## wpiea2019237-print-pdf - Section 2 describes the literature and some empirical trends. Section 3 provides the data ## Source details **Canonical URL:** [wpiea2019237-print-pdf - Section 2 describes the literature and some empirical trends. Section 3 provides the data](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019237-print-pdf.pdf) ## Other formats - [Markdown version](/-/media/files/publications/wp/2019/wpiea2019237-print-pdf.pdf.md) - [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019237-print-pdf.pdf.json) --- ### Literature and empirical facts - Prior literature: - Barsky and Kilian (2004): monetary theory of oil shocks for the 1970ís. - Gillman and Nakov (2009): Granger predictability of nominal oil prices by ináation; model shows nominal oil prices jump with jumps in ináation to keep real return to oil (and gold) investment constant. - Alquist et al. (2013): robustness of causality from ináation to oil prices (1975–2009) and Granger predictability of M1 to oil prices. - Lucas and Nicolini (2015): use an adjusted M1 (M1MMDA) by adding money market deposit accounts from M2 to M1 to find a stable money demand function; analogous focus on relevant monetary aggregates here. - Visual/graphical observations (January 1, 1947 to May 1, 2017 unless otherwise noted): - Figure 1: natural log of US-dollar series shows close comovement of WTI spot oil price (solid) and Gold Fixing Price (grey); CPI (dashed) and Monetary Base (dotted) broadly comove with these commodity series. - Post-2008 jump in the monetary base comoves with oil price; both turn downwards nearly the same time in 2014. - WTI oil price fell from 106 in June 2014 to 47 in January 2015. - Excess Reserves (EXCSRESNS) peaked in August 2014 and dropped steadily thereafter; peaked at $2.7 trillion in August, 2014 (discussion in Section 6). - Figure 2: nominal and real WTI price series and endpoint-to-endpoint trend lines yield: - nominal endpoints: $1.62 to $52.50 → 52:51:62 / 1:62 = 31:4 → 3,141% increase. - real price endpoints: 7.54 to 19.88 → 19:887:54 / 7:54 = 1:64 → 164% increase. - average annual real oil price increase over 70 years ≈ 1:4% per year (annualized rate stated). - Figure 3: natural log of CPI and monetary base (1947–2017) indicates four distinct inflation "regimes" with arc-slope calculations: - 1-1-1947 to 1-1-1960: (3:383:076)/13 = 0:0234 → 2:3% annual average. - 1-1-1960 to 1-1-1982: (4:5473:38)/12 = 0:097 (Vietnam-war acceleration). - 1-1-1982 to 8-1-2008: (5:3874:547)/26:583 = 0:0554 → 5.54% average. - 10-1-2008 to 5-1-2017: (5:505:387)/8:75 = 0:013 → 1.3% average annual inflation. - Figure 4: oil/gold price ratio (gold data from 1968) appears noisy; tested for nominal causality. - Figure 5 and 6: monetary base, monetary base minus Central Bank Liquidity Swaps (SWP), Excess Reserves minus Swaps; subtracting Swaps shows Excess Reserves turned negative in April 2008 with a negative peak in October 2008; Swaps peaked at $580 Billion in December 2008 (Figure 6 caption). ### Data sources and series - Monthly data from FRED with FRED codes and sample periods as listed: - MB: Money Base (AMBSL, 1946m1 - 2017m4) - SWP: Central Bank Liquidity Swaps (SWPT, 2003m1 - 2017m5) - DEMDEP: Demand deposits (DEMDEPSL, 1959m1 ñ2017m4) - M1: M1 Money Stock (M1SL, 1959m1 ñ2017m4) - M2: M2 Money Stock (M2SL, 1959m1 ñ2017m4) - CPIE: CPI less Energy (CPILEGSL, 1957m1 ñ2017m4) - CPI: CPI for all urban consumers (CPIAUCSL, 1947m1 ñ2017m4) - WTI: Spot Crude Oil Price WTI (WTISPLC, 1946m1 ñ2017m4) - GOLD: Gold Öxing price, London Bullion (GOLDPMGBD228NLBM, 1968m4 ñ2017m5) - M1Divis: Monetary services Index M1 (MSIM1P, 1967m1 ñ2013m12) - M2Divis: Monetary services Index M2 (MSIM2, 1967m1 ñ2013m12) - EXCH: Trade Weighted U.S. Dollar Index: Broad (TWEXB, 1973m1 ñ2017m4) - EXP5Y: 5-Year Breakeven Ináation Rate (T5YIEM, 2003m1 - 