Volatility in the U.S. natural gas futures market : an analysis of the ‘Shale Gas Boom’ and its effect on the volatility of natural gas futures in the U.S. market

(1)

Bachelor Thesis

Bachelor Finance and Organization

Volatility in the U.S. natural gas futures

market:

an analysis of the ‘Shale Gas Boom’ and its

effect

on

the volatility of natural gas futures in the U.S.

market

Abstract: This thesis

analyzes the effect of shale gas production on price volatility of natural gas futures. Four different contracts in the United States natural gas market are examined in the period 1999-2016. Statistical analysis is conducted and we find no evidence of shale gas production affecting natural gas futures price volatility.

Name: Robbert-Jan Blommestein Student number: 10678425 Date: 29-6-2016 Thesis supervisor: G. Vala Elias Pimentel de Oliveira

(2)

This document is written by Student Robbert-Jan Blommestein who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1.0 Introduction

U.S. natural gas is the biggest innovation in the energy sector in a decade according to Daniel Yergin (2009). Representing just 1% of natural gas supply in 2000, and expected to represent 69% in 2045, Yergin his statement may just be true. If he is right, then more research in this field will be beneficial, especially the impact of shale gas has not been tested empirically. Another reason that justifies more research into shale gas is that the U.S. does not even have the biggest shale gas reserves in the world: China’s reserves are 50% greater (EIA, 2015). This thesis analyzes the effect of shale gas production on futures price volatility in the United States. It aims to add to the discussion in the field of volatility analysis, a field in which oil gets the most attention by researchers. The natural gas market is one of the most volatile markets in existence. Storage levels, weather, heating demand and supply constraints and inelasticity in supply are just some of the variables affecting the price volatility. All these subjects will be examined in this paper, and their roles with regards to volatility explained through their relation with natural gas. In the period after 2007 there was a reduction in overall volatility of natural gas futures traded at the NYMEX, the New York Mercantile Exchange, as can be seen from table 1. The delivery for natural gas in contract 1 takes place one month after the trade date, this is two months for contract 2.

Table 1. Source: author’s own analysis.

This reduction in volatility is the main interest of this thesis and leads to our research question:

“Does shale gas production influence the volatility in the U.S. natural gas market?”

We answer this question by conducting literature review in which recent academic findings concerning this topic will be reviewed. A statistical analysis is then conducted, with the price volatility of futures contracts as dependent variable, and shale gas production in the list of other variables. These other Overall price volatility 1999-2006 2007-2016 % difference Contract 1 248% 214% 14% Contract 2 254% 216% 15% Contract 3 259% 218% 16% Contract 4 255% 222% 13%

(5)

variables are verified with academic literature. We do not find a significant relationship between futures price volatility and shale gas production. We do find that there is a lag-effect: volatility from one period influences volatility in another period.

The next section will give an introduction into the U.S. natural gas market, analyzing the shale gas boom, the composition of the market and its participants. It will also give a short overview of the drawbacks of mining natural gas in the U.S. Section 3 will give an overview of the literature on natural gas markets, futures trading on these markets and the variables influencing future prices volatility.

Section four discusses the data used in the statistical analysis. Section five explains the steps that were taken during the analysis. The results of the statistical analysis are reviewed in section six. Section seven then concludes the paper, after which a suggestion is made for future research.

2.0 Introduction into North American natural gas market

This section is aimed at giving the reader a better idea of the U.S. natural gas market. The sector as a whole is explained, as is the emergence of shale gas. Drawbacks to natural gas developments exist as well and are discussed in this section.

2.1 Changes in the market

The

U.S natural gas market was highly regulated and prices were controlled the government. These government agencies failed and led to a gas shortage. In 1978 the Natural Gas Policy Act was adopted in order to liberalize the market and stimulate competition. Price regulations were completely abandoned in 1989, after which the natural gas market became almost completely market-driven (Wang and Krupnick, 2013). Another change was the start of large scale mining of so called ‘shale gas’. Shale gas is natural gas which is found in low-permeable rock formations spread across North America. Shale gas has acquired a more prominent role in the U.S natural gas supply ever since 2000 when it represented just 1% of total natural gas supply. By 2010 the market share of shale gas had risen to over 20%, to 50% in 2016 and the U.S. Energy Information

(6)

Administration expects shale gas to make up 69% of total natural gas supply in the U.S. by 2040. (EIA, 2016).

