The Stock Market Reaction to Oil Price Changes

Sridhar Gogineni
Division of Finance
Michael F. Price College of Business
University of Oklahoma
Norman, OK 73019-0450

March 13, 2008

Abstract

I explore the reaction of the stock market as a whole and of different industries to daily oil price
changes. I find that the direction and magnitude of the market‟s reaction to oil price changes depend
on the magnitude of the price changes. Oil price changes most likely caused by supply shocks have a
negative impact while oil price changes most likely caused by shifts in aggregate demand have a
positive impact on the same day market returns. In addition to the returns of oil-intensive industries,
returns of industries that do not use oil to any significant extent are also sensitive to oil price changes.
Finally, I show that both the cost-side dependence and demand-side dependence on oil are important
in explaining the sensitivity of industry returns to oil price changes.

I am indebted to Louis Ederington. I am grateful for the helpful comments received from Chitru Fernando,
Vahap Uysal, Cynthia Rogers, Carlos Lamarche, Shawn Ni, Vikas Raman, Veljko Fotak, Jesus Salas, Ginka
Borisova and Anthony May. I also acknowledge support from the Center for Financial Studies and the Summer
Research Paper support fund at the University of Oklahoma. All errors are my own.
 21 September 2006. “Unchanged Rates, Oil-Price Dip Rally Stocks”

17 January 2007 “Dow Clears Another Record As Oil Prices Continue to Fall”

17 May 2007 “Industrials Push to New High on Deal Optimism, Oil‟s Fall”

19 October 2007 “Dow Slips and Dollar Slides As Oil Jumps, Job Claims Rise”

(Headlines from The Wall Street Journal)

As the above headlines illustrate, in the recent months the popular financial press has talked

repeatedly about how oil price changes are impacting the stock market. In fact, during the years 2005

and 2006, oil prices figured in the headlines1 of The Wall Street Journal on 204 days and a majority

of these attribute stock price movements the previous day to oil price changes. As reviewed below, a

considerable economics literature has been devoted to study the long-term impact of oil prices on

macroeconomic variables such as inflation, growth rates, and exchange rates. However, despite the

attention oil prices receive on a regular basis, there is very little research in the finance literature on

how the stock market reacts to oil price changes. While the financial media assumes that the stock

market is strongly influenced by oil prices, no one has measured how strong the relation is and what

factors influence this relation.

Petroleum is an essential energy source in the US accounting for 40% of total energy

requirements. In 2006, the total demand for oil in dollar terms was approximately $506 billion, which

is equivalent to 3.8% of US GDP.2 Given the importance of oil, its short-term demand price

inelasticity, and the attention oil prices receive in the financial press, an understanding of the impact

of oil price changes on market returns is essential to market participants. In this paper, I investigate

the impact of oil price changes on the stock market as a whole and on individual industries from a

financial markets perspective.

1
The search terms used are “oil prices”, “oil price”, “oil prices and stocks” and “oil price and stocks”.
2
US GDP in 2006 was $13.21 trillion. Average price per barrel and average demand were $66 and 21 million
barrels per day in 2006. This is projected to increase to 28 million barrels per day by 2030. Sources:
http://www.eia.doe.gov;https://www.cia.gov/library/publications/the-world-factbook/print/us.html

2
 The first goal of the paper is to provide a systematic investigation of the impact of oil price

changes on the US stock market. Using daily data from 1983 to 2006, I find that the overall relation

between the stock market and daily oil price changes is weak, suggesting that the financial media

overplays the supposed impact of oil prices. Further, I find that the market reaction to oil price

changes depends on the magnitude of price change. That is, I find that large oil price changes over a

one-day horizon and oil price changes during war periods have a large negative impact on the same

day market returns. On the other hand, the stock market is positively correlated with small daily oil

price changes. I do not find evidence of asymmetry in the market‟s reaction to oil price increases and

decreases. Further, I find no evidence for any over or under-reaction of the market to oil price

changes, which suggests that the market is efficient in responding to daily oil price changes.

The second goal of this paper is to examine the impact of oil price changes on the returns of

individual industries. I hypothesize and prove that the sensitivity of industries‟ returns to oil price

changes depends both on their cost-side dependence and demand-side dependence on oil, and that the

effect of these factors varies across industries. These results are a possible explanation for the finding

that in addition to the returns of oil-intensive industries3 such as the air transportation industry and the

freight transportation industry, returns of industries that do not use oil to any significant extent are

also sensitive to daily oil price changes.

To my knowledge, this is the first paper to examine the impact of daily oil price changes on

the returns of a wide array of individual industries. By doing so, I attempt to address a topic of

interest to both academics and practitioners alike. It is widely accepted that the stock returns of oil-

intensive industries are more sensitive to oil price changes than those of non oil-intensive industries.

However, no study has documented these largely subjective views of the market participants. Also,

3
In accordance with the popular financial press, I define oil-intensive industries as those with high cost-side
dependence on oil. That is, industries that use significant amount of oil in their production process. As
explained in Section IV, the cost-side and demand-side dependence measures are calculated from the
benchmark input-output accounts released by the bureau of economic analysis.

3
this is the first paper to examine how the magnitude of an industry‟s correlation with oil price changes

depends on its cost-side dependence and demand-side dependence on oil.

This paper contributes to the literature in several other ways as well. I try to differentiate

between the impact of daily oil price changes likely caused by supply shifts and those caused by

changes in expectations about future economic activity. It seems likely that most large oil price

changes over a one-day horizon are due to supply shifts. In other words, changes in expectations of

future economic activity are likely spread out over time so that their impact on oil prices in any one

day is small. On the other hand, oil supply changes and anticipated changes due to hurricanes, civil

unrest, pipeline ruptures and refinery stoppages could lead to large one-day changes in oil prices.4 If

this were the case, results indicate that the market reacts negatively to oil price changes caused by

supply shocks and positively to oil price changes caused by changes or expected changes in aggregate

demand.

Finally, as Table I illustrates, most studies examine the impact of oil price changes on

macroeconomic variables other than the stock market; and those that examine the impact on the stock

market look at the relation over longer periods while the financial press posits a same day reaction. I

use daily returns data and investigate the impact of oil price changes on the market and industry

returns from a financial markets perspective. Oil prices are reported daily and since stock prices

usually respond quickly to relevant public information, using daily data provides a more precise and

timely measure of the impact of oil price changes on the market.

The rest of the paper is organized as follows. Section II discusses related literature. Section

III presents testable hypotheses. Section IV presents the data and defines the sample. Section V

discusses the empirical results. Section VI concludes the paper.

4
While it is reasonable to assume that most big jumps in oil prices are due to supply shifts, small oil price
changes are probably due to both supply and demand shifts. However, it is impossible to distinguish between
small price changes caused by small supply shifts and those caused by changes in expectations of future
economic activity. Even though there is no theoretical argument to say that demand shifts dominate for small oil
price changes, it seems to be the case empirically.

4
II. Related Literature

A long line of empirical work in economics finds that oil price increases negatively impact

measures of macroeconomic activity. To review some of the most important studies, Hamilton (1983)

documents a significant negative relation between quarterly oil price changes and future GDP growth

in the United States. Hamilton also finds that all but one of the US recessions since World War II

have been preceded, typically with a lag of three quarters, by a dramatic increase in the price of crude

petroleum. Subsequent research by Gisser & Goodwin (1986) and Darby (1982) largely confirms

Hamilton‟s findings while Burbidge & Harrison (1984) report similar, although slightly weaker,

results using data from five OECD countries.5 Mork (1989) extends Hamilton‟s results and

documents an asymmetric relation between quarterly oil price changes and output growth. However,

Hooker (1996) finds that the oil price-macroeconomic relationship and the evidence for asymmetric

impact of oil price changes on output growth are considerably weaker in the 1990‟s.

