Oil and the macroeconomy14-01-2014
Claudio Morana has a new paper out investigating the relationship between oil prices and macroeconomic business cycles. Following below is an extended abstract, while you can read the full paper here.
Since the 1990s, the literature has focused on different features of the oil price-macroeconomy relationship, concerning its relevance, the presence of asymmetric responses, the sources of shocks and the role of the Great Moderation: to date, 10 out 11 postwar US recessions were preceded by sharp increases in oil prices (Hamilton, 2011).
Theoretically, an oil price shock may directly affect real activity through different demand and supply side channels. For instance, private consumption may contract due to discretionary income and precautionary savings effects, as households dispose of lower real income after paying energy bills and face a higher likelihood of future unemployment or longer unemployment duration (Edelstein and Kilian, 2009). Moreover, as higher and more volatile energy prices may lead consumers and firms postponing the purchase of durables and irreversible investment decisions, respectively (Bernanke, 1983; Pindyck, 1991), uncertainty effects may also be relevant; similarly through an operating costs channel, as consumers may postpone the purchase of durables (complement in use of energy) and firms save on capital usage, due to rising marginal costs and decreasing labor and capital productivity (Pindyck and Rotemberg, 1984; Rotenberg and Woodford, 1996).
Indirect effects, due to changes in consumption patterns and technologies, inducing capital/labor reallocation across sectors, may also be posited: if production factors are sector or product specific, an oil price shock, by shifting preferences in favor of more efficient durables and technologies, may then lead to labor and capital unemployment (Hamilton, 1988). Moreover, as the implementation of preemptive restrictive monetary policies might lead to stronger negative responses of real activity to oil price shocks than otherwise would occur (Bernanke et al., 1997), the monetary policy transmission mechanism may yield an additional source of asymmetry, relatively to the reallocation and uncertainty effects.
The empirical evidence on the above channels is not clear-cut. For instance support to the asymmetry feature of the oil price-macroeconomy relationship (Mork, 1989) is provided by Lee et al. (1995) and Hamilton (1996, 2003), showing that oil shocks are more likely to affect real activity when they occur in an environment of low oil price volatility or when they are not compensating previous price decreases. Moreover, consistent with the relevance of the reallocation effect, oil price shocks are found to affect differently and more strongly job destruction than job creation, and energy intensive sectors (automobile, chemicals, rubber and plastic) than other sectors (generating supply rather than demand shortages) by Engeman et al. (2011), Davis and Haltinwanger (1999), Lee and Ni (2002), Herrera et al. (2010), Ramey and Vine (2012). See also Ferder (1996), Elder and Serletis (2010), Lee et al. (2010) and Miller and Ni (2011) for supporting evidence concerning the uncertainty channel.
Yet, by comparing dynamic multipliers, measuring the response of US consumption, investment, and GDP to positive and negative energy price shocks, in support to the (symmetric) discretionary income, precautionary savings and operating costs mechanisms, no evidence of asymmetric responses is found by Edelstein and Kilian (2007, 2009) and Kilian and Vigfusson (2011a,b). Moreover, despite the evidence provided by Bernanke et al. (1997), Hamilton and Herrera (2004), Herrera and Pesavento (2009) and Kilian and Lewis (2011) dismiss any deepening effect of systematic monetary policy on US recessions since the 1970s.
Different explanations for the resilience of the global (world) economy to the recent oil price episode have also been provided, i.e. a declining oil share (Nakov and Pescatori, 2010), lower real wage rigidity (Blanchard and Galí, 2010; Blanchard and Riggi, 2009), lower volatility of oil demand and supply shocks (Baumeister and Peersman, 2009), and better anchoring of inflation expectations, in the face of demand driven oil price shocks (Kilian, 2010).
In the light of the contrasting empirical evidence and the different explanations provided in the literature for the economic effects of the 2008 oil price episode, the paper then assesses the recessionary effects of oil price shocks since the mid-1980s, by means of a large-scale factor vector autoregressive (F-VAR) model and a detailed description of global oil market-macro-finance interactions. The original contributions to the literature are as follows.