In 2015 estimations were made of shale gas reserves per country, the results showed that China, Argentina and Algeria each have bigger shale gas reserves than the United States, and are just becoming profitable to mine locally. Learning from the mistakes the United States made can prove very valuable to these countries in the future (Wijermars, 2013,).

2.2 Emergence of shale gas

Shale gas has seen large growth in production since the mid 2000’s. This subsection analyzes what the drivers behind this growth are and what can be

expected in the future.

Shale gas has been mined for a long time. There are accounts of shale gas being extracted as early as 1821 in the state of New York, although production was negligible. The currently used technique for stimulating gas production from wells, hydraulic fracturing, was first used in 1940 (Wang, Chen, Awadhesh and Rogers, 2014).

U.S. oil production peaked in 1970, up to that point there was no incentive seek alternative energy sources, and mining new natural gas was not profitable with the prices at the time. As a consequence of this, attention shifted abroad. When the oil producing countries in the Middle East imposed an oil-embargo in 1973, the U.S. government implemented a policy which strived for energy independence (Yergin, 2006). R&D programs were set up to tackle operational hurdles in mining shale gas on a large scale, and in 1995 the first profitable well was drilled. Wang et al. (2014) identify three main drivers of the increase in shale gas production since 2007.

High oil prices are the first factor. Oil prices in 2003 were $30 per barrel, and rose to $140 per barrel in 2008. The rising oil prices caused a grow in demand for natural gas because consumers started substituting oil for natural gas. Technological development is the second factor. While oil prices and demand for natural gas were rising, technological development enabled mining companies to reduce costs, thus improving profitability. This mechanism continued into 2008 when the oil price reached a peak of $140 per barrel. The increased profitability of drilling companies attracted new players into the market and supply started to rise. This rise in supply forced prices down and is

(7)

the reason for the third factor. Since natural gas prices were low, coal-fired electricity plants started converting to accept natural gas as input for their process.

2.3 Futures markets participants

The most active traders in energy products and derivatives are producers, distributors, wholesale consumers and, as a result of significant deregulation in 1989, large and complex financial institutions (Albrecht, 2009). All of these market participants have to deal with price volatility. For all these market players, it is important to examine the dynamics of energy prices and volatility in order to make correct decisions relating to risk management, hedging and investments. Hedging became easier in April 1990 when The New York Mercantile Exchange (NYMEX) started trading natural gas futures (Nymex, 2016). As of June 2016 there are currently 54 contracts being offered, each one specifying delivery for one calendar month in the future. Volumes are highest for the contracts nearest to the future (CME Group, 2016).

2.4 Drawbacks of shale production

Shale gas sometimes sounds like the solution for energy production in the U.S for some decades to come. There are many drawbacks however. Recent low prices of natural gas have led to the downfall of coal as the main source of choice for generating electricity (Louie & Pearce,2016). Burning natural gas emits 55% less CO2 when burned then coal (Kerr, 2010), but scientists are expecting this ‘shale boom’ to actually enhance CO2 emissions rather than reducing them. First explanation is that U.S. coal exports are going up (Louie & Pearce, 2016). Thus leading to CO2 emissions elsewhere in the world. Secondly, the relative cost of alternative energy goes up. Investments in clean technology are not attractive when natural gas is cheap. Researchers are worried that low gas prices will slow down the transition and distract politicians from transforming a fossil fuel driven economy to one powered by renewable energy (Knittel, Metaxoglou, Trindade, 2016). Recently there has been a lot of research has been done into the subject of contamination of groundwater, and the endangerment of drinking water by shale gas production (Vengosh et. al., 2014)

(8)

2.5 Volatility in natural gas market

This section will give an overview of the academic literature on this subject. The relevance of volatility analysis, the effects of price volatility and the causes are examined in this section.

