Forecasting Holiday Retail Sales

Holiday retail sales are a really “spikey” time series, illustrated by the following graph (click to enlarge).

HolidayRetailSales

These are monthly data from FRED and are not seasonally adjusted.

Following the National Retail Federation (NRF) convention, I define holiday retail sales to exclude retail sales by automobile dealers, gasoline stations and restaurants. The graph above includes all months of the year, but we can again follow the NRF convention and define “sales from the Holiday period” as being November and December sales.

Current Forecasts

The National Retail Federation (NRF) issues its forecast for the Holiday sales period in late October.

This year, it seems they were a tad optimistic, opting for

..sales in November and December (excluding autos, gas and restaurant sales) to increase a healthy 4.1 percent to $616.9 billion, higher than 2013’s actual 3.1 percent increase during that same time frame.

As the news release for this forecast observed, this would make the Holiday Season 2014 the first time in many years to see more than 4 percent growth – comparing to the year previous holiday periods.

The NRF is still holding to its bet (See https://nrf.com/news/retail-sales-increase-06-percent-november-line-nrf-holiday-forecast), noting that November 2014 sales come in around 3.2 percent over the total for November in 2013.

This means that December sales have to grow by about 4.8 percent on a month-over-year-previous-month basis to meet the overall, two month 4.1 percent growth.

You don’t get to this number by applying univariate automatic forecasting software. Forecast Pro, for example, suggests overall year-over-year growth this holiday season will be more like 3.3 percent, or a little lower than the 2013 growth of 3.7 percent.

Clearly, the argument for higher growth is the extra cash in consumer pockets from lower gas prices, as well as the strengthening employment outlook.

The 4.1 percent growth, incidentally, is within the 97.5 percent confidence interval for the Forecast Pro forecast, shown in the following chart.

FPHolidaySales

This forecast follows from a Box-Jenkins model with the parameters –

ARIMA(1, 1, 3)*(0, 1, 2)

In other words, Forecast Pro differences the “Holiday Sales” Retail Series and finds moving average and autoregressive terms, as well as seasonality. For a crib on ARIMA modeling and the above notation, a Duke University site is good.

I guess we will see which is right – the NRF or Forecast Pro forecast.

Components of US Retail Sales

The following graphic shows the composition of total US retail sales, and the relative sizes of the main components.

USRETAILPIE 

Retail and food service sales totaled around $5 trillion in 2012. Taking out motor vehicle and parts dealers, gas stations, and food services and drinking places considerably reduces the size of the relevant Holiday retail time series.

Forecasting Issues and Opportunities

I have not yet done the exercise, but it would be interesting to forecast the individual series in the above pie chart, and compare the sum of those forecasts with a forecast of the total.

For example, if some of the component series are best forecast with exponential smoothing, while others are best forecast with Box-Jenkins time series models, aggregation could be interesting.

Of course, in 2007-09, application of univariate methods would have performed poorly. What we cry out for here is a multivariate model, perhaps based on the Kalman filter, which specifies leading indicators. That way, we could get one or two month ahead forecasts without having to forecast the drivers or explanatory variables.

In any case, barring unforeseen catastrophes, this Holiday Season should show comfortable growth for retailers, especially online retail (more on that in a subsequent post.)

Heading picture from New York Times

Big Data and Fracking

Texas’ Barnett Shale, shown below, is the focus of recent Big Data analytics conducted by the Texas Bureau of Economic Geology.

BarnettShale

The results provide, among other things, forecasts of when natural gas production from this field will peak – suggesting at current prices that peak production may already have been reached.

The Barnett Shale study examines production data from all individual wells drilled 1995-2010 in this shale play in the Fort Worth basin – altogether more than 15,000 wells.

Well-by-well analysis leads to segmentation of natural gas and liquid production potential in 10 productivity tiers, which are then used to forecast future production.

Decline curves, such as the following, are developed for each of these productivity tiers. The per-well production decline curves were found to be inversely proportional to the square root of time for the first 8-10 years of well life, followed by exponential decline as what the geologists call “interfracture interference” began to affect production.

