Interactive Graphic for new Working Paper: 'Fake News'

My coauthor (William Pratt) and I will post our new working paper 'Fake News' here and to SSRN shortly.  It is an analysis of how false information of a takeover of Twitter on July 14 2015 was incorporated into market (stock and option) prices.

To accompany the paper, I created an interactive graphic to help understand the realtionship between Twitter's stock price and options implied volatility.  I am posting the graphic HERE, and linking to it in the journal article.  Take a look, and feedback is always welcome.

Markov Switching Volatilty in Weekly Natural Gas Price Returns (Henry Hub Spot)

In a few earlier posts (here and here) I estimated the Time Varying volatility in daily natural gas futures price (NGH4) returns.  This showed the distinct increase in the volatility of NGH4 in mid January 2014.  Now that the natural gas market has settled I thought I would try to put the recent volatility in perspective.

Pulling weekly Henry Hub spot prices from the EIA's API, I estimated the time varying weekly volatility again using a Markov Switching AR model.  A plot of the state weighted annualized volatility is below. As you can see since 2010 the natural gas market has been fairly tame.  However the recent volatility fits in quite normally with the pre-2010 behavior of gas prices.

markov_weekly_hh_vol_from_97hh_spot_prices

Time-Varying Beta: BP versus Exxon

In a recent presentation I had a plot (see below) of a Kalman filtered estimate of BP's time-varying beta coefficient (from a CAPM type market model).  The plot clearly shows the effect of the 2010 Gulf oil spill on BP's beta --  it jumps up above 2.5.  We also see BP's alpha drop below zero.

bp_beta

Such a large effect on BP's beta coefficient is understandable given what its stock price was doing at the time.

bp_price

This prompted some in the audience to wonder how large the effect was on Exxon's beta coefficient around the time of the 1989 Valdez spill.  Kalman filtered estimates of Exxon's alpha and beta around the time are below.

xom_CAPM

 

As you can see, there is surprisingly little effect on Exxon's beta.  To try and explain why I charted the stock price over the period (below) and found there was little effect on Exxon's stock.  The only decline of note in Exxon's stock was due to the 1987 crash.

exxon_price

This explains why Exxon's beta coefficient was unaffected by the spill.  However, I am not sure why there was such a large effect on BP's stock price, and no measurable effect on Exxon's.  I may take a look into it, and update this post later.

State Dependence in the Natural Gas and Rig Count Relationship

I just posted a new working paper, 'State Dependence in the Natural Gas and Rig Count Relationship' to the USAEE working paper series on SSRN.  The paper is available for download here:  http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2371753

In a nutshell, the paper finds that changes in the north American natural gas rig count does affect future changes in natural gas prices (previous  research has not found this result).  The relationship is state dependent however.  When natural gas prices are above $6.74/MMBtu then increases in the rig count will drive down natural gas prices. Below this threshold the rig count does not affect gas prices (however gas prices affect the rig count). In sum, the evidence is consistent with gas producers, 'killing the rally' in gas prices by markedly increasing gas production.