Bank Of Greece Sa · ATHEX:TELL
Bank Of Greece Sa — Technical Presentation (June 2011)
51 slides from Bank Of Greece Sa's technical presentation, June 2011
Slide 1: What Drives US Foreign Borrowing? Evidence on External Adjustment to Transitory and Permanent Shocks
What Drives US Foreign Borrowing? Evidence on External Adjustment to Transitory and Permanent Shocks Giancarlo Corsetti Panagiotis Th. Konstantinouy University of Cambridge, University of Rome III, CEPR y University of Macedonia BoG, June 2011 Corsetti, Konstantinou (Cam, UoM) US Borrowing: P vs. T
Slide 2: Introduction
Introduction Premise of our work Introduction Over the recent past, rising current account deficit in the US... Correctly measured, the US external position can be estimated Page to be as 1 of 1 high as -15% of US GDP in 2007. A picture from Milesi-Ferretti: Corsetti, Konstantinou (Cam, UoM) US Borr
Slide 3: What drives the current account? Basic theory
Introduction Premise of our work What drives the current account? Basic theory Modern international economics stress that the current account results from the intertemporal investment and consumption decisions by firms and households: CAt = St It = Changes in net foreign wealth (NFA). Models of equi
Slide 4: What drives the current account? From theory to empirics
Introduction Premise of our work What drives the current account? From theory to empirics The essential ingredients of the intertemporal-trade model 1 Because of smoothing, consumption should adjust swiftly to permanent increases in income, but be relative insulated from transitory variations – in a
Slide 5: Empirical models
Introduction Empirical Models of the current account Empirical models Early approach employs variants of Campbell (1987) and Campbell and Shiller (1987): testing frameworks build on enough restrictions to derive a present value relation equating, in expectations, the current account balance to the p
Slide 6: Restrictive assumptions (1): returns
Introduction Empirical Models of the current account Restrictive assumptions (1): returns Two main assumptions are extremely restrictive from the outset: 1 Constant return on NFA portfolio I when the return to net foreign wealth is allowed to vary stochastically according to some stationary process,
Slide 7: Restrictive assumptions (2): permanent/transitory
Introduction Empirical Models of the current account Restrictive assumptions (2): permanent/transitory 2. All shocks to (net) output are permanent – there is only one disturbance, a unit root country-specific technology shock that generates a permanent response in output! I Without being necessary t
Slide 8: Restrictive assumptions: theory-consistent data
Introduction Empirical Models of the current account Restrictive assumptions: theory-consistent data Most previous empirical models employed current account data based either on national accounting identities, or on balance of payments data. Results in Tille (2003), Lane and Milesi-Ferretti (2001, 2
Slide 9: Taking stock: two main questions
Introduction This Paper Taking stock: two main questions To understand the CA, the relevant shocks are: I returns vs. output, I transitory vs. permanent 1 What is the evidence on the relative weight of different shocks in explaining external imbalances? 2 Does the macroeconomic response to these dif
Slide 10: Key assumptions
Introduction This Paper Key assumptions Our results are based on a minimal set of assumptions (enough) that guarantee the existence of a long-run equilibrium. I All the is required is that a country’s intertemporal budget constraint holds, and ... I a balanced-growth assumption to be satisfied in th
Slide 11: An overview of the main results
Introduction Main Results An overview of the main results Virtually all variation in aggregate consumption is dominated by permanent innovations, while it is not excessively smooth vis-à-vis persistent shocks to income. Instead, consumption seems completely insulated from transitory variations in i
Slide 12: Rest of the Talk
Rest of the Talk I Rest of The Talk I Intertemporal Budget Constraint: Assumptions and Implications Empirical Framework Main Findings Extensions and Robustness Conclusions Corsetti, Konstantinou (Cam, UoM) US Borrowing: P vs. T Shocks BoG, June 2011 12 / 29
Slide 13: Intertemporal Budget Constraint: Definitions
The Intertemporal Budget Constraint Trends in Variables Intertemporal Budget Constraint: Definitions The current account (CAt ) of c country is defined as the change in the value of net foreign assets between any two periods. Let At denote (the value of the stock of) gross assets, Lt gross liabiliti
Slide 14: Intertemporal Budget Constraint: Assumptions
The Intertemporal Budget Constraint Trends in Variables Intertemporal Budget Constraint: Assumptions Let us decompose any variable Xt distinguishing between a deterministic trend exp( x;t ), and a non-deterministic-trend component Xt , i.e. Xt = Xt exp( x;t ). Under this notational convention, the c
Slide 15: Intertemporal Budget Constraint: Assumptions
The Intertemporal Budget Constraint Trends in Variables Intertemporal Budget Constraint: Assumptions The deterministic trend component of each variable w;t+i , for w = A; L; Z; C in period t + i, need not be the same of all variables! Is this necessary? ct − zt (Non−Durable & Services Consumption) a
Slide 16: Intertemporal Budget Constraint: Assumptions
The Intertemporal Budget Constraint Trends in Variables Intertemporal Budget Constraint: Assumptions So, w;t+i , might be any deterministic function of time (t + i) in principle. Is this necessary? ct − zt (Non−Durable & Services Consumption) at − zt −0.05 0.0 −0.5 −0.10 −1.0 −0.15 1980 1990 2000 19
Slide 17: Intertemporal Budget Constraint: Assumptions
The Intertemporal Budget Constraint Assumptions Intertemporal Budget Constraint: Assumptions Assumption (Balanced Growth) limk!+1 w;t+k = t+k ; for w = c; z; a; l. 1 2 (Transversality) limk!1 Et Rt;t+k At+k+1 exp a;t+k+1 Lt+i+1 exp l;t+k+1 = 0. 3 TheP expectation terms P1 1 E i=0 Rt;t+i Ct+i exp c;t
Slide 18: IBC: Under These Assumptions...