Prices of commodities are volatile and this volatility varies over time, is one of the findings of Geman and Ohana (2009). Their research on U.S natural gas volatility and forward curves indicates that the price of commodities therefore is difficult to predict. This is remarkable because in 2010, Alquist and Kilian (2010) found evidence that several large institutions, both governmental and NGO’s like the International Monetary Fund, use the prices of futures to make predictions about the spot prices. Alquist and Kilian (2010) found that both parties feel that futures prices are a more accurate predictor of spot prices than the econometric models they are able to generate themselves. They however come to the conclusion that futures prices do not predict the future spot prices in the oil market. The same methodology was applied by Mishra and Smyth (2016) on the natural gas market, they reached the same conclusion.

2.6 Effects of price volatility

In his analysis of commodity price volatility focusing on oil and natural gas, Pindyck (2004) concludes that changes in volatility alter the exposure of market participants to price risk. The price risk, he continues, directly affects the market participants’ willingness to invest in the natural gas market. Pindyck (2004) conlcudes that these investments should be split into two components: investments in derivatives and real options.

In his empirical research on natural gas price volatility Herbert (1995) finds that time to maturity had an effect on the value of futures. Futures contracts that are closer to maturity, respond more strongly to new information on expected market developments such as storage levels and supply and demand conditions. Contracts which are far from maturity react less strongly to the same news. This is in line with the findings of Pindyck (2004), who concludes that fluctuations in volatility do not persist for a long period of time. He finds that the half-life of volatility shocks is 5 to 10 weeks. Volatility affects investments in futures contracts nearer to maturity to a greater extent than contracts which are further from maturity (Pindyck, 2004).

(9)

2.6.1. Effect on real options

The choice whether or not to invest in drilling a new gas well is an example of a real option. Pindyck (2004) concludes that volatility should not have an effect on the valuation of these real options. In their research into stock markets, real options and volatility, Grullon, Lyandres and Zhadanov (2012) use the real options theory. This theory states that the value of the real option is increasing in the volatility of the process that lies under the asset. The underlying process is defined as demand volatility, cost volatility or profitability volatility. According to the real options theory volatility should enable the drilling firm to improve profits since firms can react to volatility in a way which yields them the highest payoff. Examples of actions in the natural gas market could be to postpone investments in new wells, invest in new wells or temporarily discontinue drilling as a response to volatility. In practice reacting optimally to volatility is not feasible, projects in the natural gas sector are by nature not suitable for these responses. Pindyck (2004) concludes that shocks to volatility last on average five to ten weeks, which is too short for companies within the natural gas sector to adequately respond.

2.6.2 Causes of price volatility

As mentioned before, the current U.S natural gas market is a liberalized market. The interaction between demand and supply thus determines the price. External factors as storage levels, weather, trading volumes and the maturity effect also have an effect on the determination of the price and thus volatility according to literature. All are discussed in this section. Financial speculation is also mentioned in literature as a possible influence of price, but falls beyond the scope of this paper.

2.6.3 Shocks in supply and demand

In the natural gas market, demand is often not equal to supply. Storage can help alleviate this problem and is discussed in the next section. Reasons for this inequality between demand and supply are discussed in this section. Natural gas demand is split in categories by end-user: residential, commercial, industrial and electric power. The main uses for natural gas are the generating of electricity and the heating or cooling of spaces (EIA, 2016). Demand is largest in the months bordering January. Mastrangelo (2007) finds evidence for this seasonal pattern in his analysis of spot price volatility in natural gas markets. These

(10)

months, November through March, are generally known as the heating season (EIA, 2016). In this period demand for natural gas is high because outside temperatures such that heating of residential and commercial spaces is necessary for comfort or production processes. In the cooling season (April through October) electricity demand rises as a consequence of cooling demand which takes place mostly through the use of air conditioning.

2.6.4 Storage

Kaldor (1939) was the first to document the Theory of Storage. The theory, not specific to natural gas, states that consumers and processors of a commodity receive a stream of profits from owning inventory of the commodity. These profits are called convenience, they represent the ability to profit from unexpected shocks in supply or demand as discussed in the previous section. The theory predicts that the benefits of holding inventory decline as inventory levels rise. Evidence for this theory has since been provided by multiple researchers (Working (1948), Fama and French (1987)) In their research on natural gas volatility, storage and convenience, Susmel and Thompson (1997) state that storage enables inventory holders to temporarily increase the supply when demand increases. When inventories are high, Susmel and Thompson state that a large change in inventory should correspond to a low change in convenience yield. This corresponds to a small change in the spread between spot and futures prices: the spot and futures prices change by roughly the same amount. When inventories are low, a change would trigger a larger response in the spot price then it would in the futures price (Susmel and Thompson, 1997).