2017m12) - EXPMICH: University of Michigan Ináation Expectation (MICH, 1978m1 - 2017m12) - EXCESS: Excess reserves of Depository Institutions (EXCSRESNS, 1984m2 - 2017m4) ### Testing methodology and data properties - Granger (1969) causality framework: estimate VAR - Yt = c0 + c1 Yt-1 + ... + cp Yt-p + d1 Xt-1 + ... + dp Xt-p + vt - Null H0: d1 = d2 = ... = dp = 0 (no Granger-causality) - Stationarity and integration: - Apply Augmented Dickey-Fuller (ADF) test for unit roots to determine order of integration. - If series are of differing integration orders, follow Toda and Yamamoto (1995) procedure (per Giles (2011)). - Sample and subperiod choices: - Full sample and subsamples include post-1973, post-1975, and a 2008:m9 breakpoint (when Swaps rose dramatically). - Integration results: - All series are integrated of order 1 (I(1)) except CPIE, EXP5Y, and EXPMICH which are I(2). - Standard Granger tests applied to I(1) pairs; Toda-Yamamoto used for I(2) → I(1) causality tests. ### Empirical results (Granger-causality findings) - Summary from Tables 1–4 (p-value significance indicated at <0.10; subscript "<" indicates reverse causality): - Monetary aggregates and oil prices: - MB: no Granger predictability for oil prices in reported subperiods. - MB-SWP (monetary base minus Swaps): shows Granger predictability to real oil prices in several subperiods (p values reported: e.g., Full Sample 0:0019 < 0:0602 etc. as tabulated). - M1: no Granger predictability in several subperiods; M1-SWP shows strong predictability with p-values below 1% for post-WWII subperiods and no reverse causality for 2008-2017. - M1-Divisia minus Swaps: similar predictability; M1-Divisia (without subtracting Swaps) does not. - MB+DEMDEP and MB-SWP+DEMDEP: MB+DEMDEP shows Granger predictability for the full sample only; MB-SWP+DEMDEP shows predictability in more subperiods (e.g., p-values in table). - Broader aggregates (M2, M2-SWP, M2-Divisia, M2-Divisia minus Swaps): generally no Granger predictability to real oil prices, except some cases when Swaps subtracted as detailed in tables. - Inflation and expectations: - CPIE (CPI less Energy): Granger predicts real oil prices for most subperiods (p-values in Table 1 show significance). - CPI (CPI for all urban consumers): Granger predictability for start of sample up to 2008 without reverse predictability; no predictability for 2008-2017. - EXP5Y (5-Year Breakeven Inflation Rate): shows Granger predictability with no reverse predictability for 2008-2017. - EXPMICH (Michigan one-year ahead expectations): predicts oil prices for 1991-2017 with reverse predictability; not predictive in 2008-2017 alone. - Gold and oil/gold ratio: - M1-SWP, M1-Divis-SWP, M2-SWP, M2-Divis-SWP show Granger-causality to real gold prices for some post-WWII subperiods. - M2-SWP shows causality for full sample, 1991-2017 and 2008-2017 to real gold prices (no reverse causality in 2008-2017). - Oil/gold price ratio causality results similar to oil price results; notable that in 2008-2017 the oil/gold ratio is Granger-predicted by M2-SWP while oil prices are predicted only by narrower aggregates. - Exchange rate and inflation: - MB and MB-SWP significantly cause the trade-weighted exchange rate index only in 2008-2017. - M1-SWP, M2-Divisia, and M2-Divisia-SWP Granger-predict exchange rate index in various subperiods (full sample, 1975-2017, 1991-2017). - CPI robustly Granger-predicts the exchange rate across subperiods. - MB and MB-SWP Granger-predict CPI inflation in many periods; MB-SWP shows stronger predictability to CPI in 2008-2017. - Additional numeric observations: - 5-Year Breakeven Inflation Rate fell from 2:03% on June 23, 2014 to 1:05% on January 13, 2015 (approximately halved). - WTI peaked on June 20, 2014 at $107.95 and fell to $44:08 on January 28, 2015, coinciding with collapse in inflation