USA retail sales and trend estimates for August 2015

It has been a long time since we’ve looked at the USA Retail Sales estimates. Way back in 2012: http://www.seasonaladjustment.com/2012/09/14/usa-retail-sales-for-august-2012-and-the-trend/ so it is worth a revisit.

The Census Bureau do not estimate or publish official trend estimates, but trend estimates can be derived by taking the published seasonally adjusted estimates and applying a set of Henderson filters (with a bit of code in R and ggplot2). Using the latest published data up to and including August 2015 gives

USA Retail Sales Seasonally adjusted and trend estimates

where the one month percentage change in the trend and seasonally adjusted estimates are

Dec 2014 Jan 2015 Feb 2015 Mar 2015 Apr 2015 May 2015 Jun 2015 Jul 2015 Aug 2015
Trend -0.24 -0.18 -0.01 0.26 0.49 0.58 0.53 0.45 0.36
Seasonally adjusted -0.87 -0.77 -0.53 1.54 0.03 1.18 -0.04 0.71 0.19

So underlying one month movement in the trend has been strong since March 2015 even though the seasonally adjusted one month movements have bounced around. Even with a dip in the seasonally adjusted estimate in September 2015, it shouldn’t change the fundamental view of the underlying strengh in recent periods.

Over the length of the series the median for the one month percentage change in the trend for USA retail sales is 0.4%, so the recent activity is back in line with historical growth.

For background you can get the seasonally adjusted data here: http://www.census.gov/retail/marts/www/timeseries.html

USA unemployment estimates in September 2012

Who is Jack Welch? Well, I had never heard of the guy until recently, but according to Wikipedia he is worth around 700 million dollars and has probably earned himself the right to say what he likes. And good for him. He is obviously good at what he does and is an expert in his chosen field of making money. But is he really an expert on survey and statistical methods? Has he got a degree in statistics? Has he worked at a statistics institute and knows the effort and resources that goes into compiling estimates like this?

Probably none of these things based on his recent comment about the unemployment in the USA dropping to 7.8% in September from 8.3% in July over the last two months. Jack Welch was confident enough to come out and say recently “I was right about that strange jobs report”, and “Unbelievable jobs numbers..these Chicago guys will do anything..can’t debate so change numbers”. These are quotes referenced from the article linked below.

In the article he even states as evidence that he sat through some review of a dozen companies and suddenly a dozen companies are representative of the whole of the USA economy. That’s right. Twelve (12) companies and this guy thinks he is suddenly an expert on what the unemployment number should be for a whole economy?

“I sat through business reviews of a dozen companies last week as part of my work in the private sector, and not one reported better results in the third quarter compared with the second quarter. Several stayed about the same, the rest were down slightly.”
 

Well lets look at some of the actual data rather than rely on clouded personal perspectives.

Grabbing the data from here we can do a few simple things. Warning: for the crackpots out there who just write meaningless words and don’t know how to do any statistics, hold on to your chair as this means actually looking at data in an unbiased way, doing a histogram (look it up if you don’t know what it is) and we’ll also do some trend analysis.

So, firstly a histogram of the percentage point changes in the unemployment rate. What does this tell us? Well, since January 2002, there have been 12 times that the month to month movements have been -0.2 percentage points, and 4 times where the percentage point movements has been either -0.3 or -0.4 between consecutive months. So it is not unusual over the history of this series to see the unemployment rate move a similar magnitude to what was seen between July and September 2012. The spread of the historical percentage point movements in the month to month estimates is below, and you can see that they vary (as you would expect) and they have an average and median month to month movement of around zero.

USA unemployment histogram

Back to the actual published unemployed percentage estimates. They are plotted in the graph below. Back in April 2010, the unemployment rate was 9.9%, then 9.6% and then in June was 9.4%. A drop of -0.5 percentage points over two months. Similarly, back in November 2010, the unemployment rate was 9.8, and it went to 9.4 and then 9.1 in January 2011. Again, a drop of -0.7 percentage points over the space of two months! So the most recent movements are certainly not out of the ordinary, particularly when you consider the last two years of this series.

This is all shown by the seasonally adjusted and trend estimates on the graph below. In fact the trend is a better indicator given the recent volatility of this series and that has been on a downward slide since early 2010, and more recently from July 2011.

USA unemployment seasonally adjusted and trend September 2012

So, the data looks like the unemployment rate has been recovering for a while now. This is why that firstly its best to look over a few months figures (more than two), but secondly data needs to be gathered about the whole economy. Which is exactly what a statistical organisation is best placed to do. And in the end, isn’t some good news actually good?