The Intertemporal Budget Constraint Assumptions IBC: Under These Assumptions... 1 The log ratios: ct zt , at zt and lt zt are trend-stationary, or that the variables are pair-wise cointegrated around deterministic trends. 2 It is possible to obtain a present value relation of the form ( a + l) (ct z
Slide 19: Cointegration Analysis
Empirical Framework Empirics and P-T Decomposition Cointegration Analysis In our analysis xt = [ct ; zt ; at ; lt ]0 . We first examine whether the log-ratios: ct zt ; at zt and lt zt are trend-stationary as assumed in our derivations. Under these cointegrating restrictions, we estimate a VEC Model
Slide 20: Permanent and Transitory Decomposition
Empirical Framework Empirics and P-T Decomposition Permanent and Transitory Decomposition Following Gonzalo and Granger (1995), Gonzalo and Ng (2001): a shock P1t is permanent if it has long-run effect on the level of the variables, whereas it is transitory if it does not have long-run effects on th
Slide 21: Table 2: Forecast Error Variance Decomposition (Orthogonalized Shocks)
32 THE AMERICAN ECONOMIC REVIEW MONTH YEAR TABLE 2—F ORECAST E RROR VARIANCE D ECOMPOSITION (O RTHOGONALIZED S HOCKS ) 1ctCh Et 1ctCh 1z tCh Et 1z tCh 1atCh Et 1atCh 1ltCh Et 1ltCh C A?tCh Et C A?tCh Panel A: Consumption Adjustment Coefficients set to Zero ( c D 0) h A.1: Contribution of Permanent S
Slide 22: Variance Decompositions: What's New?
Empirical Findings Variance Decompositions Variance Decompositions: What’s New? A measure of the US current account: CA?t NFAt+1 , with NFAt = exp (at ) exp (lt ). I T shocks account for the vast majority of the fluctuations in (this proxy of) the CA: between 96% and 98% at virtually all horizons; P
Slide 23: Impulse response charts: model with and without consumption adjustment
Model with αc=0 Model with αc ≠ 0 Net Output Net Output 1 1 0.8 0.8 Percent Percent 0.6 0.6 0.4 0.4 0.2 0.2 0 10 20 30 40 0 10 20 30 40 Consumption Consumption 1 1 0.8 0.8 Percent Percent 0.6 0.6 0.4 0.4 0.2 0.2 0 10 20 30 40 0 10 20 30 40 Gross Assets Gross Assets 4 4 3 3 Percent Percent 2 2 1 1 0
Slide 24: Impulse Responses to Permanent Shock
Empirical Findings Impulse Responses to Permanent Shock Impulse Responses to Permanent Shock Another Look at at and lt The correlation in the movements of at and lt is a novel empirical result in the literature, which turns out to be robust to different methodologies — e.g. it is also found by Corse
Slide 25: Transitory Shocks and Variations in Gross Positions
Empirical Findings Transitory Shocks in zt and Returns Transitory Shocks and Variations in Gross Positions In our results, temporary fluctuations are an important driver of zt , but even more so of gross positions and CA?t . Since the effect of transitory innovations on zt die out relatively quickly
Slide 26: Transitory Shocks and Variations in Gross Positions
Empirical Findings Transitory Shocks in zt and Returns Transitory Shocks and Variations in Gross Positions 0.2 aTt with αc =0 0.2 lTt with αc =0 rat − MA(4) rlt − MA(4) 0.05 0.05 0.1 0.1 0.0 0.00 0.0 0.00 −0.1 −0.1 −0.2 −0.05 −0.05 1980 1990 2000 1980 1990 2000 0.2 aTt with αc estimated freely 0.2 l
Slide 27: Extensions: Long-Horizon Regressions
Empirical Findings Externsions and Robustness Extensions Table: Long-Horizon Regressions (Approximate Current Account Expression) – Sample:1973-2004 Horizon H 1 2 3 4 8 12 16 Dependent Variable ft regressed on CA PH h=1 ct+h -0.014 -0.042 -0.062 -0.099 -0.091 0.060 -0.002 [t-stat] [-0.544] [-0.823]
Slide 28: Conclusions
Conclusions Conclusions In this paper, we carry out an analysis of the US external balance differentiating between trend and cycle components in US consumption, net output, gross foreign assets and gross foreign liabilities. I We identify permanent and transitory shocks, and analyze the dynamics of
Slide 29: Conclusions
Conclusions Conclusions In line with the IACA: consumption is ‘insulated’ from the corresponding transitory variations in output and gross asset positions. I Consumption is well described by a trend component/its variation is dominated by permanent shocks. I Consumption responds swiftly to permanent
Slide 30: International Financial Adjustment in the G-7: How Much? How Fast?