expectations. ### Interpretation and discussion - Main interpretations: - Nominal factors—money supply aggregates (especially when subtracting Central Bank Liquidity Swaps), CPI, and inflation expectations—Granger-predict real oil prices. - Money supply growth Granger-predicts inflation, consistent with prior findings. - Monetary factors also predict international gold prices, the oil-to-gold price ratio, and the U.S. dollar exchange rate index — indicating a broader nominal influence across asset prices. - Distinction between oil and gold: - Real gold prices build in shorter-term inflation expectations. - Real oil prices incorporate both long- and short-run inflation expectations. - For 2008-2017, only longer-term inflation expectations (EXP5Y) Granger-predict oil prices. - Post-Great Recession dynamics: - Subtracting Swaps reveals stronger monetary aggregate effects on asset prices. - In 2008-2017, broader aggregate (M2-SWP) provides useful information for expectations of inflation as reflected in the oil/gold ratio. - Interest on excess reserves may have sterilized part of the monetary base (preventing reserves from being lent out), giving M2 additional weight for short-term inflation expectations. - Speculative narrative on 2009–2014 oil shock: - Extra expected inflation may have been built into oil prices until 2014 but was not realized because excess reserves (partially sterilizing the monetary base) prevented monetary expansion from translating into realized inflation. - Excess reserves mirrored the level of Fed MBS purchases (MBS rose to $1:1 trillion by April 2010) and may have contributed to a temporary jump in inflation in late 2009 when reserves would have been lent absent sterilization. ### Conclusions and suggested next steps - Conclusions: - Evidence indicates nominal factors Granger-predict real oil prices, including during and after the Great Recession, once Swaps are subtracted from aggregates. - Post-2008 expansion of the monetary base increased inflation expectations; inflation expectations predict oil prices; collapse of both inflation expectations and oil prices in 2014 coincided. - Long-term inflation expectations Granger-predicted real oil prices with no reverse predictability in 2008-2017. - Suggested further research: - Incorporate monetary factors into Kilian's (2009) impulse response framework to test whether monetary variables help explain oil shocks when added to Kilian’s supply and demand components. *wpiea2019237-print-pdf (IMF working paper content: Sections 2–7 as provided).* ### Section 2 describes the literature and some empirical trends. Section 3 provides the data ### wpiea2019237-print-pdf - Section 2 describes the literature and some empirical trends. Section 3 provides the data ### Literature and empirical facts - Prior literature: - Barsky and Kilian (2004): monetary theory of oil shocks for the 1970ís. - Gillman and Nakov (2009): Granger predictability of nominal oil prices by ináation; model shows nominal oil prices jump with jumps in ináation to keep real return to oil (and gold) investment constant. - Alquist et al. (2013): robustness of causality from ináation to oil prices (1975–2009) and Granger predictability of M1 to oil prices. - Lucas and Nicolini (2015): use an adjusted M1 (M1MMDA) by adding money market deposit accounts from M2 to M1 to find a stable money demand function; analogous focus on relevant monetary aggregates here. - Visual/graphical observations (January 1, 1947 to May 1, 2017 unless otherwise noted): - Figure 1: natural log of US-dollar series shows close comovement of WTI spot oil price (solid) and Gold Fixing Price (grey); CPI (dashed) and Monetary Base (dotted) broadly comove with these commodity series. - Post-2008 jump in the monetary base comoves with oil price; both turn downwards nearly the same time in 2014. - WTI oil