Golub (1983) examines exchange rate reactions to oil price changes and notes that a

country‟s dependence on imported oil and the direction of wealth transfer associated with the oil price

change explains the reaction. Among the many other studies finding that oil price shocks impact the

economy are Davis & Haltiwanger (2001), and Keane & Prasad (1996) for employment effects;

Hamilton & Herrera (2004), Bernanke, Gertler & Watson (1997) and Barsky & Kilian (2001) on the

role of monetary policy responses to oil price shocks; Lee & Ni (2002) on demand and supply effects

on industries and Hooker (2002) on the inflationary effects of oil price shocks. Hamilton & Herrera

(2004) provide a comprehensive list of studies conducted on oil shocks.

A much smaller finance literature addresses the issue of whether stock market reactions to oil

price shocks are rational and whether oil prices have any predictability. Jones & Kaul (1996) test

whether the reaction of international stock markets to quarterly oil price shocks can be justified by

current and future changes in real cash flows. They find that US and Canadian stock markets are

rational while Japanese and UK stock markets tend to over-react to oil price shocks. Using monthly

5
US, Japan, Germany, UK and Canada.

5
data, Sadorsky (1999) finds that an oil price shock has a negative and statistically significant initial

impact on stock returns.6 Huang, Masulis and Stoll (1996) document significant causality from oil

futures prices to stock returns of individual firms, but not to aggregate market returns. On the other

hand, Chen, Roll & Ross (1986) find that the risk associated with oil price changes is not priced in the

stock market.

Three recent papers have examined whether future stock market returns can be predicted

based on past oil price changes. Dreisprong, Jacobsen & Maat (2003) test if monthly oil price changes

can predict stock market returns worldwide. Using data from eighteen developed and thirty emerging

markets, they find significant predictability in twelve of the eighteen developed markets. Emerging

markets show the same effect, though with less significance. Hong, Torous & Valkanov (2002)

document a negative relation between lagged petroleum industry returns and the U.S stock market

returns. Pollet (2002) finds that expected changes in oil prices can be used to predict excess market

returns as well as excess returns for most U.S industries.

A recent working paper by Bittlingmayer (2005) comes closest in spirit to this paper.

Bittlingmayer (2005) finds that oil price increases associated with war risk cause larger declines in

stock prices and larger increases in treasury yields than oil price increases associated with other

causes. My paper contributes to the literature by analyzing the differential impact of oil price changes

caused by likely changes in supply of oil and likely changes in expectations of future economic

activity. More specifically, I find that the stock market returns are positively correlated with oil price

changes likely caused by changes in aggregate demand. Moreover, I examine the impact of daily oil

price changes on the returns of individual industries, a question of interest to both academics and

practitioners that has not been addressed by the literature.

6
This impact lasts for approximately 4 months. In a related paper, Papapetrou (2001) estimates that real stock
returns in Greece are affected negatively by oil price increases.

6
III. Hypotheses

A. Relation between oil price changes and market returns

While a number of studies examine the relation between oil price changes and the stock

market over long horizons, if investors believe oil has an important impact on the economy, then oil

price changes should impact the stock market almost immediately as stock prices usually respond

very quickly to public information. Also, financial commentators often attribute negative (positive)

stock market movements to oil price increases (decreases) at roughly the same time. As mentioned

earlier, during the years 2005 and 2006, oil prices figured in the headlines of The Wall Street Journal

on 204 days. Given the apparent presumption in the financial press that oil prices strongly impact the

stock market, it is surprising that little research has been conducted to measure the impact of daily oil

price changes on market returns. This leads to my first hypothesis.

Hypothesis I: The stock market reacts negatively to oil price changes.

It is important to recognize that the presumed negative relation between the stock market

returns and oil price changes might be moderated or partially offset by times when oil prices and

stock prices move together. More specifically, if the market expects a boom in future economic

activity, both oil and stock prices are likely to rise and, if the market decides a recession is likely,

both oil and equity prices are likely to fall. So changes in oil and equity prices due to shifts in

expectations regarding future economic activity imply a positive correlation. On the other hand,

changes in oil prices due to supply shocks imply a negative correlation because these price changes

increase operating costs of industries where oil is a major input.

It seems likely that most large oil price changes over a one-day horizon are due to supply

shifts. In other words, changes in expectations of future economic activity are likely spread out over

time so that their impact on oil prices in any one day is small. On the other hand, oil supply changes

and anticipated changes due to wars, hurricanes, civil unrest, pipeline ruptures and repairs, refinery

stoppages and repairs, and sharp inventory changes could lead to large one-day changes in oil prices.

Appendix I lists the fifteen largest one-day oil price increases and decreases during the sample period

7
and it seems most of these changes are a result of supply shifts. If most large oil price changes are due

to shifts and anticipated shifts in supply, then we should observe a stronger negative correlation

between oil price changes and market returns when the former is large. This leads to my second

hypothesis.

Hypothesis II: While large oil price changes have a large negative impact on market returns,

the impact of small oil price changes is positive.

B. Market efficiency and oil price changes

If markets are efficient and investors correctly anticipate the impact of oil price changes on

the economy, then stock prices should adjust almost instantaneously so that these changes have no

predictive ability. On the other hand, if investors underestimate (overestimate) the true impact of an

oil price increase on the economy, then as the true impact becomes clear, stock prices will fall (rise)

in the period following the oil price change. This implies that oil price changes would have predictive

ability. Findings of previous studies indicate that oil price changes do have some degree of predictive

ability. For example, Jones and Kaul (1996) find that quarterly stock returns of most countries in their

sample including the U.S. are negatively affected by both current and lagged oil price variables.

Similarly, Pollet (2002) and Dreisprong, Jacobsen & Maat (2003) find that monthly oil price changes

have predictive ability for excess market returns and returns of most US industries. A possible

explanation is that only the stocks of industries directly dependent on oil react immediately to oil

price changes and others do not, as investors might not be able to assess the impact of oil price

changes on these industries. As they realize the wider effects of the price change, the returns of these

industries will exhibit negative correlation with lagged oil price changes. In the light of these

findings, I formulate my next hypothesis as follows.

Hypothesis III: The stock market under- reacts to daily oil price changes.

In an interesting observation, Mork (1989) using quarterly data finds an asymmetric relation

between oil prices and output growth and presents evidence that GNP growth displays a statistically

8
significant negative correlation with oil price increases and an insignificant correlation with oil price

decreases. In a related study, Mork, Olsen and Mysen (1994) confirm the asymmetry in oil price

effects on the growth rates of other OECD countries.7 If oil prices have an asymmetric effect on

growth rates and investors recognize this, it is possible that oil prices have an asymmetric impact on

stock prices as well. Hence, motivated by the results of Mork (1989) and Mork, Olsen & Mysen

(1994), I test for an asymmetry in the market‟s reaction to oil price changes. Also, Brown, Harlow &

Tinic (1988) and Campbell & Hentschel (1992) show that the stock price reaction to unfavorable

news events tends to be larger than the reaction to favorable events. Assuming an oil price increase is

an unfavorable event and an oil price decrease is a favorable event,8 I hypothesize the following.

Hypothesis IV: The stock market‟s reaction to oil price increases is larger than its reaction to

oil price decreases.

C. Sensitivity of industry returns to oil price changes

As mentioned earlier, not all industries are equally dependent on oil. At least, that seems to be

the presumption of the popular press. It is common to find news articles highlighting the impact of oil

price changes on the returns of oil-intensive industries such as the air transportation industry. On the

other hand, it is uncommon to find articles that highlight the impact of oil prices on the returns of non

oil-intensive industries such as the computers industry and the telecommunications industry. The

presumption behind these stories is that oil price hikes lead to an increase in the costs of production

for oil-intensive industries thereby hurting their profits. Table III lists the 81 industry groups used in

this study in descending order of their dependence on oil for production processes. And as Column 2

of Table III indicates, industries such as air transportation depend heavily on oil for their operations

whereas the opposite is true for industries such as computer manufacturers or telecommunications.