Futures prices tend to move towards spot prices as maturity comes closer, different changes in prices between spot and futures thus entails larger volatility. Mastrangelo (2007) tested this in his analysis on volatility in natural gas spot markets. He found that when storage levels moved away from the five year average storage level, volatility increased, although there was a lag. He suggests the lag could mean that storage at time Tt+x could be affected by volatility in Tt.

This conclusion is in line with the research of Susmel and Thompson (1997). They state that an increase in volatility increases the marginal profits of holding that inventory, thus raising demand for storage in a response to higher volatility. In more recent research on price volatility of U.S natural gas markets by Geman and Ohana (2009), a negative correlation between inventory and price volatility is found. They add to the research of Mastrangelo (2007) by concluding that this

(11)

negative correlation only holds when storage levels are below their historical average which is always during the heating season.

2.6.5 Trading volume

Serletis (2007, p. 37) mentions that although he finds evidence of trading volume affecting price volatility, there seem to be other factors influencing volatility at the same time as trading volume. It is worth noting that unlike Serletis, we actually find that volatility has dropped since 2006, while the trading volume has increased. Our findings are given body by Wen-I Chuang, Hsiang and Rauli (2012, p. 2,) and Chocholatá (2011) who have done similar research as Serletis, albeit into stock markets. They conclude that the trading volume has different effects on different markets. They do however both conclude that this influence is more often found in developed markets than in markets that are developing. As we can see from figure 3, the volume of futures that are traded has risen strongly. strongly. It may be the case that the situation described by both papers applies to the natural gas market as well, and that the natural gas market has reached a more mature time hence the effect of trading volume.

3.0 Data

The data for Henry Hub futures prices are retrieved from EIA.gov. The Henry the most important hub in North America. It connects four intrastate and nine interstate pipelines and the major pipelines from wells in the Gulf make land here. Henry Hub natural gas prices are seen as the general prices for natural gas in the North American market. Financial trading is made up of transactions in futures, swaps, options and other types of derivatives based on natural gas prices (Albrecht, 2009). Around 360.000 Henry Hub natural gas futures are traded daily at the New York Mercantile Exchange, with a market value of around 10.8 billion dollars (CME Group, 2016).

Prices are retrieved for contract 1 through 4. Contract 1 matures in the nearest month, contract 2 the month after and so on. Futures data were retrieved with daily intervals from the U.S. Energy Information Administration, which collects its data directly from the New York Mercantile Exchange. Shale production data was unfortunately only available with monthly intervals, and was also retrieved from the U.S Energy Information Administration. Natural gas storage levels and shale production data were retrieved from the U.S. Energy

(12)

Information Administration.

Cooling degree days and Heating Degree days were retrieved from Weather data depot. Weather plays an important role in natural gas demand. In order to make the estimation as accurate as possible, weather data was obtained for three cities: Kansas, Dallas and Pittsburgh. These cities are located in a semi-circle around Erath, Louisiana: the location of the Henry natural gas Hub. This was the most accurate estimate I could get since area-wide weather data was

not available.

Formulating a hypothesis is not really doable since no research has yet been done on this subject. It would be pure speculation.

Complete regression model:

LN(volatility) = β0 + β1 ln(Shaleproduction)+ β2

ln

PfuturesT

PfuturesT −1

+ β3

ln

PfuturesT −1

PfuturesT −2

+ β4 AboveStorage + β5 ln(StorageDifference) + β6

Heatingdays + β7Coolingdays β8 Feb + …. β 17 Dec +

ε

Explanations of variables can be found in the appendix.

4.0 Methodology

This section explains transformations made on the data, why we choose for using the ARIMA and Prais Winsten models, and the intermediate conclusions.

