June 2013 • 35
Are you losing control over
your cash flow?
®
For more information about ADD Energy E3 contact our national sales office at
1-800-922-0972 or visit us on the web at
Gain Control over
Cash Flow again with the
efficiencies built into
cation) traded at a 12.1-cent premium to the
April contract, which still used the heating
oil specification.
For the month of April going back to
2007, the heating oil cashmarket (New York
Harbor Barge Mean) averaged 20 points
below the NYMEX spot month contract,
and the rack differentials averaged between
4.94 and 11.2 cents, based on the dataset.
Applying our 6-cent assumption, we would
then expect the heating oil cash market at
New York Harbor after the specification
change to average 6.02 cents below the
NYMEX ULSD contract, and for rack prices
to average between 1.06 cents below to 5.2
cents above the NYMEX spot month.
In fact, the average spot differentials for
last month showed a more dramatic decline:
New York Harbor heating oil averaged
12.55 cents below the NYMEX, and outside
of Linden, heating oil rack prices ranged
between 5.42 cents under and 0.73 cents over
the NYMEX spot month. Basis began the
month prices particularly elevated at Linden
at over 22 cents, but also declined steadily to
6.33 cents by month’s end. These declines in
heating oil rack differentials are statistically
significant, again excluding Linden, both
when constructing 95 percent confidence
intervals as well as in a simple time series
regression with a binary categorical variable
for the specification change on the right-
hand side.
RUNNING THE NUMBERS
The idea behind confidence intervals is
that if we know the variance of a variable
(in this case the differential at a particular
rack), and if that variable approaches a
normal distribution as the number of obser-
vations increases, we can calculate a range
around the average we observe within which
95 percent of values should fall. Since the
average differentials (and also a 95 percent
confidence interval around them) fall out-
side and below the range where we’d expect
basis to fall within 95 percent of the time,
we have good reason to believe the drop in
averages is not a random event.
To see if this result would hold up
when accounting for other factors likely to
affect rack differentials, I ran a very basic
time series model with a statistical package
for basis at each rack location, using the
previous day’s basis, prior day NYMEX
settlement, heating degree days, the latest
weekly East Coast heating oil inventory
level, and – the variable of interest here – a
“dummy” variable with a value of “1” after
the NYMEX switchover and “0” before. The
specification change variable had a negative
coefficient (as we would expect to see) and
was statistically significant with 95 percent
confidence at five of the rack locations, and
at 90 percent in Linden. In other words,
the results suggest the specification change
contributed to lower rack basis values, and
that these results were highly unlikely to
be random even when controlling for other
factors that may have made a false relation-
ship appear to exist when just looking at
confidence intervals.
A PATTERN EMERGES
Not only do some fixed differential offers
for the upcoming season appear to have been
inflated, be cognizant that seasonal data
indicates that suppliers tend to lag prices as
the market comes off, expanding basis. With
that said, the “portals” do erase some portion
of the expansion. The correlation between
the previous day’s NYMEX price change and
the current day’s rack price change is signifi-
cantly higher for days when the market rose
versus when the market fell over the past 6.5
years. The correlation averaged 83.3 percent
across the rack locations for days when the
market advanced, and just 74.9 percent for
days when the market declined.
Again, checking for robustness of
results, I ran a similar regression to the one
above, but with rack price rather than the
rack basis, and interaction term for “up”
market days instead of the specification
change variable. The interaction basically
allows for the impact of the NYMEX
futures price on the local rack price to be
different for days when the market goes up
(versus down). This term was statistically
significant at the 99 percent level for all six
rack locations.
The analysis produces a very clear con-
sensus. Based on many of the early offers
around wet barrel differentials, there is
risk that monies will be left on the table.
In other words, you could end up paying
substantially more in contract differentials
than what will be available for rack basis
come delivery time. This factor must be
weighed into the hedging equation and
strategy while simultaneously assessing the
basis risk. The months of January through
March are the most vulnerable to basis
fluctuations on a typical heating degree day
curve. Work with your suppliers and your
hedging advisor to protect your margins
during this critical period.