price fell from 106 in June 2014 to 47 in January 2015. - Excess Reserves (EXCSRESNS) peaked in August 2014 and dropped steadily thereafter; peaked at $2.7 trillion in August, 2014 (discussion in Section 6). - Figure 2: nominal and real WTI price series and endpoint-to-endpoint trend lines yield: - nominal endpoints: $1.62 to $52.50 → 52:51:62 / 1:62 = 31:4 → 3,141% increase. - real price endpoints: 7.54 to 19.88 → 19:887:54 / 7:54 = 1:64 → 164% increase. - average annual real oil price increase over 70 years ≈ 1:4% per year (annualized rate stated). - Figure 3: natural log of CPI and monetary base (1947–2017) indicates four distinct inflation "regimes" with arc-slope calculations: - 1-1-1947 to 1-1-1960: (3:383:076)/13 = 0:0234 → 2:3% annual average. - 1-1-1960 to 1-1-1982: (4:5473:38)/12 = 0:097 (Vietnam-war acceleration). - 1-1-1982 to 8-1-2008: (5:3874:547)/26:583 = 0:0554 → 5.54% average. - 10-1-2008 to 5-1-2017: (5:505:387)/8:75 = 0:013 → 1.3% average annual inflation. - Figure 4: oil/gold price ratio (gold data from 1968) appears noisy; tested for nominal causality. - Figure 5 and 6: monetary base, monetary base minus Central Bank Liquidity Swaps (SWP), Excess Reserves minus Swaps; subtracting Swaps shows Excess Reserves turned negative in April 2008 with a negative peak in October 2008; Swaps peaked at $580 Billion in December 2008 (Figure 6 caption). ### Data sources and series - Monthly data from FRED with FRED codes and sample periods as listed: - MB: Money Base (AMBSL, 1946m1 - 2017m4) - SWP: Central Bank Liquidity Swaps (SWPT, 2003m1 - 2017m5) - DEMDEP: Demand deposits (DEMDEPSL, 1959m1 ñ2017m4) - M1: M1 Money Stock (M1SL, 1959m1 ñ2017m4) - M2: M2 Money Stock (M2SL, 1959m1 ñ2017m4) - CPIE: CPI less Energy (CPILEGSL, 1957m1 ñ2017m4) - CPI: CPI for all urban consumers (CPIAUCSL, 1947m1 ñ2017m4) - WTI: Spot Crude Oil Price WTI (WTISPLC, 1946m1 ñ2017m4) - GOLD: Gold Öxing price, London Bullion (GOLDPMGBD228NLBM, 1968m4 ñ2017m5) - M1Divis: Monetary services Index M1 (MSIM1P, 1967m1 ñ2013m12) - M2Divis: Monetary services Index M2 (MSIM2, 1967m1 ñ2013m12) - EXCH: Trade Weighted U.S. Dollar Index: Broad (TWEXB, 1973m1 ñ2017m4) - EXP5Y: 5-Year Breakeven Ináation Rate (T5YIEM, 2003m1 - 2017m12) - EXPMICH: University of Michigan Ináation Expectation (MICH, 1978m1 - 2017m12) - EXCESS: Excess reserves of Depository Institutions (EXCSRESNS, 1984m2 - 2017m4) ### Testing methodology and data properties - Granger (1969) causality framework: estimate VAR - Yt = c0 + c1 Yt-1 + ... + cp Yt-p + d1 Xt-1 + ... + dp Xt-p + vt - Null H0: d1 = d2 = ... = dp = 0 (no Granger-causality) - Stationarity and integration: - Apply Augmented Dickey-Fuller (ADF) test for unit roots to determine order of integration. - If series are of differing integration orders, follow Toda and Yamamoto (1995) procedure (per Giles (2011)). - Sample and subperiod choices: - Full sample and subsamples include post-1973, post-1975, and a 2008:m9 breakpoint (when Swaps rose dramatically). - Integration results: - All series are integrated of order 1 (I(1)) except CPIE, EXP5Y, and EXPMICH which are I(2). - Standard Granger tests applied to I(1) pairs; Toda-Yamamoto used for I(2) → I(1) causality tests. ### Empirical results (Granger-causality findings) - Summary from Tables 1–4 (p-value significance indicated at <0.10; subscript "<" indicates reverse causality): - Monetary aggregates and oil prices: - MB: no Granger predictability for oil prices in reported subperiods. - MB-SWP (monetary base minus Swaps): shows Granger predictability to real oil prices in several subperiods (p values reported: e.g., Full Sample 0:0019 < 0:0602 etc. as tabulated). - M1: no Granger predictability in several subperiods; M1-SWP shows strong predictability with p-values below 1% for post-WWII subperiods and no reverse causality for 2008-2017. - M1-Divisia