Futures Markets, Oil Prices, and the Intertemporal Approach to the Current Account

Transcription

1 Futures Markets, Oil Prices, and the Intertemporal Approach to the Current Account LAMES November 21, 2008

2 Intertemporal Approach to the Current Account

3 Intertemporal Approach to the Current Account Dynamic, optimizing model of the current account suggesting that the persistence of income shocks play a key role in understanding the current account.

4 Intertemporal Approach to the Current Account Dynamic, optimizing model of the current account suggesting that the persistence of income shocks play a key role in understanding the current account. Two key challenges in testing the theory:

5 Intertemporal Approach to the Current Account Dynamic, optimizing model of the current account suggesting that the persistence of income shocks play a key role in understanding the current account. Two key challenges in testing the theory: 1 Identification of exogenous income shocks.

6 Intertemporal Approach to the Current Account Dynamic, optimizing model of the current account suggesting that the persistence of income shocks play a key role in understanding the current account. Two key challenges in testing the theory: 1 Identification of exogenous income shocks. 2 Distinguishing between permanent/persistent and transitory shocks.

7 My Solution: Petroleum Exporter-The Simplest Case

8 My Solution: Petroleum Exporter-The Simplest Case Petroleum exporter is small Price fluctuations constitute exogenous income shocks +

9 My Solution: Petroleum Exporter-The Simplest Case Petroleum exporter is small Price fluctuations constitute exogenous income shocks New approach to identifying persistent and transient innovations to petroleum prices using futures markets. + Yields a transparent framework to evaluate the implications of the intertemporal approach.

10 Key Results

11 Key Results The marginal propensity to consume out of persistent price shocks is significantly higher than the marginal propensity to consume out of transitory price shocks.

12 Key Results The marginal propensity to consume out of persistent price shocks is significantly higher than the marginal propensity to consume out of transitory price shocks. There is no evidence of a significant marginal propensity to consume out of transitory price shocks.

13 Key Results The marginal propensity to consume out of persistent price shocks is significantly higher than the marginal propensity to consume out of transitory price shocks. There is no evidence of a significant marginal propensity to consume out of transitory price shocks. When futures prices are not used in the identification of different price shocks evidence for the intertemporal approach is weaker.

14 Petroleum Prices Log of the spot price is assumed to be given by: p c,t = ψ t + χ t where ψ t is the permanent component and χ t is the transitory component. (Schwartz and Smith, 2000) and (Herce, Parsons and Ready, 2006) ψ t = µ c + ρψ t 1 + ε ψ,t χ t = φχ t 1 + ε χ,t Benchmark: Purely permanent (ρ = 1) and purely transitory shocks (0 < φ < 1)

15 Using Futures Prices to Identify Oil Price Shocks (1) Futures prices contain information about future spot prices: p c,t = ψ t + χ t f t,t+n = E t (ψ t+n + χ t+n ) ω n f t,t+n futures contract that expires in n periods ω n constant risk premium

16 Futures Term Structure with Permanent and Transitory Shocks phi=0.90 Impulse Response to Perm. and Trans. Shocks Transitory Component phi=0.95 Permanent Component

17 Using Futures Prices to Identify Oil Price Shocks (2) Express all the observables (spot and futures prices) as a state-space model and estimate the parameters of the model using maximum likelihood. p c,t = ψ t + χ t f t,t+n = nµ c + ρ n ψ t + φ n χ t ω n Calculate permanent and transitory components of petroleum prices using the Kalman Filter.

18 Data-Petroleum Prices Crude oil futures traded in NYMEX (since 1983). Monthly averages of spot prices, 3, 6, 9, 12 and 15 months ahead futures prices in the estimation. Data Source: West Texas Intermediate (WTI) spot price data from the Energy Information Administration. Futures prices are constructed using historical end of day futures price data from Price-Data.com

19 Permanent and Transitory Comp. of Petroleum Prices Price ($/barrel) Apr-83 Apr-84 Apr-85 Apr-86 Apr-87 Apr-88 Apr-89 Apr-90 Apr-91 Apr-92 Apr-93 Apr-94 Apr-95 Apr-96 Apr-97 Apr-98 Apr-99 Apr-00 Apr-01 Apr-02 Apr-03 Apr-04 Apr-05 Apr-06 Permanent Component Predicted Spot Price Actual Spot Price

20 Consensus Forecasts Expected Change In Spot Prices (% of Spot Prices) (12-month - 3-month Forecast) Oct Oct-90 Oct-91 Oct-92 Oct-93 Oct-94 Oct-95 Oct-96 Oct-97 Oct-98 Oct-99 Oct-00 Oct-01 Oct-02 Oct-03 Oct-04 Oct Consensus Forecasts Model

21 A Model Of the Current Account For a Petroleum Exporter

22 A Model Of the Current Account For a Petroleum Exporter 1 A country that only produces petroleum and consumes imported goods from the rest of the world.