TierDCurves

A write-up of the Barnett Shale study by its lead researchers is available to the public in two parts at the following URL’s:

http://www.beg.utexas.edu/info/docs/OGJ_SFSGAS_pt1.pdf

http://www.beg.utexas.edu/info/docs/OGJ_SFSGAS_pt2.pdf

Econometric analysis of well production, based on porosity and a range of other geologic and well parameters is contained in a followup report Panel Analysis of Well Production History in the Barnett Shale conducted under the auspices of Rice University.

Natural Gas Production Forecasts

Among the most amazing conclusions for me are the predictions regarding total natural gas production at various prices, shown below.

Barnetshalecurvelater

This results from a forecast of field development (drilling) which involved a period of backcasting 2011-2012 to calibrate the BEG economic and production forecast models.

Essentially, it this low price regime continues through 2015, there is a high likelihood we will see declining production in the Barnett field as a whole.

Of course, there are other major fields – the Bakken, the Marcellus, the Eagle-Ford, and a host of smaller, newer fields.

But the Barnett Shale study provides good parameters for estimating EUR (estimate ultimate recovery) in these other fields, as well as time profiles of production at various prices.

Forecasting Shale Oil/Gas Decline Rates

Forecasting and data analytics increasingly are recognized as valued partners in nonconventional oil and gas production.

Fracking and US Oil/Gas Production

“Video Friday” here presented a YouTube with Brian Ellis – a Michigan University engineer – discussing hydraulic fracturing and horizontal drilling (“fracking”).

USannualoilprod

Fracking produced the hockey stick at the end of this series.

These new technologies also are responsible for a bonanza of natural gas, so much that it often has nowhere to go – given the limited pipeline infrastructure and LNG processing facilities.

shalegasprod

Rapid Decline Curves for Fracking Oil and Gas

In contrast to conventional wells, hydraulic fracturing and horizontal drilling (“fracking”) produces oil and gas wells with rapid decline curves.

Here’s an illustration from the Penn State Department of Energy and Mineral Engineering site,

Pennstatedeclinecurve

The two legends at the bottom refer to EUR’s– estimated ultimate recovery times (click to enlarge).

Conventional oil fields typically have decline rates on the order of 5 percent per year.

Shale oil and gas wells, on the other hand, may produce 50 percent or more of their total EUR in their first year of operation.

There are physical science fundamentals behind this, explained, for example, in

Decline and depletion rates of oil production: a comprehensive investigation

You can talk, for example, of shale production being characterized by a Transient Flow Period followed by Boundary Dominated Flow (BDF).

And these rapid decline rates have received a lot of recent attention in the media:

Could The ‘Shale Oil Miracle’ Be Just A Pipe Dream?

Wells That Fizzle Are a ‘Potential Show Stopper’ for the Shale Boom

Is the U.S. Shale Boom Going Bust?

Forecasting and Data Analytics

One forecasting problem in this context, therefore, is simply to take histories from wells and forecast their EUR’s.

Increasingly, software solutions are applying automatic fitting methods to well data to derive decline curves and other shale oil and gas field parameters.

Here is an interesting product called Value Navigator.

This whole subject is developing rapidly, and huge changes in the US industry are expected, if oil and gas prices continue below $60 a barrel and $4 MMBtu.

The forecasting problem may shift from well and oil field optimization to evaluation of the wider consequences of recent funding of the shale oil and gas boom. But, again, the analytics are available to do this, to a large extent, and I want to post up some of what I have discovered in this regard.

Video Friday – Fracking

Here is Brian Ellis from Michigan University Engineering with a look at the technology of hydraulic fracturing (fracking) and horizontal drilling – the innovations that recently pushed US oil production near the 10 million barrel per day mark.

I’m putting this up, rather than other, often excellent film clips showing people lighting water from their kitchen taps because the scale of shale oil and gas production has become so large. There really is a huge tradeoff between current employment and business activity and long term environmental effects.