Firstly, in terms of model specification, the modeling approach allows for unobserved factors, related to global macro-financial conditions, estimated using data for fifty countries, including OECD and emerging countries; following the lead of Morana (2013), proxy variables for expectations about future global fundamentals and economic/financial fragility conditions, i.e., risk aversion, size, value, momentum, stock’s liquidity, and leverage factors, are also directly included in the information set; global oil reserves, production, consumption, (OECD) inventories, refineries margins, and proxies for oil price uncertainty and excess speculation in the oil futures market, are finally employed to account for physical and financial oil market conditions. To our knowledge this is the first paper in the oil price-macroeconomy literature to directly modeling market expectations about future fundamentals and to seek understanding of the global macro-financial effects of oil price shocks from a broad empirical perspective, involving real activity and fiscal/monetary policy responses, as well as labor and financial markets developments, within a joint assessment.
Secondly, the proposed modeling strategy grants the benefits of a richer description of oil market-macroeconomy interactions than achieved in previous small scale VAR investigations (Kilian, 2009; Baumeister and Peersman, 2008, 2009), still attained using the same tools of econometric policy analysis, and the multi-country/global perspective yield by dynamic panel data analysis, yet improving upon the latter in terms of empirical reliability: different from the fixed effect estimator for panel-VAR model, the PC-VAR approach implemented in the paper (Morana, 2012) neither shows downward bias nor is inconsistent when the coefficients on the lagged endogenous variables differ across countries (Pesaran and Smith, 1995; Holtz-Eakin et al., 1988).
Ten oil market structural shocks, related to both supply and demand side conditions, are then identified and their macro-financial consequences studied in details. New evidences on the oil market-macroeconomy relationship are shed, as we are unaware, for instance of previous results concerning the macro-financial effects of oil market non fundamental financial speculation shocks, as well as oil reserves and refineries margins shocks.
Thirdly, by controlling for the macro-financial factors driving flow and financial oil demand, more accurate identification of purely oil market supply side shocks should be achieved within the proposed framework; this feature is particularly desirable in the light of recent results, suggesting that recessionary effects should stem from supply driven oil price shocks only (Kilian, 2009).
The main results are as follows. Firstly, oil market shocks exercise stronger effects on macro-financial variables in the long- than in the short-term; for instance, figures for real activity are 20% and 10%, respectively. In particular, supply side disturbances yield the largest contribution to macro-financial fluctuations, i.e., 12% and 9% for real activity and (core) inflation, 30% to 35% for real stock and housing prices; consistent with their macro-financial effects, the latter shocks also sizably account for fluctuations in the policy variables, i.e. public expenditure (13%), liquidity (40%), and the real interest rate (7%).
Secondly, symmetric transmission mechanisms, as described by the discretionary income, precautionary savings and operating costs channels, are supported by the empirical evidence; similarly for the uncertainty channel, which, within the framework considered, is not necessarily asymmetric. Indeed, by comparing the effects of positive and negative net production shocks, weak evidence of asymmetric impacts on real activity can be found, the latter responding more strongly to negative than positive shocks in the very short-term only. Real effects of oil market speculative and consumption/inventories preferences shocks are also found.
Thirdly, the above mechanisms might account for the recessionary effects associated with some recent oil price shock episodes. In particular, during the first Persian Gulf War, oil market supply side shocks contributed to the 1990:2-1993:3 recession (-1.2%), and, at a lower extent, during the second Persian Gulf War, to the 2000:4-2003:2 recession (-0.24%). Oil market supply side conditions also exacerbated the recessionary effects of the subprime financial crisis in 2008 (-1.19%). Oil market speculative, preferences and volatility shocks also sizably contributed to slowing down real activity over the three episodes investigated.
Finally, we find that the resilience of the global economy to the 2008 oil price shock may be related to both an endogenous contraction in real wages and the implementation of expansionary stabilization policies, in the face of a mostly demand (macro)-driven oil price shock.