First the daily price change is calculated: Daily logarithmic price change-> ∆Pt =

ln

PfuturesT

PfuturesT −1

To calculate monthly volatility, the standard deviation of these price changes is taken an multiplied by the square root of the number of trading days in month i.

(13)

Formula: Volatilityi :

∑

i=1 N t

(

¿

∆ P t−Δ ´p)

N t−1

√

¿

*

√

N t−1

In order to be able to assess the estimated coefficients as elasticities, a natural logarithm transformation on the variables was carried out. The amount of days per month is the number of observed trading days.

We regressed the data using an OLS model. The results did not seem right, so an Durbin Watson test was conducted (table 2). There was positive autocorrelation in the data, which needed correcting. We corrected for this using a Prais-Winsten model, and an ARIMA model to check if all the correct steps were taken, since both should yield approximately the same results. Results of the regressions can be seen from table 5 through 8. We ran four regressions per contract. In the first regression all monthly dummies are excluded. In the second, third and fourth regression model they are included. Results of regression one and two have no meaning however, because there was significant autocorrelation.

5.0 Results

In this section we will analyze the results obtained from our regression analysis described in the methodology section. Results of this regression can be seen in

tables 5 through 8 in the appendix.

The four different models are run per contract.

From regression models one and two, nothing can be concluded. Although we see that the Price variables are mostly significant, the autocorrelation in the model makes it too unreliable to infer any results from.

In model three the complete regression model is run using the Prais-Winsten model. We see that in most of the models Price and Price (lagged) are significant, which is in line with literature. Mastrangelo (2007) finds the same results for the Price variable. Shale gas production is significant for contract number 2. For all the other contracts shale production is not significant, so no conclusion about a general effect of shale gas production can be based of this single case of significance.

(14)

Model four then uses the ARIMA model, which yields approximately the same outcomes as the Prais-Winsten, which is a good outcome since it means the Prais-Winsten model was applied correctly.

6.0 Conclusion

In this thesis I investigate the effect of shale gas mining on the price volatility of natural gas futures. We first give a short introduction into the U.S. natural gas market and discuss the composition of the market, its history and the future expectations. We then turned to literature to review the main findings of academic papers written on the subject of price volatility. We find that the vast majority of papers on price volatility are written about the oil markets. However, building on the work of Smyth and Mishra (2016), Pindyck (2004) and Mastrangelo (2007) allowed us to get a solid foundation for the empirical part of this thesis.

After reviewing the literature, data collected from the U.S. Energy Information Administration (2016) was analyzed. We used four futures contracts and their volatilities in our analysis, the first contract maturing in the nearest month and every following contract a month thereafter. We started off with regular OLS analysis, which led us to conclude that there is strong autocorrelation in the residuals of our OLS model. This was confirmed when we performed the DurbinWatson test (table 2 and 3). To compensate for this we used the Prais-Winsten model, and the ARIMA model to check if the correct steps had been taken, since both should yield approximately the same results, which was the case.

From our results we can infer that the Price variables, which are the averages of historical prices have a significant effect on price volatility, which is in line with literature. We can however not conclude that shale gas production

(15)

7.0 Bibliography

Albrecht, W. (2009). Price Transparency in the US Natural Gas Market. Natural

Gas Supply Association.

Alquist, Ron, and Kilian Lutz, 2010, What do we learn from the price of crude oil futures? Journal of Applied Econometrics Vol 25: 539–573

Alterman, S (2010). Natural gas price volatility in the UK and North America. The oxford institute for energy studies, February 2010.

Chocholatá, M., 2011. Trading volume and volatility of stock returns: evidence from some European and Asian stock markets. Quantitative methods in economics. Vol. 12, No. 1, p.35.

Chuan, W.I., Liu, H.H., Susmel, R., 2012.The bivariate GARCH approach to investigating the relation between stock returns, trading volume and return volatility.Global Finance Journal, Vol. 23, issue 1, 1-15.

CME Group, 2016. Natural Gas (Henry Hub) Last-day Financial Futures Quotes. CME Group, 2016. Henry Hub Natural Gas Futures Contract Specs. Dahl C.A., 1993. A survey of energy demand elasticities in support of development of the NEMS. U.S. Department of Energy: p.6, 107

Energy Information Administration. Annual Energy Outlook 2016 Early Release: Annotated Summaryof Two Cases.