7
Oil price increases seem to slow down economic growth in the U.S. to a greater extent than in Germany,
France and Japan, all of which are more dependent on imported oil than the U.S.
8
Furthermore, it seems oil price increases receive more attention in the financial press than oil price decreases.
For example, during 2005, news related to oil price increases figured in the headlines of The Wall Street Journal
on 74 days while news related to oil price decreases figured in the headlines on 47 days.

9
 The small academic literature examining the impact of oil shocks on industries includes Lee

and Ni (2002) who study the long-run supply and demand effects of oil shocks using industry level

data.9 Hong, Torous & Valkanov (2002) and Pollet (2002) study whether monthly market and

individual industry returns can be predicted using oil prices. To the best of my knowledge, there are

no studies in the finance literature that quantify the cost-side effects of oil price changes on individual

industries‟ returns. Given this fact and the presumptions made by the financial press, I hypothesize

the following:

Hypothesis V: There is a positive relationship between an industry‟s cost-side dependence

on oil and its sensitivity to oil price changes.

It is important to recognize that while the cost-side effects of oil prices are important, they do

not fully explain the sensitivity of industries‟ returns to oil price changes. In other words, to gauge the

overall impact of oil price changes, we need to take into account how oil prices affect the demand for

an industry‟s products. That is, even if a particular industry does not use oil to any significant extent

in its production process, its customer industries might be sensitive to oil price changes,10 which in

turn cause the original industry‟s returns to be sensitive to oil prices as well. For example, in spite of

using less oil in its production process, returns of the electric lighting and wiring equipment industry

(number 39 in Table III) exhibit a strong negative correlation with daily oil price changes. This is

probably because the main customers of this industry include automobile manufacturers, air

transportation and retail trade, all of which are heavily dependent on oil. This observation leads to my

next hypothesis:

Hypothesis VI: There is a positive relationship between an industry‟s demand-side

dependence on oil and its sensitivity to oil price changes.

9
In this paper, the authors use monthly industry level output data for a select group of industries.
10
It is important to realize that while the returns of most customer industries are negatively correlated with oil
price changes, returns of some industries might be positively correlated with oil price changes.

10
IV. Sample Data and Descriptive Statistics

Daily value-weighted returns of NYSE/NASDAQ/AMEX index are obtained from CRSP. I

use these returns as a measure of stock market returns. Daily cash price data of light crude oil are

obtained from a data vendor, Normans historical data.11 Daily value-weighted industry returns data

are calculated using individual firm returns obtained from CRSP. The sample period spans from April

1983 to December 2006. Data used to calculate industries‟ own dependence and their customers‟

dependence on oil are obtained from the Benchmark Input-Output accounts published by the Bureau

of Economic Analysis.

After matching daily market returns with the corresponding oil returns, there are 5,948 daily

observations. Table II presents the descriptive statistics for oil price returns and market returns. Both

the mean oil return and the mean market return are close to zero. However, the standard deviation of

daily oil returns is 2.4%, nearly one and a half times higher than that of market returns. Panel B

presents the autocorrelation patterns of the oil and market returns. Consistent with existing

evidence,12 daily market returns exhibit significant positive first-order autocorrelation and a negative

second-order autocorrelation. On the other hand, daily oil returns exhibit weak autocorrelation

patterns with significant second and third-order autocorrelations.

I use the benchmark input-output (IO) accounts published by the Bureau of Economic

Analysis13 to calculate industries‟ dependence on oil. In particular, I rely on the Use table of the

benchmark accounts.14 Using this data and the methodology described in Shahrur (2005), I compute

an industry‟s cost-side dependence on oil as follows.

11
Cash price data is the actual settlement price used at NYMEX. Results are robust to using crude futures price
data rolled by volume. www.normanshistoricaldata.com
12
See, for example, Campbell, Lo and McKinlay (1997).
13
http://www.bea.gov/bea/dn2/i-o.htm. The names of the file and table used are NDNO313 and IOusesum.xls
and are available for download at the Bureau of Economic Analysis website.
14
For any pair of supplier and customer industries, the Use table reports estimates of the dollar value of the
supplier industry‟s output that is used as input in the production of the customer industry‟s output.

11
 Column 2 of Table III presents the measures of cost-side dependence on oil for the 81

industry groups used in this study.15 Industries are listed in descending order of their cost-side

dependence on oil. While it is relatively straightforward to calculate estimates of industries‟ cost-side

dependence on oil, it is not the case with estimates of industries‟ demand-side dependence on oil. A

possible solution is to consider the extent to which an industry‟s customers depend on oil. That is,

even if a particular industry does not use oil to any significant extent, its customer industries might

depend on oil heavily. And as oil price increases, these customer industries are hurt which in turn

affects the demand for the products of the supplier industry. However, it might be the case some

customer industries might benefit from oil price increases. Therefore results should be read with

caution. Column 3 of Table III presents a measure of the importance of oil to an industry‟s customers.

For each industry, I identify the main customers based on the methodology described in Fan and Lang

(2000) and Shahrur (2005).16 Customer dependence on oil is then calculated as follows: I multiply the

proportion of output sold to each of the main customer industries with their respective cost-side

dependencies on oil. The sum of these values is used as a measure of customer dependence on oil. I

use this variable as a proxy for industry‟s demand-side dependence on oil.17

Value-weighted returns of each industry group are calculated as follows: Daily closing price,

shares outstanding and returns data for each firm in an industry are obtained from CRSP.18 Relative

weights of each of the firms are calculated as market value of the firm divided by total market value

of the industry. Weighted returns are then calculated as the product of relative weights and returns of

15
I exclude new construction, federal government enterprises, state and local government enterprises, non-
comparable imports, scrap, used and second hand goods and general government industry groups due to the
unavailability of returns data from CRSP. Further, I exclude petroleum refining and related products, metallic
ores mining, coal mining, crude petroleum and natural gas and gas production and distribution (utilities)
industries. This is because for these industries, an oil price increase will have a negative effect on the supply
side and a positive effect on the demand side. It is therefore difficult to distinguish between these two
contradicting effects.
16
Any industry that buys 3% or more of supplier industry‟s output is classified as a main customer. As in
Shahrur(2005) I repeat this process with 1% and 5% cutoff points. Results did not vary significantly and are
available on request.
17
A major limitation of this measure is that data on main customers are missing for some industries. For
instance, the input-output tables do not have information about households that are a major customer group for
industries such as restaurants and hotel, retail etc.
18
SIC codes used to identify firms in an industry are available on request.

12
the individual firms. The sum of all the individual weighted returns gives the valued weighted return

for an industry on a given day.

V. Empirical Results

A. Relation between oil price changes and market returns

As a first step towards exploring the effects of oil price changes on market returns, I estimate

the following regression:

Rst Rot t (1)

where Rst is the return on the value weighted NYSE/NASDAQ/AMEX index and Ro t is the oil

return on day „t‟. As results presented in Panel A of Table IV indicate, market returns covary

negatively and significantly with oil price changes providing support for hypothesis I. However, oil

price changes are associated with small market returns on average. For example, the estimated

coefficients imply that a 5% rise in oil prices is associated with a market decline of about 9.3 basis

points on average. Further, R-squared is less than 1% indicating only a small portion of the variation

in daily stock market returns is explained by variation in daily oil prices. It seems that despite the

attention received by media, on average the impact of oil price changes on the stock market is small

and economically insignificant.

It may be the case that while the market reacts negatively to oil price changes caused by shifts

in the supply of oil, oil prices and the market likely react similarly to shifts in the market‟s

expectation of future aggregate demand and economic activity. So shifts in oil and equity prices due

to shifts in expectations regarding future economic activity imply a positive correlation while shifts in

oil prices due to changes in expected oil supply imply a negative correlation. I present some evidence

to support this. It seems likely that most large oil price shifts over one-day horizons are due to supply

shifts. In other words, changes in expectations of future economic activity are likely spread out over

time so that their impact on oil prices in any one day is small. On the other hand, oil supply changes

13
and anticipated changes due to wars, hurricanes, civil unrest, pipeline ruptures and repairs, refinery

stoppages and repairs, and sharp inventory changes could lead to large one-day changes in oil prices.

