econ 191: background material for prof. eichengreen's ...webfac/eichengreen/e191... ·...
TRANSCRIPT
Econ 191: Background material for Prof. Eichengreen’sLecture on International Currencies
Owen Zidar
University of California, Berkeley
September 11, 2012
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 1 / 33
Announcements
Twitter: Follow @economics191
Sample Question: Turn in sample research question now please
Attendance: Please sign attendance sheet that is going around
Office Hours: Sign up online - need to see GSI once by Oct 2.
Next Week: Professor Eichengreen lecturing on internationalcurrencies
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 2 / 33
Overview of Today’s Lecture
1 Historical Context: Brief History of the Evolution of theInternational Monetary System
2 Tools: Referesher on Multivariate Regression
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 3 / 33
1. Big Picture - History of International Monetary System1
Gold Standard 1821-1914
Interwar period
Bretton Woods Fixed Exchange Rates: 1946-1973
After Bretton Woods
1Sources: ritholtz.com/blog/ for exchange rate images and InternationalEconomics by Krugman and Obstfeld
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 4 / 33
I. Gold Standard: 1821-1914
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 5 / 33
I. Gold Standard
You can “cash in” your paper notes for predetermined amount of goldSome Issues:
Limits monetary growth to gold stock, which can’t be adjusted inresponse to economic conditions
Global shortage of gold put downward pressure on prices
E.g. In 1896, the U.S. price level was roughly 40% below its 1869 level
Farmers were squeezed between declining prices for crops and thefixed dollar payments for their mortgages and other debts.
William Jennings Bryan, Cross of Gold, and Wizard of Oz
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 6 / 33
II. WWI ‘Dirty Float’ 1915-1925
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 7 / 33
II. Interwar Period
War is Expensive: Many countries suspended gold standard, US dida few year later when it entered WWI
Inflation Taxes
Strength in US: US returned to Gold in 1919
Genoa: 1922 conference establishing gold exchange standard2
Restricted Monetary Flexibility: 1925 Britain returned to the goldstandard. Contractionary policy to get back to prewar price level.
2Countries could hold securities convertible into gold issued by othercountries as reserves
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 8 / 33
II. Interwar Gold-Exchange Standard: 1926-1931
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 9 / 33
II. Pre-Bretton Woods Floats: 1932-1944
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 10 / 33
III. Bretton Woods: 1945-1971
Fixed exchange rates against the dollar
Unvarying dollar price of gold: $35 an ounce
Member countries held international reserves in dollars
They had the right to sell dollars for gold at the official price3
They also had the right to adjust their peg to the dollar (pending IMFagreement)
3Bretton Woods differed from interwar gold exchange standard in that theright to sell dollars to the Treasury for gold was limited to official foreigncreditors (governments and central banks
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 11 / 33
III. Bretton Woods: 1945-1971
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 12 / 33
IV. Smithsonian Agreement: 1972-1973
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 13 / 33
III. Floats and Pegs: 1974-1979
US macro policies in the late 1960s helped cause the breakdown ofBretton Woods by early 1973
Most advanced countries abandoned stated pegs, while manydeveloping countries stayed pegged
1974-1975: Stagflation4 led most gov’ts to use monetary and fiscalpolicy
Many attempts to coordinate pseudo-fixed exchange rates withinEurope
4Stagnant growth, high inflation. Higher commodity prices increasedinflation & depressed output. Expectations of future inflation fed into wages &prices
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 14 / 33
IV. Floats and Pegs: 1974-1979
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 15 / 33
IV. ERM, Euro, ... : 1980-
Dollar Fluctuates: 1976-1979 Weak Dollar, Paul Volker & Deficits,Joint Intervention to depreciate in 1985
ERM semi-pegged system: member banks would intervene tomaintain 2.25% band. UK joined in 1990 and exited 2 years after.
German Reunification: 1990 reunification set off inflationarypressures, divergent needs in Europe and speculative attacks (leadingto 15% bands in 1993).