Herbert, J.H (1995). Trading volume, maturity and natural gas price volatility. Energy economics, Vol. 17 No.4, 1995, 293-299.

Grullon, G, Lyandres, E and Zhadanov, A., 2012, Real Options, Volatility, and Stock Returns. The Journal of Finance. Vol.67, 1499-1537

Geman, H., Ohana. S, 2009. Forward curves, scarcity and price volatility in oil and natural gas markets. Energy Economics, Vol. 31, 576-585

Knittel, Metaxoglou, Trindade, 2016. Are we fracked? The impact of falling gas prices and the implications for coal-to-gas switching and carbon emissions. Oxford Review Economic Policy, Vol. 32 No.2, 241-259.

Kawai, M. 1983. Price volatility of storable commodities under rational expectations in spot and future markets. International economic review, Vol. 24, No 2.

(16)

Kerr, R.A. (2010) Natural gas from shale bursts onto the scene. Science, Vol. 328, Issue 5986, p.p. 1624-1626

Louie, E.P., Pearce, J.N., 2016 Retraining Investment for U.S. Transition from Coal to Solar Photovoltaic Employment. Energy Economics,2016. Available June 11 June 2016.

Mastrangelo, E., 2007. An analysis of price volatility in natural gas markets. Energy Information Administration, Office of Oil and Gas.

Mishra, V., Smyth,R. 2016. Are natural gas spot and futures prices predictable? Economic Modelling, Vol. 64, 178-186

Osborn, S. G., Vengosh, A., Warner, N. R. & Jackson, R. B., 2011. Proc. National Academy of Sciences USA, Vol 108, p.p. 8172–8176.

Pindyck, R. 2004. Volatility in natural gas and oil markets. The Journal of Energy and Development, Vol. 30, No. 1

Qiang, W., Xi, C., Awadhesh, N., Rogers, H. 2014. Natural gas from shale formation – The evolution, evidences and challenges of shale gas revolution in United States. Renewable and Sustainable Energy Reviews, Volume 30, Pages 11. Rangarajan K.,Sundaram Sanjiv, R., 2016. Derivatives: principles and practice, second edition. McGrawhill Education, 2016, 2 Penn Plaza, New York, NY 10121. Serletis, A., 2007. Quantitative and Empirical Analysis of Energy Markets. World Scientific Series on Energy and Resource Economics. World Scientific Publishing Co. Pte. Ltd, Singapore.

Susmel, Raul, Thompson, Andrew, 1997. Volatility, Storage and Convenience: Evidence from Natural Gas Markets. The Journal of Futures Markets, Vol. 17 No. 1, 17-43

Vengosh, Jackson, Warner, Darrah, Kondash. (2014) A Critical Review of the Risks to Water Resources from Unconventional Shale Gas Development and Hydraulic Fracturing in the United States. Environ. Sci. Technol., Vol. 48 No. 15, pp 8334– 8348.

United States Federal Trade Commission (1938). Title 15 – Commerce and Trade – Chapter 15B – Natural Gas.

U.S. Energy Information Administration, 2015. World Shale Resource Assessments. WEB

Walls, W.D., 1995. An Econometric Analysis of the Market for Natural Gas Futures. The Energy Journal, Vol. 16, No. 1, p.p. 71-83

(17)

Wang, Q., Chen, X., Awadhesh, N., Rogers, H. (2014). Natural gas from shale formation – The evolution, evidences and challenges of shale gas revolution in United States. Renewable and Sustainable Energy Reviews Vol. 30, 1-28.

WeatherDataDepot, 2016. Heating degree days Kansas, Texas.

Wijermars, R., 2013. Economic appraisal of shale gas plays in Continental Europe. Applied Energy, 2013, Vol 106, p. 101.