While it is reasonable to assume that most big jumps in oil prices are due to supply shifts, small oil

price changes are probably due to both supply and demand shifts. However, it is impossible to

distinguish between small price changes caused by small supply shifts and those caused by changes in

expectations of future economic activity. Even though there is no theoretical argument to say that

demand shifts dominate for small oil price changes, it seems to be the case empirically. Appendix I

lists the fifteen largest daily oil price increases and decreases over the sample period and it is evident

that most of these price changes are a result of supply shocks or news pertaining to supply

disruptions.

If most large oil price changes are due to shifts and anticipated shifts in supply, then we

should observe a stronger negative correlation between oil price changes and market returns when the

former is large. I find fairly strong evidence that this is the case. More specifically, I estimate the

following regression to document the relation between the stock market returns and the magnitude of

oil price changes:

Rst 1 Rot 2 ( Rot * Rot ) t

(2)

where Rot is the oil return in absolute terms. All other variables are defined earlier. 2 < 0 indicates

that large oil price changes have a large negative impact on market returns. Results are presented in

the first two rows of Panel B, Table IV. The coefficient estimate on 2 is negative and significant

providing support to hypothesis II that large oil price changes have a large negative impact on the

stock market returns.

In order to provide further evidence for hypothesis II, I examine the impact of large and small

oil price changes by dividing the sample into two parts based on the magnitude of the price change. I

14
replace Rot in equation (2) with a dummy variable „D‟ that equals 1 for the top 25% of absolute oil

returns19 and 0 otherwise. More specifically, I estimate the following regression:

Rst 1 Rot 2 ( D * Rot ) t (3)

The estimated coefficients and the associated t-statistics are presented in the third and fourth

rows of Panel B. That estimated coefficients imply that the market reacts positively to small oil price

changes and negatively to large and small oil price changes over a one-day horizon. Similar results

were found when the top 20% of absolute oil returns are used as a cutoff. Results are not presented

for brevity. If most large daily oil price changes are due to supply shocks, which seem to be the case

empirically, then results provide some evidence that it is the disruptions or expected disruptions in the

supply of oil that is a major concern to the market. Along the same lines, positive correlation between

market returns and oil price changes when the latter is small indicate that these changes are likely the

result of changing expectations about future economic activity.

An alternative way to confirm the negative impact of supply shock induced oil price changes

is to examine the market reaction to oil price changes during the periods when most daily oil price

changes are caused due to supply shocks: US involvement in an armed conflict in the Middle East

provides an ideal example. During a war, one would expect most oil price changes to be caused by

changing expectations regarding oil supply. This is because the Middle East accounts for nearly 41%

of the world oil production and any unrest in this region increases the uncertainty regarding oil

supply. To document the effects of oil price changes caused due to oil supply, I estimate the following

regression:

Rst 1 Rot 2 ( war * Rot ) t (4)

here war is a dummy variable that equals 1 if the US is involved in an armed conflict in the Middle

East.20 Assuming that war proxies for periods in which most oil price changes are caused due to

19
Oil return cutoff for the top 25% in absolute terms is 2.14%.

15
changes in expectations regarding oil supply, 2 < 0 reestablishes the claim that the market reacts

negatively to oil price changes caused by supply disruptions. Results are presented in Panel C of

Table IV. It is evident that the stock market‟s reaction to daily oil price changes is more negative

during periods of war than other times. The estimated coefficients imply that a 5% rise in oil prices is

associated with a market decline of about 26.4 basis points on average during the war periods, almost

three times higher than the 9.3 basis point decline implied from an earlier result when such a

distinction isn‟t made.

To summarize, while oil price changes and market returns are negatively correlated, oil price

changes are associated with small market returns on average. However, I find that large oil price

changes and oil price changes during periods of war have a large negative impact on market returns.

If most large oil price changes over a one-day horizon and most oil price changes during periods of

war proxy for oil price changes caused due to supply disruptions, these results can be interpreted as

the stock market being very sensitive to oil price changes caused by supply shifts.

B. Market efficiency and oil price changes

In this section, I examine whether the market fully captures the information in oil price

changes and whether there is an asymmetry in the market‟s reaction to oil price increases and

decreases. As discussed above, some previous studies argue that past oil price changes have

predictive power for future market returns. This implies that the market under-reacts at the time to

information in oil price changes (hypothesis III). To test this, I estimate the following regression:

Rst 1 Rot 2 Rot 1 3 Rot 2 4 Rot 3 5 Rot 4 6 Rot 5 t (5)

i <0 for i > 1 indicates under-reaction while i > 0 indicates over-reaction. As results in Panel A of

Table V indicate, there is no evidence of under-reaction or over-reaction of the market to daily oil

price changes. My results, therefore, contradict the findings of Pollet (2002) and those of Dreisprong,

20
I use media reports to identify the periods when the US is involved in a war.

16
Jacobsen & Matt (2005). A possible explanation is that these studies use monthly data whereas I use

daily returns data. While oil prices have predictability over longer horizons, it seems that the stock

market is efficient in responding to daily oil price changes.

Brown, Harlow & Tinic (1988) show that the stock price reaction to unfavorable news events

tends to be larger than the stock price reaction to favorable events. As mentioned earlier, the

asymmetric response of macroeconomics factors to oil price increases and decreases suggests that in

general oil price increases are viewed as unfavorable events. Along the same lines, I test for an

asymmetry in the stock market‟s reaction to oil price changes (Hypothesis IV). More specifically, I

estimate the following regression:

Rst 1 Rot 2 ( Rot * D) t (6)

Here D is a dummy variable that is equal to 1 if oil return is positive on day „t‟ and 0 otherwise. All

other variables are defined earlier. The effect of oil price decreases on stock returns is 1 and the

effect of oil price increases is measured by 1 2 . i <0 and 1 2 > 2 in absolute terms

indicates that the market reacts more negatively to oil price increases than to oil price decreases. As

results in Panel B of Table V indicate, there is no evidence of an asymmetry in the market‟s reaction

to oil price increases and decreases. It seems the market does not interpret all oil price increases as

bad news.

C. Sensitivity of industry returns to oil price changes

Next, I examine the impact of oil price changes on the returns of individual industries. As a

first step, I estimate regression specification (7) for each of the 81 industry groups used in this study

Rit 1 Rot 2 Rst t (7)

Here Rit is the return of industry „i‟ on day „t‟. Rot is the oil return and Rst is the value-weighted

market return on day „t‟. I include market returns as an independent variable as it allows for the

17
estimation of oil‟s incremental impact. That is, 2 picks up the impact of changes in expected

aggregate demand or economic activity on the industry so 1 picks up the impact of changes in oil

prices not caused by changes in expected future economic activity.

I report the slope coefficient and the associated t-statistic on oil returns variable in columns 3

and 4 of Table III. As speculated by the financial media, daily returns of oil-intensive industries such

as the air transportation industry exhibit a high negative correlation with oil price changes. For

example, a 5% increase in oil prices is associated with a negative stock return of 0.26% or 26 basis

points for the air transportation industry. On the other hand, it is surprising to see that the returns of a

few oil-intensive industries such as agricultural products, industrial and other chemicals and electric

utilities etc. are not very sensitive to oil price changes. Industries in the bottom quartile of Table III

(the non oil-intensive group) mostly span the services, financial and telecommunication sectors. It is

interesting to note that returns of most industries in this group react negatively to oil price changes as

well and some of them are more sensitive to oil price changes than oil-intensive industries. For

example, stock returns of the drug industry and insurance industry (numbers 64 and 81 respectively)

are more sensitive to oil price changes than the returns of paints and allied products industry (number

13) which is relatively more oil-intensive.