23 A Model Of the Current Account For a Petroleum Exporter 1 A country that only produces petroleum and consumes imported goods from the rest of the world. 2 Optimization Problem: maxu = E t i=0 βi u(c t+i )

24 A Model Of the Current Account For a Petroleum Exporter 1 A country that only produces petroleum and consumes imported goods from the rest of the world. 2 Optimization Problem: maxu = E t i=0 βi u(c t+i ) 3 Period Budget Constraint: B t = (1 + r)b t 1 + Q C,t 1 C t 1

25 A Model Of the Current Account For a Petroleum Exporter 1 A country that only produces petroleum and consumes imported goods from the rest of the world. 2 Optimization Problem: maxu = E t i=0 βi u(c t+i ) 3 Period Budget Constraint: B t = (1 + r)b t 1 + Q C,t 1 C t 1 4 Intertemporal Budget Constraint: (1 + r)b t + i=0 (1/1 + r)i E t Q C,t+i = i=0 (1/1 + r)i E t C t+i

26 A Model Of the Current Account For a Petroleum Exporter 1 A country that only produces petroleum and consumes imported goods from the rest of the world. 2 Optimization Problem: maxu = E t i=0 βi u(c t+i ) 3 Period Budget Constraint: B t = (1 + r)b t 1 + Q C,t 1 C t 1 4 Intertemporal Budget Constraint: (1 + r)b t + i=0 (1/1 + r)i E t Q C,t+i = i=0 (1/1 + r)i E t C t+i 5 Income: Q C,t = X C,t (P C,t /P M,t )

27 Optimal Import Consumption Response C t = r 1 + r i=0 ( ) 1 i (E t E t 1) Q C,t+i 1 + r C t Q C,t 1 θ ψ ε ψ,t + θ χ ε χ,t + e t θ ψ 1 and θ χ 0

28 Petroleum Exporters Country % of Exports % of World Production OPEC Member ( ) ( ) since Nigeria Oman Angola Libya Congo Gabon Iran Venezuela Qatar Syria Algeria Ecuador Norway Cameroon Trinidad and Tobago Egypt Colombia Indonesia Mexico Average Data from UNCTAD Handbook of Statistics. Data from International Energy Annual 2004 published by Energy Information Administration.

29 Estimating the Marginal Propensity to Consume The innovations to permanent and transitory components are converted to annual frequency and adjusted for other exports. The marginal propensities to consume out of permanent and transitory shocks are estmated using the adjusted innovations ε ψ,t,i and ε χ,t,i : C t,i Q t 1,i = c i + θ 1 ε ψ,t,i + θ 2 ε χ,t,i + ɛ t,i

30 Estimates of Marginal Propensity To Consume C t,i Q t 1,i = c i + θ 1 ε ψ,t,i + θ 2 ε χ,t,i + ɛ t,i Sample θ 1 (Std. Error) θ 2 (Std. Error) θ 1 = θ 2 Num. of p-value Obs. 1 All Countries 0.329*** (0.127) (0.147) Excluding * (0.196) (0.180) Excluding Iran 0.345*** (0.110) (0.125) Excluding Norway, Nigeria, 0.334*** (0.104) (0.122) Iran, Venezuela and Mexico 5 Opec Members (0.256) (0.296) Other Petroleum Exporters 0.331*** (0.073) (0.082) Fixed effects were incorporated in all the regressions even though their values are not reported in the table. Pooled OLS estimates with correlated panels corrected standard errors. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level

31 MPC Estimates-With and Without Long Horizon Futures Prices Marginal Propensity To Consume (With and Without Futures Prices) C t,i Q t 1,i = c i + θ 1 ε ψ,t,i + θ 2 ε χ,t,i + ɛ t,i Parameters 3 Month Futures Prices Full Set of Futures Prices θ *** (0.125) (0.127) θ (0.209) (0.147) θ 1 = θ 2 (p-value) Fixed effects were incorporated in all the regressions even though their values are not reported in the table. Pooled OLS estimates with correlated panels corrected standard errors. *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level

32 Marginal Propensity to Import Out of Oil Price Shocks Parameter MPC P MPC T r = 0.04, ρ = r = 0.04, ρ = r = 0.04, ρ = r = 0.06, ρ = r = 0.06, ρ = r = 0.06, ρ =

33 Implications

34 Implications Persistence of shocks is important in explaining current account fluctuations.

35 Implications Persistence of shocks is important in explaining current account fluctuations. Despite the fact that petroleum is highly durable, fluctuations in petroleum prices are mostly transitory.