Price, rather than environmental concerns, are likely to be the crucial factor in any scaleback.

At the same time, there is the possibility of further technical advance in the US shale oil and gas technologies, advances which may push extraction prices lower, giving the industry a longer lease during what may be a year or more of lower oil prices.

Fracking and its possible dynamics are critical to a lot of business activity and, thus, forecasting in the US.

China – Trade Colossus or Assembly Site?

There is a fascinating paper – How the iPhone Widens the United States Trade Deficit with the People’s Republic of China. In this Asian Bank Development Institute (ADBI) white paper, Yuqing Ying and his coauthor document the value chain for an Apple iPhone:

IPhone

The source for this breakout, incidentally, is a “teardown” performed by the IT market research company iSupply, still accessible at –https://technology.ihs.com/389273/iphone-3g-s-carries-17896-bom-and-manufacturing-cost-isuppli-teardown-reveals. In other words, iSupply physically took apart an iPhone to identify the manufacturers of the components.

The Paradox

After estimating that, in

2009 iPhones contributed US$1.9 billion to the trade deficit, equivalent to about 0.8% of the total US trade deficit with the PRC,

the authors go on to point out that

..most of the export value and the deficit due to the iPhone are attributed to imported parts and components from third countries and have nothing to do with the PRC. Chinese workers simply put all these parts and components together and contribute only US$6.5 to each iPhone, about 3.6% of the total manufacturing cost (e.g., the shipping price). The traditional way of measuring trade credits all of the US$178.96 to the PRC when an iPhone is shipped to the US, thus exaggerating the export volume as well as the imbalance. Decomposing the value added along the value chain of iPhone manufacturing suggests that, of the US$2.0 billion worth of iPhones exported from the PRC, 96.4% in fact amounts to transfers from Germany (US$326 million), Japan (US$670 million), Korea (US$259 million), the US (US$108 million), and other countries (US$ 542 million). All of these countries are involved in the iPhone production chain.

Yuqing Xing builds on the paradox in his more recent China’s High-Tech Exports: The Myth and Reality published in 2014 in MIT’s Asian Economic Papers.

Prevailing trade statistics are inconsistent with trade based on global supply chains and mistakenly credit entire values of assembled high-tech products to China. China’s real contribution to the reported 82 percent high-tech exports is labor not technology. High-tech products, mainly made of imported parts and components, should be called “Assembled High-tech.” To accurately measure high-tech exports, the value-added approach should be utilized with detailed analysis on the value chains distributions across countries. Furthermore, if assembly is the only source of value-added by Chinese workers, in terms of technological contribution these assembled high-tech exports are indifferent to labor-intensive products, and so they should be excluded from the high-tech classification.

MNEs, in particular Taiwanese IT firms in China, have performed an important role in the rapid expansion of high-tech exports. The trend of production fragmentation and outsourcing activities of MNEs in information and communication technology has benefitted China significantly, because of its huge labor endowment. The small share of indigenous firms in high-tech exports implies that China has yet to become a real competitor of the United States, EU, and Japan. That China is the number one high-tech exporter is thus a myth rather than a reality.

Ying and Yang

This perspective – that it is really “value-added” that we should focus on, rather than the total dollar volume of trade coming in or going out of a country – is interesting, but I can’t help but think there is a disconnect when you consider actual Chinese foreign exchange reserves, shown below (source – http://www.stats.gov.cn/tjsj/ndsj/2013/indexeh.htm).

ChinaFER

So currently China holds nearly 3.5 trillion dollars in foreign exchange reserves – most of which, but not all, is comprised of US dollars.

This is a huge amount of money, on the order of five percent of total global GDP.

How could China have accumulated this merely by being an assembly site for high tech and other products (see Five Facts about Value-Added Exports and Implications for Macroeconomics and Trade Research)? How can this be attributable just to mistakes in counting the origin of the many components in goods coming from China? Don’t those products have to come in and be counted as imports?

There is a mystery here, which it would be good to resolve.

Assembly photo at top from Apple Insider