USA retail sales for August 2012 and the trend

The latest USA retail estimates were published today for August 2012. In some previous posts, a Henderson filter was used to calculate a trend estimate and looked at the impact of revisions to both the seasonally adjusted and trend estimates. With the new estimate for August 2012 the analysis can be updated.

The seasonally adjusted estimate for August was reported to have risen by 0.9% when compared July 2012. This comes after a similar increase in the seasonally adjusted estimates last month. So two increases in a row. With a trend line calculated using a Henderson filter this looks like:

Trend estimate for USA retail trade series up to August 2012

So even with the two very large kick ups in the seasonally adjusted estimates, the trend has coped with this. This can help us put things into perspective and not get too carried away. While two recent large increases in the seasonally adjusted estimates are a good sign, normally we should take the rule of thumb that once is random, twice is coincidence and three times makes a trend. So in this respect, the impact of the two recent increases in the seasonally adjusted estimates are moderated down when considering the trend estimate.

The data we’ve used is:

Nov 2011 Dec 2011 Jan 2012 Feb 2012 Mar 2012 Apr 2012 May 2012 Jun 2012 Jul 2012 Aug 2012
Trend 0.63 0.62 0.53 0.35 0.16 0.03 -0.04 -0.03 0.01 0.02
Seasonally adjusted 0.47 0.04 0.64 1.03 0.37 -0.51 -0.12 -0.74 0.63 0.89

where the trend growth over the last five months is pretty much flat, even while the seasonally adjusted estimates have bounced all over the place. This compares to the previous months results of:

Nov 2011 Dec 2011 Jan 2012 Feb 2012 Mar 2012 Apr 2012 May 2012 Jun 2012 Jul 2012 Aug 2012
Trend 0.64 0.63 0.53 0.34 0.17 0.04 -0.05 -0.08 -0.12 n.a.
Seasonally adjusted 0.47 0.04 0.64 1.03 0.37 -0.51 -0.12 -0.73 0.81 n.a.

And how about the revisions? There are some revisions to the recent seasonally adjusted estimates, e.g. previously published of 0.81% now revised down to 0.63%. We can see that the last three trend estimates have been reasonably robust to the changes in the seasonally adjusted estimates. The blue line on the chart below shows the latest trend estimate for August 2012, with a slight revision upwards compared to the previous trend estimates for June and July 2012. This is primarily caused by the kick up in the most recent August 2012 estimate.

Revision to USA trend and seasonally adjusted estimates up to August 2012

You can also see how the seasonally adjusted estimates have been revised slightly with each release. As I mentioned previously, to get improved quality and the most up-to-date estimates, revisions to the original and seasonally adjusted should be taken on each month. Continual revisions mean continuous improvements. Who wouldn’t want the most up to date data?

With a similar increase in the seasonally adjusted estimates next month, we’d likely see some bigger revisions to the trend estimate. So currently we could say the trend growth in the USA retail sales over the last few months is relatively stable. So while there are some good signs in the seasonally adjusted estimates don’t get too excited.

USA retail sales for July 2012 and the trend

The USA retail trade estimates were released earlier today. Check out the full release here: http://www.census.gov/retail/.

Last month, we took the published seasonally adjusted data up to June 2012 and used a Henderson filter to estimate a trend series. See this post: http://www.seasonaladjustment.com/2012/07/18/three-falls-in-a-row-for-usa-retail-trade-estimates/ for all the details.

At the time we said “retail activity in the US has started to reach a turning point. But. And this is the big but. We’ll need more data to make sure. This is because it is difficult to understand if what we are seeing is due to random variation or a change in direction of the underlying trend.”

Now that this extra time point is available, how does this change the picture? A full time series with data up to July 2012 is below.

USA retail up to July 2012

The most recent seasonally adjusted estimate for July 2012 shows a 0.8% rise on the month. Calcuating a new trend estimate gives:

Nov 2011 Dec 2011 Jan 2012 Feb 2012 Mar 2012 Apr 2012 May 2012 Jun 2012 Jul 2012
Trend 0.64 0.63 0.53 0.34 0.17 0.04 -0.05 -0.08 -0.12
Seasonally adjusted 0.47 0.04 0.64 1.03 0.37 -0.51 -0.12 -0.73 0.81

Which compares to estimates that were calculated last month of

Nov 2011 Dec 2011 Jan 2012 Feb 2012 Mar 2012 Apr 2012 May 2012 Jun 2012 Jul 2012
Trend 0.66 0.63 0.50 0.33 0.19 0.08 -0.01 -0.07 n.a.
Seasonally adjusted 0.47 0.04 0.64 1.03 0.37 -0.51 -0.17 -0.48 n.a.