36 Implications Persistence of shocks is important in explaining current account fluctuations. Despite the fact that petroleum is highly durable, fluctuations in petroleum prices are mostly transitory. Most petroleum price shocks therefore have a large effect on the current accounts of petroleum exporters.

37 Caveats

38 Caveats Non-Tradables, Other Exported Goods

39 Caveats Non-Tradables, Other Exported Goods Correlation between petroleum price shocks and import price shocks Re-importing refined petroleum Pass-through to the prices of imported goods

40 Caveats Non-Tradables, Other Exported Goods Correlation between petroleum price shocks and import price shocks Re-importing refined petroleum Pass-through to the prices of imported goods Correlation between petroleum price shocks and output of petroleum OPEC Investment in drilling and exploration

41 Conclusions

42 Conclusions The marginal propensity to consume out of permanent shocks is higher than the marginal propensity to consume out of transitory shocks.

43 Conclusions The marginal propensity to consume out of permanent shocks is higher than the marginal propensity to consume out of transitory shocks. There is no evidence of a significant marginal propensity to consume out of transitory shocks.

44 Conclusions The marginal propensity to consume out of permanent shocks is higher than the marginal propensity to consume out of transitory shocks. There is no evidence of a significant marginal propensity to consume out of transitory shocks. When futures prices are not used in the identification of different shocks, the evidence for the intertemporal approach is weaker.

45 Research Agenda Extension to other commodities. Long-run fluctuations in commodity prices and the real exchange rate dynamics in commodity exporting countries.

46 Petroleum Prices-With and Without Futures Prices Price ($/barrel) Apr-83 Apr-84 Apr-85 Apr-86 Apr-87 Apr-88 Apr-89 Apr-90 Apr-91 Apr-92 Apr-93 Apr-94 Apr-95 Apr-96 Apr-97 Apr-98 Apr-99 Apr-00 Apr-01 Apr-02 Apr-03 Apr-04 Apr-05 Apr-06 With Futures Prices Without Futures Prices With 3 Month Futures

47 Crude Oil Prices (January 1995-October 2008) $ per barrel Jan 95 Jun 95 Nov 95 Apr 96 Sep 96 Feb 97 Jul 97 Dec 97 May 98 Oct 98 Mar 99 Aug 99 Jan 00 Jun 00 Nov 00 Apr 01 Sep 01 Feb 02 Jul 02 Dec 02 May 03 Oct 03 Mar 04 Aug 04 Jan 05 Jun 05 Nov 05 Apr 06 Sep 06 Feb 07 Jul 07 Dec 07 May 08 Oct 08 permanent component spot price

48 Why Petroleum? Many countries have a large fraction of their exports come from petroleum. Commodity Average % of Exports Number of Countries Crude Petroleum Cocoa 47 3 Cotton 41 7 Copper 37 3 Coffee 35 12

49 Mincer-Zarnowitz Forecast Efficiency Regressions p c,t+n p c,t = α + β(f t,t+n p c,t ) + ε t Future α β R 2 Num. of α=0 and β=1 (std. error) (std. error) Obs. p-value 3 month (0.010) (0.377) 6 month (0.014) (0.246) 9 month (0.016) (0.200) 12 month (0.021) (0.174) 15 month (0.021) (0.173) * Standard errors are HAC standard errors.

50 Properties of Different Futures Prices Var (Change in ln(future Prices)) Contract Maturity

51 Estimates of the Model Parameters For Petroleum Prices Parameter Estimate Std. Error φ (0.0023) µ (0.0020) σψ (0.0002) σχ (0.0005)

52 Marginal Propensity to Import Out of Oil Price Shocks (1) We identify innovations to p c,t+i and want to write of shocks to petroleum prices: C t Q C,t 1 r(1 + µ q ) 1 + r (1 + µ q )ρ ε ψ,t + Ct Q C,t 1 in terms r(1 + µ q ) 1 + r (1 + µ q )φ ε χ,t + e t where e t contains all the innovations to other components of income. Given the estimate for φ and under reasonable assumptions for µ q and r: r(1+µ q) 1+r (1+µ q)ρ 1 if ρ = 1 and r(1+µ q) 1+r (1+µ q)φ 0.