minus Swaps: similar predictability; M1-Divisia (without subtracting Swaps) does not. - MB+DEMDEP and MB-SWP+DEMDEP: MB+DEMDEP shows Granger predictability for the full sample only; MB-SWP+DEMDEP shows predictability in more subperiods (e.g., p-values in table). - Broader aggregates (M2, M2-SWP, M2-Divisia, M2-Divisia minus Swaps): generally no Granger predictability to real oil prices, except some cases when Swaps subtracted as detailed in tables. - Inflation and expectations: - CPIE (CPI less Energy): Granger predicts real oil prices for most subperiods (p-values in Table 1 show significance). - CPI (CPI for all urban consumers): Granger predictability for start of sample up to 2008 without reverse predictability; no predictability for 2008-2017. - EXP5Y (5-Year Breakeven Inflation Rate): shows Granger predictability with no reverse predictability for 2008-2017. - EXPMICH (Michigan one-year ahead expectations): predicts oil prices for 1991-2017 with reverse predictability; not predictive in 2008-2017 alone. - Gold and oil/gold ratio: - M1-SWP, M1-Divis-SWP, M2-SWP, M2-Divis-SWP show Granger-causality to real gold prices for some post-WWII subperiods. - M2-SWP shows causality for full sample, 1991-2017 and 2008-2017 to real gold prices (no reverse causality in 2008-2017). - Oil/gold price ratio causality results similar to oil price results; notable that in 2008-2017 the oil/gold ratio is Granger-predicted by M2-SWP while oil prices are predicted only by narrower aggregates. - Exchange rate and inflation: - MB and MB-SWP significantly cause the trade-weighted exchange rate index only in 2008-2017. - M1-SWP, M2-Divisia, and M2-Divisia-SWP Granger-predict exchange rate index in various subperiods (full sample, 1975-2017, 1991-2017). - CPI robustly Granger-predicts the exchange rate across subperiods. - MB and MB-SWP Granger-predict CPI inflation in many periods; MB-SWP shows stronger predictability to CPI in 2008-2017. - Additional numeric observations: - 5-Year Breakeven Inflation Rate fell from 2:03% on June 23, 2014 to 1:05% on January 13, 2015 (approximately halved). - WTI peaked on June 20, 2014 at $107.95 and fell to $44:08 on January 28, 2015, coinciding with collapse in inflation expectations. ### Interpretation and discussion - Main interpretations: - Nominal factors—money supply aggregates (especially when subtracting Central Bank Liquidity Swaps), CPI, and inflation expectations—Granger-predict real oil prices. - Money supply growth Granger-predicts inflation, consistent with prior findings. - Monetary factors also predict international gold prices, the oil-to-gold price ratio, and the U.S. dollar exchange rate index — indicating a broader nominal influence across asset prices. - Distinction between oil and gold: - Real gold prices build in shorter-term inflation expectations. - Real oil prices incorporate both long- and short-run inflation expectations. - For 2008-2017, only longer-term inflation expectations (EXP5Y) Granger-predict oil prices. - Post-Great Recession dynamics: - Subtracting Swaps reveals stronger monetary aggregate effects on asset prices. - In 2008-2017, broader aggregate (M2-SWP) provides useful information for expectations of inflation as reflected in the oil/gold ratio. - Interest on excess reserves may have sterilized part of the monetary base (preventing reserves from being lent out), giving M2 additional weight for short-term inflation expectations. - Speculative narrative on 2009–2014 oil shock: - Extra expected inflation may have been built into oil prices until 2014 but was not realized because excess reserves (partially sterilizing the monetary base) prevented monetary expansion from translating into realized inflation. - Excess reserves mirrored the level of Fed MBS purchases (MBS rose to $1:1 trillion by April 2010) and may have contributed to a