To sum, of the 81 industries, stock returns of 65 are negatively correlated with daily oil price

changes (out of these, 37 industries exhibit a statistically significant correlation at the 10% level). In

the analysis that follows, I measure the extent to which the magnitude of industries‟ correlation with

oil prices depends on their cost-side dependence on oil (hypothesis V). More specifically, I estimate

the following cross-sectional regression:

ˆ cos t side dependencei (8)

i ,1 1 i

Here, ˆ is the coefficient estimate on oil return for industry „i‟ obtained from regression
i ,1

specification (7). It measures the average change in that industry‟s return given a 1% change in oil

18
prices. Large negative estimates of 1 imply that for industries exhibiting a negative correlation with

oil price changes, an increase in the cost-side dependence on oil makes the correlation more negative.

Along the same lines, for industries exhibiting a positive correlation with oil price changes, an

increase in the cost-side dependence on oil makes the correlation less positive. Results are presented

in Panel A of Table VI and provide support to hypothesis V. The estimated coefficients imply that a

10% increase in the cost-side dependence leads to a 1.09% or 109 basis point change in the

magnitude of an industry‟s correlation with oil price changes.

As mentioned earlier, while cost-side dependence on oil is important, it is not the only factor

explaining the magnitude of industries‟ correlation with oil price changes. In other words, even if a

particular industry does not use oil to a significant extent, its customer industries might depend on oil

heavily. And as oil price increases, these customer industries are hurt which in turn affects the

demand for the products of the supplier industry. To test how the magnitude of industries‟ correlation

with oil price changes is affected by their demand-side dependence on oil (hypothesis VI), I estimate

the following regression:

ˆ cos t side dependencei demand side dependencei (9)

i ,1 1 2 i

where demand-side dependence is weighted average of the cost-side dependencies of an industry‟s

main customer industries presented in Table III. This measure is used as a proxy for that industry‟s

demand-side effects of oil price changes. Large negative estimates of 2 imply that for industries

exhibiting a negative correlation with oil price changes, an increase in the demand-side dependence

on oil makes the correlation more negative. However, as results presented in Panel B of Table VI

indicate, this seems not to be the case.21 A possible explanation for this finding is that data on

individuals and households who are the major customers for some industries such as the retail

industry is absent in the benchmark input-output tables. Hence, there is a possibility that the demand-

21
Similar results were obtained in a regression without the cost-side dependence as an independent variable.
Results are not presented for brevity

19
side dependence might be under-representing the true measure of demand-side effects of oil price

changes.

While analyzing the role played by industries‟ cost-side and demand-side dependencies in

explaining the correlation between their returns and oil price changes, it is important to recognize that

the effects of these factors may vary across industries. That is, the effect of these factors on industries

that exhibit a high negative correlation with oil price changes might be different from those that

exhibit a positive correlation with oil price changes. In order to capture these variations, I use the

quantile regression technique pioneered by Koenker and Bassett (1978). This technique allows us to

estimate the effect of the independent variables on the outcome of interest not only in the center of the

distribution but also in the upper and lower tails of the conditional distribution of the response

variable. By using OLS estimation we can analyze the effects of cost-side dependence and demand-

side dependence on the magnitude of industries‟ sensitivity to oil price changes, but the estimators

summarize the conditional mean effect. In other words, OLS is not informative about how these

factors impact the lower and upper tails of the conditional distribution of industries‟ sensitivities to oil

price changes. More specifically, I estimate the following regression specification:

ˆ | xi' cos t side dependencei demand side dependencei (10)

i ,1 1 2 i

for the set of quantiles = {0.25, 0.5, 0.75}. For example, for = 0.25, the quantile regression

estimator could be considered to estimate the effect of the independent variables on the lower tail of

the conditional distribution of ˆi ,1 . As the results presented in Panel C of Table VI indicate, the

impact of cost-side dependence and demand-side dependence varies as we move from lower quanitles

to upper quantiles. An intuitive way to interpret these results is the following: as previously

documented in Table III, returns of some industries are negatively correlated while returns of others

are positively correlated with daily oil price changes. The estimated coefficients for = 0.25 in

Panel C, Table VI indicate that for industries that have a high negative correlation between their

returns and oil price changes, the magnitude of the correlation depends on both cost-side and demand-

20
side dependence. Negative coefficient estimates indicate that oil price increases negatively affect

these industries both on the supply side and demand side. On the other hand, estimates for = 0.75 in

Panel C, Table VI indicate that for industries that have a slight negative or positive correlation

between their returns and daily oil price changes (the upper quantile), the magnitude of correlation

depends only on demand-side dependence. On average, the magnitude of correlation increases as the

demand-side dependence increases. It is possibly because the main customers of these industries gain

from oil price increases. For example, returns of the pipeline, freight forwarders and related services

industry (number 23 in Table III) are positively correlated with daily oil price changes (i.e., it

corresponds to the upper quintile of ˆi ,1 ). The main customers of this industry include petroleum

refining and petro-chemical industries which profit from oil price increases. This in turn, increases the

demand for the products offered by the pipeline, freight forwarders and related services industry.

VI. Conclusion

This paper provides a comprehensive study of the stock market‟s reaction to oil price

changes. Considering the current level of interest in oil prices and their supposed impact on the stock

market, the findings of this paper might be of interest to academicians and practitioners alike. The

conclusions and contributions that I consider most important follow.

While the stock market is sensitive to oil price changes, financial media seems to over play

the supposed impact of oil prices. Oil price changes most likely caused due to supply shifts, i.e., large

price changes over a one-day horizon and price changes during the periods when the US is involved

in an armed conflict in the Middle East have a large negative impact on the same day market returns.

On the other hand, oil price changes most likely caused by changes in expectations of future

economic activity, i.e., small oil price changes over a one-day horizon are positively correlated with

the market returns. I find no evidence of over- or under-reaction of the market to oil price changes,

suggesting that the market is efficient in responding to daily oil price changes. Overall, results

21
indicate that oil price changes that are likely caused by supply shifts have the largest impact on stock

market.

I document the relationship between stock returns of individual industries and oil price

changes, a topic of interest that has not been addressed in the literature. In addition to oil-intensive

industries, returns of some industries which virtually use no oil are also significantly correlated with

oil price changes. I find that the magnitude of the correlation between industries‟ returns and oil price

changes depends both on the cost-side dependence and demand-side dependence on oil. Furthermore,

the effects of these factors vary across industries.

22
References

Barsky, B. Robert, and Lutz Kilian (2001), “Do We Really Know that Oil Caused the Great
Stagflation? A Monetary Alternative”, NBER Working Paper Series 8389

Bernanke, S. Ben, Gertler, Mark and Watson, (1997), “Systematic Monetary Policy and the Effects of
Oil Price Shocks” Brookings Papers on Economic Activity Vol.1, pp91-157

Bittlingmayer, George (2005), “Oil and Stocks; Is it War Risk?” Working Paper Series

Brown, C. Keith, Harlow W.V., and Seha M. Tinic, (1988), “Risk Aversion, Uncertain Information
and Market Efficiency”, “Journal of Financial Economics”, Vol. 22 pp 355-385

Burbidge, John and Alan Harrison, (1984), “Testing for the Effects of Oil-Price Rises Using Vector
Autoregressions”, International Economic Review, Vol. 25, No.2 pp 459-484

Campbell, J. and L. Hentschel, (1992), “No News is Good News: An Asymmetric Model of Changing
Volatility in Stock Returns”, Journal of Financial Economics, Vol. 31, pp 281-318

Campbell, J., Lo, A.W. and Craig McKinlay, (1997), “The Econometrics of Financial Markets”,
Princeton University Press, Princeton, NJ 1997

Chen, N., Roll, R., and Stephen A. Ross, (1986), “Economic Forces and the Stock Market”, The
Journal of Business, Vol. 59, No.3 pp 383-403

Davis, J Steven., and John Haltiwanger (2001). “Sectoral Job Creation and Destruction Responses to
Oil Price Changes.” Journal of Monetary Economics 48, 465–512.