Yergin, D. (2006). Ensuring Energy Security. Foreign Affairs. Vol. 85, 69-82

Zhongmin, W., Krupnick, A. (2013). A Retrospective Review of Shale Gas Development in the United States. Resources for the future. April 2013

(18)

8.0 Appendix

Variable

DurbinWatson statistic

Model 1

1.1283

Model 2

1.2231

Model 3

1.2585

Model 4

1.3049

Table 2: Durbin Watson test on models

Variable

DurbinWatson statistic

Model 1

2.087 Model 2

2.06 Model 3

2.1151

Model 4

2.1235

Table 3: Durbin Watson test on Prais- Winsten corrected models

(19)

This table shows the descriptive statistics for our sample and individual factors that make up the regression model. The natural log values of Volatility, Shale gas production, Price, Price (lagged), Trading Volume, and absolute storage difference are taken. All variables run from January 1999 through March 2016. t-statistics are reported below the coefficients in parentheses. Coefficients marked with ***, **, and * are significant at the 1%, 5%, and 10% level, respectively.

Panel A: Summary statistics Contract 1

Variable Observations Mean

Standa rd Deviati

on Min Max

Volatility 207 -0.972 0.642 -2.359 1.012

Shale Gas Production 207 0.005 0.011 -0.042 0.058

Futures Price 207 1.090 0.507 0.118 2.396

Futures Price (lagged) 207 1.090 0.508 0.097 2.415

Absolute storage

difference 207 -1.57 1.013 -5.84 0.066

Volume traded 207 0.011 0.491 -1.67 1.31

Heating days 207 246.03 314 0 1048

Cooling days 207 194.43 227.22 0 751.5

Panel B: Summary statistics contract 2

Variable Observati ons Mean Standa rd Deviati on Min Max Volatility 207 -1.065 0.641 -2.42 0.677

Futures Price 207 1.110 0.512 0.118 2.42

Futures Price (lagged) 207 0.347 0.178 0.069 0.696

Absolute storage

difference 207 -1.57 1.013 -5.84 0.066

Volume traded 207 0.011 0.909 -7.570 0.696

Heating days 207 246.03 314 0 1048

(20)

Table 4 continued. Panel C: Summary statistics contract 3

Stand ard Deviat

ion Min Max

Volatility 207 -1.091 0.641 -2.427

0.67 7

Shale Gas Production 207 0.005 0.011 -0.042

0.05 8 Futures Price 207 1.124 0.516 0.118 2.45 4 Futures Price (lagged) 207 0.362 0.187 0.094 0.73 7 Absolute storage difference 207 -1.57 1.013 -5.84 0.06 6 Volume traded 207 0.013 0.623 -2.053 1.78 3 Heating days 207 246.03 314 0 1048 Cooling days 207 194.43 227.22 0 751. 5

Panel D: summary statistics contract 4

Stand ard Deviat

ion Min Max

Volatility 207 -1.169 0.628 -2.723

-0.41 8

Futures Price 207 1.132 0.517 0.118 2.438 Futures Price (lagged) 207 0.381 0.190 0.111 0.754 Absolute storage difference 207 -1.57 1.013 -5.84 0.066 Volume traded 207 0.014 0.788 -2.369 1.949 Heating days 207 246.03 314 0 1048 Cooling days 207 194.43 227.22 0 751.5

(21)

Table 5: Impact of shale gas on price volatility of contract 1

The table shows the regression results for Futures Contract 1. Regressions are as defined in the appendix. The natural log values of Price, Price (lagged), Trading Volume, Above average storage and storage difference are taken. Monthly dummies are not included in this table but were added to models 2,3 and 4. The sample period runs from January 1999 through March 2015. Coefficients marked with ***, **, and * are significant at the 1%, 5%, and 10% level, respectively. t- scores are included in parentheses for model 1 through 3, model 4 utilizes an ARIMA model, so between parentheses are the z-scores.