temporary jump in inflation in late 2009 when reserves would have been lent absent sterilization. ### Conclusions and suggested next steps - Conclusions: - Evidence indicates nominal factors Granger-predict real oil prices, including during and after the Great Recession, once Swaps are subtracted from aggregates. - Post-2008 expansion of the monetary base increased inflation expectations; inflation expectations predict oil prices; collapse of both inflation expectations and oil prices in 2014 coincided. - Long-term inflation expectations Granger-predicted real oil prices with no reverse predictability in 2008-2017. - Suggested further research: - Incorporate monetary factors into Kilian's (2009) impulse response framework to test whether monetary variables help explain oil shocks when added to Kilian’s supply and demand components. *Italicized source: wpiea2019237-print-pdf (IMF working paper content: Sections 2–7 as provided).* ### References ### wpiea2019237-print-pdf - References ### References - Alquist, Ron and Kilian, Lutz and Vigfusson, Robert J., 2013. "Forecasting the Price of Oil," Handbook of Economic Forecasting, Elsevier. - Barsky, Robert B. and Kilian, Lutz, 2004. "Oil and the Macroeconomy Since the 1970s," Journal of Economic Perspectives, 18 (4, Fall):115-134. - Baumeister, Christiane and Kilian, Lutz . 2016, "Forty Years of Oil Price Fluctuations: Why the Price of Oil May Still Surprise Us", Journal of Economic Perspectives; 30 (1, Winter): 139-160. - Blinder, Alan S. and Lutz Kilian, 2009, "Comments and Discussion on Causes and Consequences of the Oil Shock of 2007ñ08.", Brookings Papers on Economic Activity, Vol. 2009, Spring; pp. 262-283. - Bordo, Michael D. and Humpage, Owen F., and Schwartz, Anna J. 2014. "The Evolution of the Federal Reserve Swap Lines since 1962", NBER Working Paper No. 20755, December. - Friedman, Milton, 1994, Money Mischief: Episodes in Monetary History, Harcourt Brace & Company, New York. - Giles, David, 2011. "Testing for Granger Causality", Econometrics Beat: Dave Gilesí Blog; April 29, 2011. - Gillman, Max & Nakov, Anton, 2009. "Monetary e§ects on nominal oil prices," The North American Journal of Economics and Finance, 20 (3, December): 239-254. - Gram, Phil and Saving, Thomas R. 2017. "A Booming Economy Will Challenge the Fed", Wall Street Journal, Dec. 13. - Granger, C. W. J. 1969. "Investigating Causal Relations by Econometric Models and Cross-spectral Methods." Econometrica, 37 (3, Aug.): 424-438. - Hamilton, J. D. 1983. "Oil and the macroeconomy since World War II". Journal of Political Economy, 91(2), 228ñ248. - Hamilton, J. D., 2009, "Causes and Consequences of the Oil Shock of 2007ñ08", Brook- ings Papers on Economic Activity, Vol. 2009, Spring; pp. 215-261. - Haug, A. A., and W. G. Dewald. 2012. ìMoney, Output and Ináation in the Longer Term: Major Industrial Countries, 1880 ñ2001.îEconomic Inquiry 50 (July), 773-787. - Kilian, Lutz, 2009. "Not All Oil Price Shocks Are Alike: Disentangling Demand and Supply Shocks in the Crude Oil Market," American Economic Review, vol. 99(3), pages 16 - 1053-1069, June. - Lucas, Robert Jr., 1988. "On the mechanics of economic development," Journal of Monetary Economics, 22 (1, July): 3-42. - Lucas, Robert E. and Nicolini, Juan Pablo, 2015. "On the stability of money demand," Journal of Monetary Economics, 73(C): 48-65. - Mabro, Robert, 1984. "On Oil Price Concepts", Oxford Institute for Energy Studies, WPM3. - Mankiw, Gregory, 2014. Principles of Economics, 7th Edition, South Western College Publishing, Nashville. - Toda, Hiro Y. and Yamamoto, Taku. 1995. "Statistical inference in vector autoregres- sions with possibly integrated processes". Journal of Econometrics, 66 (1-2): 225-250. - 17 *wpiea2019237-print-pdf - References* --- _Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019237-print-pdf.pdf_