Darby, R. Michael, (1982), “The Price of Oil and World Inflation and Recession”, The American
Economics Review, Vol. 72, No. 4 pp 738-751

Driesprong, G., Jacobsen, B. and Benjiman Maat. (2003), “Striking Oil: Another Puzzle?, Working
Paper, Erasmus University Rotterdam

Fan, Josehp P. H. and Lang, Larry H P., (2000), “The Measurement of Relatedness: An Application
to Corporate Diversification”, The Journal of Business Vol. 73 No 4, pp 629-660

Gisser, Micha, and Goodwin, Thomas H., (1986), “Crude Oil and the Macroeconomy: Tests of Some
Popular Notions: Note”, Journal of Money, Credit and Banking, Vol. 18, No.1 pp 95-103

Golub, Stephen S. (1983), “Oil Prices and Exchange Rates”, The Economic Journal, Vol.93, No. 371,
pp 576-593

Hamilton, James D. (1983), “Oil and the Macroeconomy since World War II”, The Journal of
Political Economy, Vol 91, No.2, pp 228-248

Hamilton, James D., Herrera, M. Ana (2004), “Oil Shocks and Aggregate Macroeconomic Behavior:
The Role of Monetary Policy”, Journal of Money, Credit & Banking Vol 36, No.2, pp 265-
286

23
Hong, H., Torous, W., and Rossen Valkanov. (2002), “Do Industries Lead the Stock Market? Gradual
Diffusion of Information and Cross-Asset Return Predictability”, Working Paper, Stanford
Univeristy & UCLA

Hooker, Mark A., (1996), “What happened to the oil price-macroeconomy relationship?”, Journal of
Monetary Economics, No.38, pp 195-213

Hooker, Mark A., (2002), “Are Oil Shocks Inflationary? Asymmetric and Nonlinear Specifications
versus Changes in Regime”, Journal of Money, Credit and Banking, Vol 34, No.2, pp 540-
561

Huang, R.D.; Masulis, R.W.; Stoll, H.R. (1996), Energy shocks and financial markets, Journal of
Futures Markets, 16, 1-27

Jones, Charles M., and Gautam Kaul, (1996), “Oil and the Stock Markets”, The Journal of Finance,
Vol LI, No.2, pp 463-491

Keane, P. Michael, and Eswar Prasad (1996). “The Employment and Wage Effects of Oil Price
Changes: A Sectoral Analysis.” Review of Economics and Statistics 78, 389–400.

Koenker, R. and Gilbert Basset, (1978), “Regression Quantiles”, Econometrica, Vol. 46, No.1, pp 33-
50

Lee, Kiseok and Shawn Ni (2002). “On The Dynamic Effects of Oil Shocks: A Study Using Industry
Level Data”, Journal of Monetary Economics

Mork, A. Knut, (1989), “Oil and the Macroeconomy When Prices Go Up and Down: An Extension of
Hamilton‟s Results”, The Journal of Political Economy, Vol. 97, No. 3, pp 740-744

Mork, K.A., Olsen, O. and Mysen, H.T. (1994), Macroeconomic responses to oil price increases and
decreases in seven OECD countries. Energy Journal 15 4, 19-35.

Papapetrou, E. (2001), “Oil price shocks, stock market, economic activity and employment in
Greece”, Energy Economics Vol. 23, No. 5 pp 511-532.

Pollet, Joshua. (2002),“Predicting Asset Returns With Expected Oil Price Changes”, Working
Paper, Harvard Univeristy

Sadorsky, P. (1999), “Oil price shocks and stock market activity”, Energy Economics, No. 2, pp449-
469

Shahrur, H (2005), “Industry structure and horizontal takeovers: Analysis of wealth effects on rivals,
suppliers, and corporate customers”, Journal of Financial Economics, 76, pp61-98

24
 Table I
Summary of Data Used in Previous Empirical Studies on the Effect of Oil Price Shocks
Study Sample Period Frequency Key dependent variable/variables Focus
Darby (1982) 1957 – 1976 Quarterly Employment level or real output, money supply, Oil prices and world inflation
govt. expenditures and ratio of exports to GNP

Hamilton (1983) 1948 – 1972 Quarterly Six-variable system presented by Sims (1980 Oil price – macroeconomy
b)* relationship
Golub (1983) 1972 – 1980 Annual Exchange rates Response of FOREX markets
to oil price changes
Burbidge & Harrison (1984) 1973-1982 Monthly Macroeconomic variables from OECD Oil price – marcroeconomy
Countries relationship in OECD
countries
Chen, Roll & Ross (1986) 1953 – 1983 Monthly Returns on NYSE index, growth rates in Asset pricing
industrial production, measures of inflation,
measures of risk premium etc.

Gisser & Goodwin (1986) 1961 – 1982 Quarterly Six-variable system presented by Sims (1980 Oil price – macroeconomy
b)* relationship

Mork (1989) 1949 – 1988 Quarterly Six-variable system presented by Sims (1980 Oil price – macroeconomy
b)* asymmetric relationship
Kaul & Seyhun (1990) 1947 – 1985 Annual Growth rate of output and market returns Effect of relative price
variability on output and
stock market
Hooker (1996) 1948 – 1994 Quarterly GDP and unemployment rate Oil price – macroeconomy
relationship
Huang, Masulis & Stoll (1996) 1983 – 1990 Daily (futures) S&P 500 index returns, industry returns Relation between oil futures
and stock returns

Jones & Koul (1996) 1947 – 1991 Quarterly Returns of market indices of several countries Market‟s ability to evaluate
and Cash flows the impact of oil shocks

Keane & Prasad (1996) 1966 – 1981 Survey data Weekly wage rates and proxies for human Employment and wage
capital effects of oil price changes
Bernanke, Gertler & Watson (1997) 1965 – 1995 Monthly GDP, GDP deflator and federal funds rate Effects of systematic
Study Sample Period Frequency Key dependent variable/variables Focus
monetary policy and oil price
shocks
Sardosky (1999) 1947 – 1996 Monthly Index of Industrial Production, interest rates & Impact of oil prices and
real stock returns volatility on stock returns

Ciner (2001) 1983 – 2000 Daily (futures) Same as Huang, Masulis & Stoll (1996) Non-linear linkages between
energy shocks and financial
markets
Barsky & Kilian (2001) 1960 – 2001 Annual Growth rate of GDP Role of oil price shocks in
causing stagflation
Davis & Haltiwanger (2001) 1972 – 1988 Quarterly Job flows between industries Oil shocks and employment
effects
Lee & Ni (2002) 1959 – 1997 Monthly Industry level output Output responses to oil price
shocks

Hamilton & Herrera (2002) 1965 – 1995 Monthly Same as Bernanke et.al (1997) Oil price – macroeconomy
relationship and the role of
monetary policy
Hong, Torous & Valkanov (2002) 1972 – 2001 Monthly Thirty-four industry portfolios Predictability of market by
industries
Hooker (2002) 1962 – 2000 Quarterly Rate of inflation Oil price changes and
inflation
Pollet (2004) 1973 – 2002 Monthly Value-weighted market and industry returns Predictive ability of expected
oil price changes

Bittlingmayer (2005) 1983 – 2004 Daily Returns on S&P 500 index War risk and impact of oil
price changes

Driesprong, Jacobsen & Maat (2005) 1973 – 2003 Monthly Returns on market indices of several countries Predictive ability of oil price
and world market index changes

*
This system includes two output variables (real GNP and unemployment rate), three price variables (implicit price deflator for nonfarm business
income, hourly compensation per worker, and import prices) and Money supply

26
 Table II
Descriptive Statistics

This table reports descriptive statistics of oil and market returns. Daily oil price data are obtained
from Normans‟ historical data. Market Return is the return on valued-weighted
NYSE/NASDAQ/AMEX index obtained from CRSP. Panel A presents the summary statistics. Panel
B presents the autocorrelation patterns of these variables. For each variable, the first row presents the
estimates while the second row presents the associated t-statistics. Coefficients significant at 10%, 5%
and 1% are marked with ***, ** and * respectively. Daily data from April 1983 to December 2006 are
used.

