• (2) Disaggregate OF identity matters?
• Empirical differential price impact across
  trader types shows which types best informed
• (3) Cross Currency / trades info for /?
• Empirical pattern of cross-market effects shows
  whether info specific to , , or
• (4) Macro expectations OF proxy changes?
• Price depends expected future macro
• OF measures expectational votes over time?

10
Does Macro News Drive Order Flow?
Total Variation 10 20 40 30
_______ 100

• (1) Public news impounded immediately and
  directly
• (2) Public news impounded via OF
• (3) OF unrelated to macro news (à la Evans
  Lyons 2002 JPE)
• (4) Dont know

See Evans and Lyons, How is Macro News
Transmitted to Exchange Rates (NBER 9433, Jan.
2003) for estimation details.
11
Taxonomy of Information Types

• Traditional split public vs private
• Public to be impounded in price without order
  flow role, not only must data be commonly
  observed, but traders must also agree on ER
  implications (i.e., plenty of room for OF role).
• Private as noted, many sub-categories within
• Private information 2x2

Note in a risk neutral world, only the payoff
column is relevant.
12
From Causes to Consequences

• Thus far work mostly on identifying the causes of
  order flows information role
• E.g., identifying the underlying information that
  causes flow.
• But what are the consequences of order flow
  conveying information?
• Application International Currencies
• Why do some currencies play a disproportionate
  international role?
• Big question with a long history
• Does the finding that order flow conveys
  information have any implications?

13
Information Approach to Intl Currencies

• Paper (with Michael Moore) addresses a specific
  dimension of the international role of
  currencies the pattern of cross-currency
  exchange.
• Currency competition outcomes driven by relative
  transaction costs across markets.
• Information approach is radically different from
  two traditional approaches.
• Market-size approach (e.g., Krugman 1980 JMCB,
  Rey 2001 RES) transaction costs are a decreasing
  function of total volume traded.
• Emphasis on resulting increasing returns and
  multiple equilibria more trade reduces trans
  costs, which promotes more trade.
• Marketmaker-risk approach (e.g., Black 1991 JIMF
  and Hartmann 1998 CUP book) adds exchange rate
  volatility to the determinants of transaction
  costs.

14
Analytical Results

• Indeterminacy Info dimension resolves
  traditional indeterminacy of currency trade
  patterns (by mitigating the concentrating force
  of market size economies).
• Asset Trade Pattern of currency trade not
  driven by the pattern of real trade, as in
  traditional models, but instead by the pattern of
  asset trade.
• Exchange Rate Whether transactions are executed
  directly or indirectly effects exchange rate
  determination (because these trading methods do
  not reveal the same information).
• Missing Markets Model provides a new rationale
  for why some currency pairs never trade directly
  information is not sufficiently symmetric to
  induce participation.

15
Empirical Results

• Paper is first integrated analysis of
  transactions in a triangle of markets (, , )
• Predictions
• Coefficient signs
• Matrix positive semi-definite
• Matrix symmetric
• Table 1
• Coefficient signs all correct (and significant)
• Semi-definiteness borne out
• Problems with symmetry
• 86 is ratio of indirect to direct / trading

16
Concluding Remarks

• Strength of New Micro is its empirical basis
• Caveat of data availability
• Theoretical work behind, but catching up
• E.g., Hau Rey (here), Bacchetta van Wincoop
  (2002)
• Much being done, both causes consequences
• Outward reaching
• New macro models
• Asset pricing
• Traditional portfolio balance models
• Behavioral finance