International Financial Adjustment in the G-7: How Much? How Fast? Roberto Cardarelli Panagiotis Th. Konstantinouy International Monetary Fund, Asia and Pacific Department, Washington DC, USA y University of Macedonia, Department of Economics, Thessaloniki, Greece BoG, June 2011 Cardarelli, Konstant
Slide 31: Two Complementary Approaches: The traditional view
Two Complementary Approaches: The traditional As external imbalances occur because of savings and investment decisions, current imbalances merely reflect high investment relative to national savings (CA < 0 and TB < 0). In the near future the will be some re-balancing by higher production of economi
Slide 32: The alternative (complementary) view
The alternative (complementary) view Recent financial integration has led to increases in gross assets and liabilities positions, resulting in country portfolios that may be heavily affected by fluctuations in asset prices. These valuation effects have been overlooked thus far, both from theory and
Slide 33: A New Theory
A New Theory Tille (2008) focuses on how valuation effects affect the transmission of monetary shocks. Blanchard, Giavazzi, and Sa (2005) set up a portfolio problem with imperfect asset substitutability and examine the role of the exchange rate in the valuation channel. Devereux and Saito (2006) emp
Slide 34: How Much Can Valuation Effects Do?
How Much Can Valuation Effects Do? Gourinchas and Rey (2007a) provide a framework to analyze the question of external adjustment, and characterize two adjustment channels. 1 The traditional TRADE CHANNEL : current imbalances are related to future trade surpluses. 2 The VALUATION CHANNEL : expected f
Slide 35: This Paper ...
This Paper ... Examines the relative importance of the valuation and trade channels of external adjustment for all G7 countries. We investigate how much the two channels contribute to the process of cyclical external adjustment, hence their relative importance. We also assess the horizons at which c
Slide 36: This Paper ...
This Paper ... Puts together data for the G7 countries on a comparable basis We built a comprehensive dataset for all G7 economies on a comparable basis... I Stocks of foreign assets and liabilities on an annual basis are put together in Lane and Milesi-Ferretti (2007a). I We construct quarterly, ma
Slide 37: This Paper ...
This Paper ... Makes a methodological contribution in constructing measures of cyclical imbalances We work with a modified set of assumptions, relative to Gourinchas and Rey (2007a) which imply that I Gross positions, exports and imports relative to GDP might be trending over time. I The ratios of A
Slide 38: This Paper ...
This Paper ... Makes a methodological contribution in measuring the amount and speed of adjustment due to trade flows and due to valuation effects Following the method suggested by Konstantinou (2010), we assess the relative importance (how much?) and the horizon (how fast?) the two channels operate
Slide 39: In This Paper ...