Variable (1) (2) (3) (4) Shale production -0.8.02*** -6.794** -1.70 -1.268 (-3.09) (-2.50) (0.347) (-0.43) Price futures 0.714*** 0.683** * 0.720*** 0.738** * (4.30) (3.74) (4.60) (4.43)

Price futures (lagged) 0.239 0.255 0.261* 0.249

1.44 (1.44) (2.01) (1.40) Traded volume 0.045 0.022 0.042 -0.005 (0.75) (0.32) (0.97) (-0.11) Above average storage 0.19 0.043 0.012 0.142 (2.38) (0.47) (1.48) (1.27)

Absolute storage diff. 0.009 0.001 0.012 0.016

(0.32) (0.18) (0.43) (0.38)

Heating days 0.000 0.00 0.00 0.000

(1.17) (0.18) (0.55 (0.48)

Cooling days -0.000 -0.001* -0.001 -0.001

(22)

Variable (1) (2) (3) (4) Shale production -1.534 -0.297 1.07*** -.31 (-0.60) (-0.11) (0.63) (-0.09) Price futures 0.920 0.926** * 0.924*** 0.927*** (17.66) (17.17) (13.19) (15.01)

Price futures (lagged) -0.803

-0.739** * -0.768 -0.738*** (-4.86) (-4.41) (-3.12) (-4.20) Traded volume 0.009 0.012 0.074 (0.34) (0.42 (1.75) Above average storage 0.179 0.088 0.119 0.089 (2.96) (1.04) (1.38) (1.06)

Absolute storage diff. -0.024 -0.024 -0.013 -0.023

(-0.87) (-0.83) (-0.43) (-0.84)

Heating days 0.000 0.000 -0.000 0.000

(2.70) (0.69) (-0.15) (0.56)

Cooling days 0.000 -0.001 -0.001 -0.001

(23)

Variable (1) (2) (3) (4) Shale production -1.44 -0.989 0.451 1.49 (-0.62) (-0.40) (0.22) (0.49) Price 0.937*** 0.933** * 0.932*** 0.937*** (19.54) (18.43) (13.05) (10.78) Price (lagged) -0.757*** -0.692** * -0.701 -0.732*** (-5.19) (-4.56) (-3.34) (-2.96) Traded volume -0.03 -0.007 0.018 0.004 (-0.79) (-0.16) (0.50) (0.1) Above average storage 0.144 0.058 0.053 0.097 (-0.62) (0.73) (0.64) (1.13)

Absolute storage diff. -0.016 -0.014 -0.01 -0.01

(-0.65) (-0.53) (-0.23) (-0.32)

Heating days 0.000 0.000 0.000 0.000

(3.14) (0.88) (0.62) (0.35)

Cooling days 0.000 -0.000 -0.000 -0.000

(24)

Variable (1) (2) (3) (4) Shale production -2.26 -2.647 -1.28 -1.11 (-0.94) (-1.07) (-0.77) -0.40 Price futures 0.929*** 0.917** * 0.913*** 0.912*** (19.02) (18.20) (12.63) (11.26)

Price futures (lagged) -0.665 -0.622 -0.619*** -0.637***

(-4.54) (-4.12) (-3.01) (-2.32) Traded volume 0.002 -0.022 -0.011 -0.0.23 (0.09) (-0.53) (-0.37) (0.56) Above average storage 0.100 0.001 0.031 0.055 (1.75) (0.537) (0.38) (0.57)

Absolute storage diff. -0.006 0.001 0.010 0.012

(-0.24) (0.05) (0.39) (0.36)

Heating days 0.000 0.000 0.000 0.000

(1.20) (1.14) (0.76) (1.06)

Cooling days 0.000 -0.000 -0.000 -0.000

(25)

(26)

Explanation of variables

- AboveStorage:a dummy variable which has value 1 if the storage at time t is greater than the 5 year average.

- Storagedifference: the absolute value of the percentage difference between the storage level at time t and the 5 year average storage level - Heating days: The amount of days where heating is need within a time

period, divided by the amount of days needed in a similar period the year before.

- Cooling days: the amount of days where cooling is required in a month divided by the amount needed a year before

- The variables β7 through β17 are dummy variables representing the months

in a year. These are added to account for a seasonality effect.

- Shale: a dummy variable which equals 0 until December 2006 and 1 from January 2007 onwards

- Shaleproduction: the amount of shale gas produced in the lower 48 states of the United States measured in millions of cubic feet.

Volatility in the U.S. natural gas futures market : an analysis of the ‘Shale Gas Boom’ and its effect on the volatility of natural gas futures in the U.S. market

Bachelor Thesis

Bachelor Finance and Organization