And pictorially this gives a revision to both the seasonally adjusted and trend estimates as below, where the green line is the old trend estimate using data up to June 2012, blue line is the new trend estimate using data up to July 2012, the red line is the latest seasonally adjusted data, and the light red line is the seasonally adjusted data at the release for June 2012.

USA retail up to July 2012 revision to June 2012

So, it was worth the wait for one more time point. Even with the promising kick up in the seasonally adjusted estimates of 0.8% in the most recent month (July 2012), retail trend growth in the United States has definitely flattened out.

With this new time point, there are two important observations. The first is the revision policy used by the USA for their seasonally adjusted estimates. From this chart it clearly shows that they have only revised the last couple of data points in the seasonally adjusted estimates, and the data points a year ago. This is a bit sneaky as we don’t have the latest up-to-date estimates for the whole series at each time point. But I’m pretty sure they would’ve used this in their internal calculations. However it would be more useful to revise the whole of the dataset to give everyone the most up to date information. The second point is that even with the data for July 2012 showing a large kick up in seasonally adjusted terms, the trend estimate that is derived is relatively unchanged. This can be one of the main reasons people shy away from the trend, but in this case, the story is pretty much the same as the trend estimate we observed last month. So rather than be excited about this 0.8% increase in the seasonally adjusted estimates, we should be mindful of the underlying trend.

Three falls in a row for USA retail trade estimates

The United States reported their Retail trade statistics the other day (July 16 2012). And quite rightly this picked up a bit of press where it was reported that they fell for the third month in a row. As is usually the case, these falls refer to changes in the seasonally adjusted estimates. And three falls in a row is starting to look like something bad for all those retailers (and perhaps the wider economy).

But given that the seasonally adjusted estimates still, by definition, contain a degree of volatility and also an underlying trend, we can go one better and derive our own smoothed estimate of the seasonally adjusted estimates. This will help us cut through the volatility and check out the underlying direction of the data.

We derived a trend estimate in the following way.

  1. Downloaded the data from here: http://www.census.gov/retail/marts/www/timeseries.html
  2. Plugged them into R (statistical package)
  3. Applied a 13 term Henderson filter to the full seasonally adjusted data to generate a trend estimate
  4. Used the ggplot2 package in R, which produces very nice plots (but can take some effort to get the data into the right format, e.g. a dataframe with all the right bits)
  5. And we get the following picture with a trend line…

usa-retail-july2012

I’ll leave the interpretation to the so called experts but it could be that with these three falls in a row in the seasonally adjusted estimates, retail activity in the US has started to reach a turning point. But. And this is the big but. We’ll need more data to make sure. This is because it is difficult to understand if what we are seeing is due to random variation or a change in direction of the underlying trend.

Just for completeness, the following table gives the one month percentage change in the different estimates. You can see that the one month change in the seasonally adjusted estimates can jump around, but the one month change in the trend is cutting through this noise and indicating a possible turning point.

Nov 2011 Dec 2011 Jan 2012 Feb 2012 Mar 2012 Apr 2012 May 2012 Jun 2012
Trend 0.66 0.63 0.50 0.33 0.19 0.08 -0.01 -0.07
Seasonally adjusted 0.47 0.04 0.64 1.03 0.37 -0.51 -0.17 -0.48

Other approaches could of course be used with different filters being applied and these would give slightly different results depending on the type and length of the filter used. It would have also been more useful if there was an official estimate of the trend as it would’ve saved some time as this is can be produced as a by-product of the seasonal adjustment process. One good thing about the data that can be downloaded from the census site is the availability of the seasonal factors, and also the sampling variability of the estimates. This is something you don’t often see being produced. So this is a big plus to have.

Also note that there are a few different estimates floating around in the dataset, particularly: advance estimates, preliminary estimates, revised estimates, and then suppressed and also not available. So this can potentially be a bit confusing as each of these estimates will have different characteristics. This is something to keep in mind if you’re grabbing the latest information from any data source is that it can often be revised as new data becomes available. Actually – this is really a good thing as it means that we at least have the best, latest and most up-to-date information.