Changes in macroeconomic variables have been attempted to be hypothesized only to be limited by econometric and data related limitations. However, over the years economists have studied possible avenues to explain the changes in these variables using naturally occurring changes and their relationships with any resulting changes in macroeconomic variables. Such a field was thought of by Sims with impacts on macroeconomic variables in general and pursued by Hamilton and others who studied oil prices in particular and how they affected macroeconomic variables.
Sims, in his paper ‘Macroeconomics and Reality’ (1980), uses six variables : i) Money Supply ii) Real GNP iii) Unemployment iv) Wages v) Price Level vi) Import Prices. He treats all the variables as endogenous because he believes that all variables move simultaneously and that the economy is dynamic. Sims discusses a general strategy for estimating profligately parameterized macromodels, and presents results for a relatively small scale application. In order to pursue such an approach, Sims first step was to develop a class of multivariate time series models which will serve as the unstructured first-stage models.
What Sim actually did was to fit to quarterly an unconstrained vector autoregression the postwar time series for the US. And West Germany on money, GNP, unemployment rate, price level, and import price index. Since the model being estimated by Sim is an autoregression, the distributed theory on which the tests are based is asymptotic. It follows chi-squared distribution for the likelihood ratio test statistics. Out of the six data series used in the model for each country, each series except unemployment was logged, and the regressions all included time trends. Sims uses the impulse response functions to describe the movements of an economy overtime given a shock to the system.
He says that analysis of the system’s response to typical random shocks appears to be the best descriptive device. The ‘typical shocks’ whose effects are discussed in the paper are positive residuals of one standard deviation unit in each equation of the system. In the paper, the residuals are also referred to as the ‘innovations’. In order to be bale to see the distinct patterns of movement the system may display, it was useful for Sim to transform them to orthogonal form.
Hamilton (1983) discusses the causes of the recessions in the US after world war 11. His study focused on the correlation between the dramatic increases in the prices of crude petroleum and the recession that followed just about every time the prices would increase. He took this starkly evident coincidence under his study in order to see if its merely a coincidence the increases in the prices of the petroleum truly is a contributing factor to the recession of the US economy.
The poor performance of the US economy has been very well documented. Poor economic performance always coincided with the period of rapidly rising energy prices. The energy prices and the recession were not only secularly correlated, but their correlation was cyclical. Hamilton in his paper also discusses three possible hypotheses that could explain this coincidence.
Hypothesis 1: The correlation represents a historical coincidence; that is, the factors truly responsible for recession just happened to occur at about the same time as the oil price increases.
Hypothesis 2: The correlation results from an endogenous explanatory variable; that is, there is some third set of influences that in fact caused both the oil price increases and the recessions.
Hypothesis 3: At least some of the recessions in the US prior to 1973 were causally influenced by an exogenous increase in the price of crude petroleum.
Hamilton bases his investigation using Sims’s six variable quarterly vector autoregressive model. He uses very straightforward and uncontroversial methods to test the first hypothesis. He simply applies the traditional tests for absence of statistical correlation. He tests the second hypothesis using two complementary methodologies. The first focuses on institutional and historical detail and the second on statistical evidence. He talks about the historical events leading up to the increase in oil prices in detail. Further, he talks about his second methodology through which he sought to test for the endogeneity of crude oil prices based on the suggestion of Granger. The second hypothesis has the statistically refutable implication that no other series should ‘Granger-cause’ oil prices.
He further asserts that ‘if instead hypothesis 2 is true- that is, if some third set of variables in fact caused both the oil prices increase and the recessions- then one should be able to identify unusual behavior in some of the key macro series in evidence prior to the oil price increases, which could have contributed significantly to the prediction of subsequent changes in oil prices.'(Hamilton). The Granger Causality test is useful in determining whether one time series is important in forecasting another time series.
At the starting point of the analysis, Hamilton examines the role of oil in a version of six-variable system which Sims initially suggested as the compact approximation to macroeconomic reality. Rigorous work has followed and various regressions have been done. Different hypothesis were tested for granger causality. The results were statistically proven. Seven out of eight postwar recessions in the US have been preceded by a dramatic increase in the price of crude oil. The first hypothesis seems to be having few grounds for claiming that the correlation between oil prices and output represents just a statistical coincidence. For the second hypothesis, there was little support for the proposition that over the period 1948-72, some third set of influences was responsible for both oil price increases and the subsequent recession.