In This Paper ... We find that 1 There are strong valuation effects for Japan, U.S., and (under some restrictive assumptions) Germany. 2 The majority of external imbalances correction takes place through trade flows (the traditional view) for all G7 countries, while there are no important valuation
Slide 40: The Basic Framework
The Basic Framework Consider the accumulation identity for NFA between t and t + 1 : NFAt+1 = RF;t+1 (NFAt + NXt ); (1) Let us start by dividing the accumulation identity (1) by Y, the level of output. Defining Z~t Zt =Yt we obtain ~ t+1 A L ~t ~t+1 = RF;t+1 = t+1 A ~t + X L ~t ~t ; M (2) where t+1
Slide 41: The Basic Framework
The Basic Framework Assumption 1 Let "zt ln Z ~t =Zt , ^rF;t+1 ln (RF;t+1 =RF;t+1 ), and "t y = ln t = t . We assume that ^rF;t+1 , "zt , and "t y are zero-mean stationary processes with a small bounded support. 2 The deterministic trend components Zt admit a decomposition of the form z Zt = Z t + d
Slide 42: The Basic Framework
The Basic Framework Under these assumptions it is possible to derive a first-order approximation of (2) around (3), which satisfies: 1 nxat+1 nxat + rF;t+1 + nxt+1 ; (4) where nxat j a j "at l "lt + j x j "xt j m j "m t ; y nxt+1 j x j "xt+1 j m j "m t+1 "t+1 ; and (5) rF;t+1 j a j rt+1 a l l rt+1 w
Slide 43: The Basic Framework
The Basic Framework Hence, we can iterate equation (6) forward, and take conditional expectations to obtain +1 X h nxat Et [rF;t+h + nxt+h ] (6) h=1 Consider a country that is a net debtor (nxat < 0): External adjustment may come through future increases in net exports Et nxt+h > 0 (the trade channe
Slide 44: Empirical Methodology
Empirical Methodology In order to proceed we need estimates of z and . I Gourinchas and Rey (2007a) use the HP-filter to extract the trend used in calculating e.g. a and the cyclical component, at . Then calculate a = A=(A L) as a function of the HP-trends and employ the previously obtained cyclical
Slide 45: Empirical Methodology
Empirical Methodology We proceed in two steps: 1. Let wt = (nxat ; rt ; nxt )0 : Then a VAR(1) takes the form: wt+1 = Awt +vt ; from which we may obtain forecasts Et (wt+i ) = Ai wt : Then +1 X h nxat Et [rF;t+h + nxt+h ] ; h=1 may be expressed as e0nxa wt = e0r A (I A) wt e0 nx A (I A) wt = nxart +
Slide 46: Empirical Methodology
Empirical Methodology 1 2. Iterating forward for H periods nxat+1 nxat + rF;t+1 + nxt+1 ;, we obtain: XH XH h h H+1 nxat = rF;t+h nxt+h + nxat+H+1 : h=1 h=1 Then H X H X h h Var (nxat ) Cov (rF;t+h ; nxat ) Cov ( nxt+h ; nxat ) h=1 h=1 + H+1 Cov (nxat+H ; nxat ) which decomposes the variance of nxa
Slide 47: Empirical Findings (1): VAR Results
Empirical Findings (1): VAR Results 0.2 CANADA: nxat nxaΔnx nxart nxapred FRANCE: nxat nxaΔnx nxart nxapred t t t t 0.2 0.1 0.0 0.0 −0.1 −0.2 −0.2 1970 1980 1990 2000 1980 1990 2000 ITALY: nxat nxaΔnx t nxart nxapred t 0.50 JAPAN: nxat nxaΔnx t nxart nxapred t 0.2 0.25 0.0 0.00 −0.25 −0.2 −0.50 1980
Slide 48: Empirical Findings (2): Variance Decomposition
Empirical Findings (2): Variance Decomposition Variance Component Quarters: 1-4 Quarters: 5-8 Quarters: 9-12 Quarters: 13– Total (Variable) Panel A: Canada rF ( 100) 0.004 1.52% -0.003 -0.98% 0.003 1.00% 0.004 1.54% [t-stat] [0.873] [-0.600] [0.807] [0.471] nx ( 100) 0.125 45.21% 0.048 17.42% 0.054
Slide 49: Empirical Findings (2): Variance Decomposition
Empirical Findings (2): Variance Decomposition Variance Component Quarters: 1-4 Quarters: 5-8 Quarters: 9-12 Quarters: 13– Total (Variable) Panel D: Japan rF ( 100) 0.203 29.09% 0.041 5.93% -0.015 -2.10% 0.230 32.93% [t-stat] [1.814] [0.569] [-0.494] [1.768] nx ( 100) 0.365 52.25% 0.165 23.66% 0.036
Slide 50: Variance Decomposition-Cumulative Contributions
Variance Decomposition-Cumulative Contributions E:\FundStuff\Figures\NPVarDecModified.eps.gwg 12/02/09 01:28:18 0.10 CANADA: r - Contribution CANADA: Δ n x - Contribution F 1.0 0.05 0.00 0.5 -0.05 0.0 0 1 2 3 0 1 2 3 0.5 FRANCE: r - Contribution 2 FRANCE: Δ n x - Contribution F 0.0 1 0 0 1 2 3 0 1 2
Slide 51: Summary
Summary Using a newly developed dataset, we have examined the relative importance of the trade and valuation channels for the process of external adjustment. We found that the VALUATION channel is important for Japan, the U.S. and Germany (under some restrictive assumptions - results not shown here)