Panel A: Summary Statistics

Statistic Oil Market
Return return
Observations 5951 5995
Minimum -0.3310 -0.1714
Maximum 0.2197 0.0866
Mean 0.0004 0.0005
Deviation 0.0238 0.0096

Panel B: Autocorrelation Patterns

Variable Intercept Lag1 Lag2 Lag3 Lag4 Lag5 Rsquared
Oil return 0.0005 -0.0120 -0.0542 -0.0324 -0.0046 -0.0454 0.0058
1.5079 -0.9293 -4.1814*** -2.4978*** -0.3512 -3.4995***
Market return 0.0005 0.0656 -0.0368 -0.0199 -0.0015 -0.0049 0.0059
*** ***
4.0619 5.0776 -2.8384 -1.5373 -0.1156 -0.3752
 Table III
Measures of Industries‟ Dependence on Oil
This table presents details about the 81 industry groups used in this study ranked by their cost-side
dependence on oil. Cost-side dependence$ indicates the total oil used by an industry as a percentage
of total output. Demand-side dependence$ is a measure of the importance of oil to an industry‟s main
customers. It is calculated as the weighted average of customer industries‟ cost-side dependence on
oil. These measures are calculated by using data from Benchmark input-output accounts published by
the Bureau of Economic Analysis. Slope$ and t-statistic are the estimated slope coefficient and the
associated t-value on oil returns variable in regression specification (7). Coefficients significant at
10%, 5% and 1% are marked with ***, ** and * respectively. Daily data from April 1983 to December
2006 are used. SIC codes used to calculate industry returns are available on request.

Rit 1 Rot 2 Rst t (7)

Here Rit is the value-weighted return of industry „i‟ on day „t‟. Rot is the daily oil return and Rst is
the value weighted market return on day „t‟ .
$
: these values are multiplied by 102 for improved readability.

Industry Name Cost-side Demand-side slope t-stat

dependence dependence
1 Air transportation 19.60 3.47 -5.270 -9.378***
2 Railroads and related services; passenger ground 12.06 6.55 -0.270 -0.58
transportation
3 Nonmetallic minerals mining 6.33 2.06 1.910 2.65***
4 Motor freight transportation and warehousing 5.97 2.61 -2.730 -5.30***
5 Water and sanitary services 5.50 2.49 -1.500 -3.30***
6 Other agricultural products 5.24 0.86 -0.970 -1.61
7 Electric services (utilities) 5.22 2.05 -0.100 -0.27
8 Industrial and other chemicals 3.27 1.68 -0.460 -1.2
9 Forestry and fishery products 2.77 1.00 0.570 0.37
10 Automotive repair and services 2.15 1.53 -1.550 -2.00**
11 Water transportation 2.10 1.58 0.380 0.53
12 Cleaning and toilet preparations 1.89 0.58 -3.250 -6.46***
13 Paints and allied products 1.69 1.11 -1.630 -3.40***
14 Stone and clay products 1.51 1.50 -0.220 -0.42
15 Agricultural fertilizers and chemicals 1.47 2.20 1.950 3.27***
16 Retail trade 1.44 1.68 -4.350 -10.21***
17 Wholesale trade 1.30 1.56 -1.280 -3.82***
18 Agricultural, forestry, and fishery services 1.27 1.93 -0.520 -0.45
19 Paperboard containers and boxes 1.22 2.26 -1.240 -0.99
20 Other business and professional services, except medical 1.17 3.34 -0.360 -0.67
21 Electrical industrial equipment and apparatus 1.13 1.29 -0.920 -1.72*
22 Livestock and livestock products 0.98 0.38 -0.430 -0.5
23 Pipelines, freight forwarders, and related services 0.93 0.81 2.210 2.88***

28
24 Glass and glass products 0.82 0.36 -0.590 -0.93
25 Personal and repair services (except auto) 0.82 2.10 -0.290 -0.48
26 Educational and social services, and membership 0.82 2.03 -1.590 -2.25**
organizations
27 Primary nonferrous metals manufacturing 0.80 0.30 -0.170 -0.26
28 Primary iron and steel manufacturing 0.79 0.59 5.490 9.47***
29 Paper and allied products, except containers 0.79 0.50 -2.050 -4.49***
30 Lumber and wood products 0.65 0.39 -1.050 -1.6
31 Furniture and fixtures 0.64 0.21 -1.930 -4.36***
32 Miscellaneous manufacturing 0.62 0.46 -2.030 -4.33***
33 Plastics and synthetic materials 0.52 0.48 -2.170 -3.39***
34 Hotels and lodging places 0.50 0.64 -2.050 -3.63***
35 Other fabricated metal products 0.48 0.87 -1.790 -3.23***
36 Metalworking machinery and equipment 0.48 1.98 -0.400 -0.69
37 Miscellaneous textile goods and floor coverings 0.48 1.04 0.810 0.45
38 Footwear, leather, and leather products 0.48 0.66 -0.970 -1.58
39 Electric lighting and wiring equipment 0.47 0.53 -2.000 -2.86***
40 Amusements 0.47 0.50 -0.890 -1.08
41 Miscellaneous machinery, except electrical 0.45 0.78 0.040 0.05
42 Materials handling machinery and equipment 0.44 0.27 0.860 1.31
43 Other transportation equipment 0.42 0.59 -1.100 -2.23**
44 Health services 0.42 0.72 -1.670 -2.80***
45 General industrial machinery and equipment 0.41 0.72 -1.350 -2.57***
46 Eating and drinking places 0.39 3.29 -2.660 -4.97***
47 Broad and narrow fabrics, yarn and thread mills 0.39 0.35 0.630 0.9
48 Heating, plumbing, and fabricated structural metal products 0.38 0.73 -0.530 -0.95
49 Ophthalmic and photographic equipment 0.38 0.76 -1.430 -2.36**
50 Household appliances 0.38 0.74 -3.000 -4.78***
51 Ordnance and accessories 0.37 1.09 -1.160 -1.5
52 Rubber and miscellaneous plastics products 0.36 0.24 -0.700 -1.57
53 Real estate and royalties 0.34 0.42 -0.260 -0.63
54 Newspapers and periodicals 0.33 0.26 -1.010 -2.85***
55 Special industry machinery and equipment 0.32 0.36 -0.420 -0.43
56 Food and kindred products 0.32 1.50 -2.740 -6.83***
57 Miscellaneous fabricated textile products 0.31 1.79 0.810 0.45
58 Other printing and publishing 0.30 0.43 -0.480 -1.32
59 Screw machine products and stampings 0.27 0.34 -0.640 -0.95
60 Farm, construction, and mining machinery 0.27 2.48 2.670 4.69***
61 Tobacco products 0.26 0.91 -2.310 -3.10***
62 Aircraft and parts 0.24 0.72 -2.390 -5.35***
63 Scientific and controlling instruments 0.23 0.95 -1.190 -4.02***
64 Drugs 0.23 0.28 -2.610 -5.66***

29
65 Miscellaneous electrical machinery and supplies 0.23 0.82 -2.720 -4.92***
66 Apparel 0.22 0.58 -2.200 -4.14***
67 Metal containers 0.22 1.34 0.700 0.48
68 Legal, engineering, accounting, and related services 0.22 2.75 1.650 2.53**
69 Communications, except radio and TV 0.21 0.93 -1.280 -3.41***
70 Finance 0.21 0.37 -1.230 -4.67***
71 Truck and bus bodies, trailers, and motor vehicles parts 0.21 1.33 -1.930 -3.92***
72 Advertising 0.20 0.47 -0.180 -0.3
73 Service industry machinery 0.19 0.34 -0.260 -0.39
74 Engines and turbines 0.16 2.05 -0.910 -1.67*
75 Computer and data processing services 0.14 2.22 -0.590 -1.02
76 Electronic components and accessories 0.13 2.66 1.110 1.85*
77 Radio and TV broadcasting 0.11 6.62 -1.590 -3.19***
78 Motor vehicles (passenger cars and trucks) 0.10 1.98 -1.960 -3.40***
79 Audio, video, and communication equipment 0.10 4.86 0.250 0.48
80 Computer and office equipment 0.07 1.49 -0.760 -1.28
81 Insurance 0.07 5.39 -2.110 -6.77***