Mork, on the other hand, extends the findings of Hamilton and further investigates the correlation between the oil prices and GNP growth in the US Data. Hamilton had demonstrated a strong relation between oil prices and recession in the US. Mork’s work essentially focuses on the fact that Hamilton’s extensive study has mainly pertained to a period in which all the large oil price movements were upwards and doesn’t answer the question whether the correlation exists in periods of price decline. He also used the model presented by Sim’s making the required changes to it. His strategy mainly included adding the real price of oil to the six-variable equation and then testing this specification for the stability of the coefficients before and after the collapse of the oil market in 1985-86.
Then he allows for an asymmetric response to oil price changes by specifying real price increase and decrease as separate variable and test the individual significance, symmetry of the responses and stability of this specification over the sample. Mork, however believed that PPI used by Hamilton for crude oil is misleading because it reflects only the controlled prices of domestically produced oil. Mork then derives a separate index to measure crude oil which also includes the refiner acquisition cost (RAC).
However, the data tested by Mork do not identify any significant effects of oil price declines. They do show convincing results that the effects of oil price declines are different from those of price increases. Mork also claims ‘these results confirm that the negative correlation with oil price increases is not an artifact of Hamilton’s data. On the other hand, an asymmetry in the responses is quite apparent in the correlation with price decreases is significantly different and perhaps zero. Further research is needed to verify this view empirically.’ (Mork)
Hooker(1996) too studied the correlation between the oil price changes and the macroeconomic variables. However, in his study, he claimed that oil prices no longer Granger cause many US macroeconomic indicator variables in data after 1973. Hooker argues that it is ironical that the usage of oil prices as an explanatory variable has increased even though the significance of the oil’s impact on the economy seems to have dramatically decreased. Hooker in his paper tries to explore three possible explanations for his claim using the VAR methodology that Hamilton had originally established except that he used aa VAR(5), using separate models for GDP and unemployment.
The first is that there were structural breaks in many US macro series around 1973 and failing to account for this weakens the role of oil. The second is that oil prices were exogenous before 1973, but no longer are, so the granger causality assigns a smaller and more specification-specific role to oil prices. The third is that there are important asymmetries between the effects of oil price increases and decreases. Oil prices are entered two different ways: in nominal log-differences and in real log-levels.
Three different tests are performed: the basic Chow test for whether each coefficient in one subsample is equal to its counterpart in the other, a modified Chow test which does not restrict the intercept to remain fixed across subsamples, and a third leaving oil prices out of the VAR. The results are such that they’re dramatically different from the findings of other published papers on similar topic. Neither unemployment nor the fact that oil prices Granger cause a variety of U.S. macroeconomic indicator variables in data up to 1973 but not in data from then to the present is shown to be robust. A number of potential explanations are investigated: that sample stability issues are important, that oil prices are now endogenous to the U.S. economy, and that linear VAR equations misspecify the nature of the oil
price-macroeconomy interaction. None of these hypotheses are supported by the data.(Hooker)
Hamilton published an article in the same year, responding to Hooker’s claim with a new measure the ‘net increase in oil price’. He shows that by using the new variable as a measure for oil prices, we actually reject the null hypothesis when testing for a structural break in 1973 by testing a change in oil price coefficients. He also shows that unlike what hooker shows, using the same VAR and lags as hooker, net oil prices do granger cause before 1973 as well as over the whole sample period(unlike what hooker shows). He argues that the new measures support morks claims where he lacks because it makes up for the rapid changes (volatility) in oil prices that might alter the actual impacts of an increase in oil prices with a following decrease. He supports any insignificant results of his for the post 1973 period with the unavailability of data instead of it proving any thing. He supports his views from his 1983 paper that oil prices shocks do precede recessions.