53 Adjusting Price Shocks Before we can estimate the following reduced form equation using OLS: C t Q C,t 1 = c + θ 1 ε ψ,t + θ 2 ε χ,t + e t Need to account for other exports ( ) C t QC,t 1 = c + θ 1 ε ψ,t Q t 1 Q t 1 Need to aggregate to annual frequency. ( QC,t 1 + θ 2 ε χ,t Q t 1 ) + e t

54 Data UN National Income Accounts to construct measures of real imports and exports. UN COMTRADE and UNCTAD Handbook Statistics databases for the value and quantity of commodity exports

55 What Difference Futures Prices Make? When futures prices are not included, the predicted permanent component is larger. Futures prices help identifying transitory shocks. Parameter Without Futures With Futures φ (0.0023) (0.0023) µ (0.0037) (0.0020) σψ (0.0039) (0.0002) σχ (0.0038) (0.0005) Var( s t ) due to ε ψ

56 Model Fit Future Mean Error Mean Absolute Error Spot month month month month month

57 Parameter Estimates of the Empirical Model For Petroleum Prices Parameter Estimate Std. Error φ (0.0023) µ (0.0020) σψ (0.0002) σχ (0.0005) ω (0.0026) ω (0.0017) ω (0.0025) ω (0.0038) ω (0.0032)

58 Marginal Propensity to Consume-Individual Countries Country θ 1 (Std. Err.) θ 2 (Std. Err.) R 2 Nigeria (0.530) (0.500) Oman (0.161) (0.199) 0.06 Angola (0.320) (0.345) Libya (0.206) (0.250) Congo (0.207) (0.247) 0.11 Gabon (0.301) (0.275) Iran (0.429) (0.419) 0.11 Venezuela (0.313) (0.413) Qatar (0.318) (0.299) 0.08 Syria (0.280) (0.256) 0.10 Algeria (0.341) (0.415) Ecuador (0.305) (0.505) Norway (0.123) (0.135) Cameroon (0.559) (0.480) Trinidad and Tobago (0.727) (1.352) Egypt (0.432) (0.724) Colombia (0.990) (0.770) 0.17 Mexico (0.663) (1.580) Indonesia (0.683) (1.564) Number of observations is 22 for all countries except Syria which has 20 observations. A constant was included in all regressions even though their values are not reported in the table.

59 [petroleum exporters] have spent a smaller share of their latest windfall on imports of goods and services than during previous oil shocks,... even though the futures markets expect oil prices to stay high. Recycling the Petrodollars, Economist, November 10, 2005.

60 If we had assumed a process for Q C,t with a purely permanent and a purely transitory component, we would have: C t = ε P,t + r 1 + r ε T,t where ε P,t : permanent shock and ε T,t : transitory shock

61 Market Commentaries: Collapse of OPEC Quotas in 1986 Futures prices predict large permanent shocks Market awaits meeting of OPEC ministers, but no action to stabilize prices expected... Meanwhile, more companies slash budgets, staff. March 10, No big oil price rebound seen after decline March 17, OPEC struggles to prop up oil prices March 24, Source: Oil & Gas Journal

62 Market Commentaries: Gulf Crisis, August-September 1990 Futures prices predict large transitory shocks.... Other analysts believe the situation will stabilize soon and prices will again return to roughly pre-invasion levels... Oil prices approaching $ 50/bbl could not be sustained for long. Even a price approaching $ 30/bbl probably would bring into play market forces that would undermine that price level. August 13, 1990 Source: Oil & Gas Journal

63 Market Commentaries: Price Hikes of Futures prices predict a large permanent component in the price hikes. Oil s new era February 21, Oil prices establish new, higher plateau, analysts say May 9, Source: Oil & Gas Journal

64 A Wrinkle Derived for level, need implication for logs (Campbell and Deaton (1989)): C t = r ( ) 1 i 1 + r Σ i=0 (E t E t 1)Q C,t+i 1 + r C t r(1 + µ q) Q C,t 1 r µ q ( ) 1 + i µq (E t E t 1) q c,t+i i=0 1 + r where q c,t+i = p c,t+i p m,t+i + y c,t+i, q c,t = log(q C,t ), p c,t+i = log(p C,t ), p m,t+i = log(p M,t ) and y c,t+i = log(y C,t )