30
 Table IV
Regression Analysis of the Effect of Oil Price Changes on Market Returns

Panel A presents the estimates of regression specification (1)

Rst Rot t (1)

Panel B presents the estimates of regression specifications (2) & (3)

Rst 1 Rot 2 ( Rot * Rot ) t

(2)

Rst 1 Rot 2 ( D * Rot ) t (3)

Panel C presents the estimates of regression specification (4)

Rst 1 Rot 2 ( war * Rot ) t (4)

Here Rst is the value weighted market return on day „t‟ and Rot is the percentage change in daily oil
price. Rot is the oil returns in absolute terms. D is a dummy variable that equals 1 for the top 25% of
absolute oil returns. war is a dummy variable that equals 1 if US is in a war or 0 otherwise. In each
panel, the first row presents the estimates while the second row presents the associated t-statistics.
Coefficients significant at 10%, 5% and 1% are marked with ***, ** and * respectively. Daily returns
data from April 1983 to December 2006 are used. Estimated coefficients are multiplied by 10 2 for
improved readability.

Panel A: Results of specification (1)

R2
0.0497 -1.866 0.0021
3.99*** -3.56***

Panel B: Results of specifications (2) & (3)

Specification 1 2
R2
Regression (2) 0.0483 0.9238 -40.08 0.0064
3.89*** 1.61 -5.03***
Regression (3) 0.0491 3.963 -6.79 0.0056
3.95*** 2.85*** -4.53***

Panel C: Results of specification (4)

1 2
R2
0.0501 -0.0472 -5.231 0.0059
4.03*** -0.07 -4.77***

31
 Table V
Regression Analysis for Under/Over-reaction and Asymmetry of Market to Oil Price Returns

Panel A contains the estimates of the following regression specification:

Rst 1 Rot 2 Rot 1 3 Rot 2 4 Rot 3 5 Rot 4 6 Rot 5 t (5)

Panel B contains the estimates of the following regression specification:

Rs t 1 Rot 2 ( Rot * D) t (6)

Here Rst is the value weighted market return on day „t‟ and Rot is the percentage change in daily oil
price. D is a dummy variable that takes a value of 1 if the return of oil is positive on day „t‟ and 0
otherwise. For each variable, the first row presents the coefficient estimates while the second row
presents the associated t-statistics. Coefficients significant at 10%, 5% and 1% are marked with ***, **
and * respectively. Daily returns data from April 1983 to December 2006 are used. Estimated
coefficients are multiplied by 102 for improved readability.

Panel A: Results of specification (5)

1 2 4
R2
3 5 6
0.0501 -1.831 0.6525 0.4969 -0.0071 -0.7692 0.0634 0.0030
3.90*** -3.47*** 1.22 0.93 -0.01 -1.43 0.12

Panel B: Results of specification (6)

1 2
R2
0.0447 -2.172 0.6250 0.0022
2.67*** -2.51*** 0.44

32
 Table VI
Determinants of Industries‟ Sensitivity to Oil Price Changes

Panel A presents results of the following cross-sectional regression specification

ˆ cos t side dependencei (8)
i ,1 1

Panel B presents the results of the following cross-sectional regression specification

ˆ cos t side dependencei demand side dependencei (9)
i ,1 1 2

Panel C presents the results of the following quantile regression specification:

ˆ | xi' cos t side dependencei demand side dependencei (10)

i ,1 1 2

for the set of quantiles = {0.25, 0.5, 0.75}.

Here, ˆ is the coefficient estimate on oil return for industry „i‟ obtained from regression
i ,1
specification (7). In each panel, the first row presents the estimates while the second row presents the
associated t-statistics. Coefficients significant at 10%, 5% and 1% are marked with ***, ** and *
respectively. Estimated coefficients are multiplied by 102 for improved readability.

Panel A: Results of specification (8)

1
R2
0.722 -10.994 0.038
-3.80*** -1.76*

Panel B: Results of specification (9)

1 2
R2
-0.803 -10.713 7.331 0.041
-3.20*** -1.70* 0.49

Panel C: Results of specification (10)

Specification 1 2
R2
= 0.25 -1.449 -19.132 -18.567 0.043
*** ***
-5.28 -5.55 -1.65
= 0. 5 -1.125 -1.518 24.986 0.012
***
-4.00 0.22 1.53
= 0.75 -0.609 -0.218 45.083 0.015
-1.50 -0.03 1.82*

33
 Appendix A
This table contains the largest daily oil price changes during the sample period. Panel A presents the fifteen largest oil price decreases while Panel B presents the
fifteen largest oil price increases.

Panel A: 15 largest drops in Oil Prices

Date Oil Return Stock Return News

19910117 -0.402 0.0341 Initial successful strike against Iraq. The drop remained steady. No sudden reversal
19901022 -0.1744 0.0078 Oil price tumbles on rumors of peace in the middle east.
19860722 -0.1739 0.0085 Technical Factors. Rumors that Saudi might increase oil production. Discord among on going OPEC meetings
20010924 -0.1653 0.0345 Fears of recession rise after the terrorist attacks
19910128 -0.163 0.0008 Nothing particular
19980423 -0.1556 -0.0089 Four major oil companies reported lower earnings.
19860408 -0.1382 0.0202 Over supply in the short run.
19860120 -0.1306 -0.0036 Over production by OPEC and mild weather curbed demand.
19900827 -0.1298 0.03 Speculation of a resolution in Middle east and an increase in oil production by OPEC ministers
19860623 -0.1283 -0.0058 Nothing particular. News about an increased onshore drilling activity
19960223 -0.1278 0.0007 Nothing particular
19901130 -0.1227 0.0167 OPEC is pumping oil at the highest levels and unless war disrupts supply, there is an over supply.
20001221 -0.1215 0.0062 Nothing particular
20011115 -0.1213 0.0007 Traders faced the prospect of a price war among global producers
19900808 -0.1179 0.0106 Crude-Oil Prices Fall as Saudis and Others Plan to Boost Output to Offset Shortages.

Panel B: 15 largest hikes in Oil Prices

Date Oil Return Stock Return News

20000124 0.0962 -0.0217 Nothing Particular

20010426 0.0977 0.0094 Nothing Particular
19860224 0.0991 -0.0011 Issues with Saudi Arabia declining responsibility for Oil Price plunge happening for the last six months
19900822 0.1014 -0.0152 Fear of war in the Persian Gulf escalated.
20011226 0.1039 0.0044 Crude Oil Rises on Prospect of OPEC Output Cuts
19910107 0.1053 -0.0162 Renewed war fears.
19910121 0.1058 -0.0012 Nothing Particular
19910114 0.1072 -0.0087 War worries swept world oil markets on Jan 14, 1991 and drove petroleum prices $2 to $4 a barrel higher.
19860407 0.1168 -0.0017 Political: Reagan Aides Dispute Bush and Affirm Free-Market Oil Policy; Prices Up Again
19980622 0.1191 0.0014 Speculation about the three largest exporters to U.S cutting output
19980323 0.1251 -0.0038 The surprise production cutback by the Organization of Petroleum Exporting Countries.
19900806 0.1469 -0.0307 Prices surged around the world as Iraqi troops occupying Kuwait appeared to be digging in on Saudi Arabia's border.
19910122 0.1506 -0.0065 Iraqis Set Fire to Oil Sites In Kuwait
19980427 0.1829 -0.02 Nothing particular
19860805 0.1986 0.004 Oil prices, continuing a sharp turnaround, soared to $15 a barrel in U.S. markets as the OPEC cuts production

35