Inflation Convergence: European countries peg to deutche markwithin Europe’s fixed exchange rate mechanism EMS5
Euro: Maastrict Treaty and a common currency in 1999
Asian Crisis of 1997: Many stopped adhering to dollar pegs after
5European Monetary SystemOwen Zidar (Econ 191) Background for Eichengreen September 11, 2012 16 / 33
IV. ERM, Euro, ... : 1980-
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 17 / 33
Part 2: Multivariate Regression Refresher6
Contents
1 Preliminaries
2 Big picture idea: summarize data
3 Linear model4 Multivariate
Frisch Waugh interpretationOmitted Variable BiasReading reg output
5 Methods: D-D and IV
6 Probit
6Some of these notes are based on Mostly Harmless Econometrics andCharlie Gibbon’s 140 notes
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 18 / 33
1. Preliminaries
1 Expectations
E(Y ) = Y1 × Prob1 + Y2 × Prob2 + ...+ Yj × Probj (1)
2 VarianceVar(Y ) = E[Y − µy ]2 (2)
3 CovarianceCov(X ,Y ) = E[X − µx ][Y − µy ] (3)
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 19 / 33
2. Big picture
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 20 / 33
3. Linear model
Decomposition
yi = E[yi |xi ] + (yi − E[yi |xi ])︸ ︷︷ ︸≡εi
(4)
= β0 + β1xi + εi (5)
Example: Height by gender
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 21 / 33
3. Linear model Continued
What is β1? β1 = Cov(x ,y)Var(x)
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 22 / 33
4. Multivariate model
yi = β0 + β1x1i + β2x2i + ...+ +βkxki + εi
(6)
βk =Cov(y , x̃k)
Var(x̃k)(7)
where x̃k is the residual regression of xk on all the other covariates.
Takeaway: βk is the relationship between y and the part of xk thatisn’t related to other x’s
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 23 / 33
4. Multivariate model: Omitted Variable Bias
Suppose that the true model is
yi = γ0 + γ1Schoolingi + γ2Abilityi + ε̃i (8)
Also suppose that Abilityi can be written:
Abilityi = α0 + α1Schoolingi + ηi (9)
What happens if we omit ability and run a “short” regression?
yi = β0 + β1Schoolingi + εi (10)
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 24 / 33
4. Multivariate model: Omitted Variable Bias cont.
What happens if we omit ability and run a “short” regression?Plug in ability expression from (10) into true model to see
yi = γ0 + γ1Schoolingi + γ2(α0 + α1Schoolingi + ηi ) + ε̃i
yi = (γ0 + γ2)︸ ︷︷ ︸β1
+ (γ1 + γ2α1)︸ ︷︷ ︸β2
Schoolingi + (ηi + ε̃i )︸ ︷︷ ︸εi
Takeaways
β2 = γ1 + γ2α1
Short equals long plus the effect of omitted times the regression ofomitted on included.
Does γ2α1 = 0?
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 25 / 33
4. Reading regression output
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 26 / 33
5. Methods: Difference in Difference
Effect = [After-Before]Treatment − [After − Before]Control
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 27 / 33
5. Methods: Instrumental Variables
Setupyi = γ0 + γ1Schoolingi + γ2Abilityi + ε̃i︸ ︷︷ ︸
≡vi
(11)
Exclusion Restriction:7 An instrument, zi , is correlated with Schoolingibut uncorrelated with other determinants of dependent variables
Cov(vi , zi ) = 0
Cov(yi − γ0 − γ1Schooling,zi ) = 0
Cov(yi , zi )− γ1Cov(Schooling,zi ) = 0
Cov(yi , zi )
Cov(Schooling,zi )= γ1
7because zi can be “excluded” from the causal modelOwen Zidar (Econ 191) Background for Eichengreen September 11, 2012 28 / 33
5. Methods: Instrumental Variables continued
Implementing this: First stage and reduced form
Schoolingi = θ0 + θ1zi + ηi (12)
yi = π0 + π1zi + ui (13)
γ1 =Cov(yi , zi )
Cov(Schooling,zi )
γ1 =Cov(yi , zi )/Var(zi )
Cov(Schooling,zi )/Var(zi )
γ1 =π1θ1
Caution: Important that first stage is strong (rule of thumb is F > 10)
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 29 / 33
6. Probit8
What if limited dependent variable? Yi = 0 or Yi = 1
8Graph from Ashenfelter et al’s Statistics and Econometrics BookOwen Zidar (Econ 191) Background for Eichengreen September 11, 2012 30 / 33
6. Probit (continued)
Setup:
E(Yi ) = Pr(Yi = 0)0︸ ︷︷ ︸=0
+Pr(Yi = 1)1
= Pr(Yi = 1)
≡ Φ(xiβ)︸ ︷︷ ︸probit
Marginal Effects:
∂Pi
∂xi=∂Φ(xiβ)
∂xi= βφ(xiβ)
where Φ is the c.d.f. and φ is the p.d.f. of the normal distribution9
9Φ is prob(z < xiβ), z ∼ N(0, 1), so need to standardize if actually using thisOwen Zidar (Econ 191) Background for Eichengreen September 11, 2012 31 / 33
6. Probit continued
What does this look like in a graph?
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 32 / 33
Next Time
Professor Eichengreen on International Currencies
Owen Zidar (Econ 191) Background for Eichengreen September 11, 2012 33 / 33