Further studies were under taken by Cunado for European countries. In that paper, the impact of oil price on inflation and Industrial production index is reviewed using different proxies of oil prices( national real price, price relative to other commodities for the whole world, +ve changes in oil prices). Possible role of exchange rate fluctuations is discovered which explains localised effects of oil price changes magnified or minimized by weaker or stronger currencies respectivley. Also short term and long terms impacts of oil prices are studied on the inflation and IPI. Oil prices’ impacts on inflation and IP index are calculated over 1960-1999.
Cointegration tests show a relationship between oil prices and inflation for UK and Ireland but they are suspected to be because of exchange rate fluctuations and not a long term relationship between the two. However, allowing for structural breaks in the period shows cointegration for most of the countries, showing a longterm relationship between the two which is also proved by granger causality tests, that show that oil prices granger cause inflation rates. No such relationship of cointegration is found between oil prices and IPI index which is used as a measure of the economic activity. Granger causality tests between the two do show a short term relationship for some countries.
This is further clarified when proxies for oil prices are used. When positive changes in oil prices and NOPI are used instead of oil prices, granger causality of IPI by them is proved for most of the countries. Further, positive and negative changes in oil prices are used that prove the non linear nature of the grager causality and the asymmetric nature of the relationship. This study supports Hamilton’s hypothesis of NOPI being a better estimate to study oil price impacts. Trivariate VAR tests including inflation in the regression are also done that show that oil prices affect the economy by means other than the inflation.
Jimenez et al(2004) study similar relationships in industrialized OECD countries, reviewing impacts of oil price(measured in real terms using Brent prices relative to US producer price index) changes on oil exporters and importers. Using a VAR(7) with quarterly data over four models (linear,asymmetric,scaled and NOPI), They discover that circumstantial conditions (changes in importer/exporter status, exchange rate catastrophes, major economic transformations) aside, oil exporters follow a opposite pattern of causality compared to exporters. They experience a increase in GDP at times of price hikes and vice versa. Oil importers follow the normal impact as discussed by others above. This paper however claims that granger causality is true for both ways for oil prices and GDP growth for most of the countries, a result specific to this author’s data.
He also studies the performance of different models, concluding that non linear models give more accurate results for most countries, something proven by the Hamilton and Cunado. This study also shows that non volatile markets show more precise and greater impacts of oil price changes which has something to do with certainty levels in the respective markets. Theoretically too, scaled models that allow for more controlled variables are bound to show more precise impacts.
Therefore, Oil prices are a very important determinant of macro economic variables. As is shown by a number of economists, they at the very least granger cause economic activity which ties them up with the occurrence of recessions. As Hamilton’s theory stands to be supported yet again with the most recent recession in the United states, it is not wise to ignore any indication provided by oil prices even though direct causality cannot be proven. More data with newer techniques can help define the relationship more precisely and also relate to oil importer/exporter status. Strategies to study the relationship over the years have varied in terms of the VARs and lags studied which makes comparison tricky. However, these differences are not enough to discredit any of them even though results are vary susceptible to the empirical strategy used.
Sims, Christopher A. “Macroeconomics and Reality” Econometrica 48 (1980) 1-48.
Hamilton, James D. “Oil and the Macroeconomy Since World War II” The Journal of Political economy 91 (1983), pp. 228-248.
Mork, Knut Anton. “Oil and the Macroeconomy When Prices Go Up and Down: An Extension of Hamilton’s Results” The Journal of Political Economy 97 (1989), pp. 770-744.
Hooker, Mark A. “What Happened to the Oil Price- Macroeconomy Relationship?” Journal of Monetary Economics 38 (1996), pp. 195-213.
Hamilton, James D. “This is What Happened to the Oil Price-Macroeconomy Relationship” Journal of Monetary Economics 38 (1996), pp. 215-220.
Hamilton, James D. “What is an Oil Shock?” Journal of Econometrics 113 (2003), pp. 363-398.
Cunado, Juncal and Fernando Perez de Gracia. “Do Oil Price Shocks Matter? Evidence for Some European Countries” Energy Economics 25 (2003), pp. 137-154.
Jimenez-Rodriguez, R and Marcelo Sanchez. “Oil Price Shocks and real GDP growth: Empirical Evidence for Some OECD countries” Applied Economics 37 (2005), pp. 201-228.