80% of... What?

This Wall Street Journal article has a chart which could say so much more.

It's a bar chart.  The height of the red line shows the share, in percentage points, of asset classes that had a loss each year.  These asset classes are defined by Deutsche Bank.  It's an interesting idea, but raises a lot more questions than it answers.

(A terrific score on the GMAT will take your career to the next level.  Contact the GMAT Doctor.)

How did they pick the asset classes?  It's possible to define different portfolios to make these numbers come out any way you want.

How big are they, relative to each other?  Are any correlated with each other?  Imagine a world where there were 90 asset classes that all had a loss but are tiny in size, while most people's portfolios are in ten other asset classes that make up the majority of everyone's portfolio.  Would that be a good year or a bad year?

This chart treats any size loss the same.  Try convincing someone that a 1% loss is similar to a 25% loss.

It's not bad to have a chart that raises questions, but a more complete analysis would anticipate such questions.

Also, on a practical level, couldn't you make money by shorting some of these asset classes?

The Best Become the Worst

I really like this graph in today's Wall Street Journal.

It's from this article.  The author argues that the sectors which performed best for the first nine months of the year also had the worst September.  It's a vertical bar chart.  The blue lines are performance for the first nine months of the year, while the red lines are performance for September.  The middle line is zero, with bars to the right showing losses and those to the left showing gains.  Visually, the argument is that big blue bars facing right are usually married to big red bars facing left.

It would be more conventional to show a scatter plot to demonstrate the negative correlation, but some how the bars give the reader a better sense of relative sizes.  Dots might not adequately represent the magnitude of these losses.

My only suggestion would be to order these industries by performance during the first three quarters of the year, perhaps with the best performers (looks like a tie between technology and consumer discretionary products) on top and the worst performers (probably consumer staples) at the bottom.  It would let you see if the red lines descend in size similarly at a glance, rather than having to go case by case. 

Eyeballs vs. Simplicity: Today's Dilemna

A recent post by my colleague Leonard Kiefer at Freddie Mac caught my eye.  You can see the post on Linked In here.  You have to scroll down a little.

The chart is quite eye-catching.  Each box shows the share of cash-out refinances relative to total conventional refinances in the third quarter of each year since 1999.  If a mortgage is eligible for purchase by Freddie Mac or Fannie Mae (meaning it is below a certain dollar amount and meets certain underwriting criteria such as loan-to-value ratio and credit score), then it is considered 'conventional.'  A cash out refinance is one where the borrower has an existing mortgage for $X dollars and takes out a new loan for more than $X, using the proceeds to pay fees and have extra cash available for expenses such as health-care or tuition.  A entire house means that the share is 100% while a few scattered points means only a tiny share of conventional mortgages are cash-outs.  

The chart is interesting in that it shows a strong resurgence in the cash-out share.  I think the chart is far too complicated for showing a simple data series.  The kind of line charts in Excel would make the pattern clearer.  It's very hard, in this array format, to compare numbers from one row with those in another. 

I'm guessing Leonard chose this alternative method in order to catch eyeballs and get more clicks.  This is an example of a dilemna for those of us who work in data visualization.  We want you to look.  Sometimes we are motivated to sacrifice clarity to get your attention.  It may have been a simple matter of aesthetic taste.

Note also that this is measured as a share of conventional refinance mortgages.  The denominator, the number of conventional mortgage refinances, changes throughout this time period.  While interest rates were falling, for example, refinances were quite common while they are less so today.  It would be nice to include information on the number of refinances with this graph, perhaps on the same graph or in an accompanying chart.  

I'd love to help you understand how to use data visualization to communicate clearly, and to understand it well for business, and in particular, for the relevant section of the GMAT.  Please don't be a stranger!  Get in touch with me here.

Wave hello to more chart junk!

The NY Times has a beautiful idea for a data presentation.  The link  is here, and I've pasted the graphic below.

Each little business suit (why not some dresses?!) represents a congressional district.  The y-axis shows the how 'Republican' the district is, by indicating the share of voters that chose the Republican candidate in the last two Presidential elections, relative to the entire country.  The color shows how the district voted, with blue indicating a Democrat victory.  The theory behind the graph is that the more 'Republican' the district has been in the past, the harder it would be to elect a Democratic house member.  The higher the number of blue suits, the more impressive a win.

The takeaway from the picture is that the Democrats did pretty well with districts that are just a bit more Republican than the rest of the country, but then their victories faded out.  This kind of analysis is really cool and creative.  It's intuitive to represent a difficult victory this way.

The concept is a bit complicated - look at how many words it took to explain it!  It  is especially worthwhile to stay simple when the concept is complex. 

Here's some suggestions:

• Get rid of the blue wave, as it's cute but adds nothing to our understanding.  
• A vertical bar chart would do a better job than the chauvinist blue suits.  
• Why did they order the suits in such a manner that some reds wind up to the left of blues, while others are on the right side.  Does that teach something, or is it just a distraction?
• The chart is from last week.  I wonder if it would have been hard for the NY Times to update it, as it is still prominently displayed on their web site and a lot of the races are yellow for 'undetermined.'

I have to admit that I like the hill, though.  It's cute. 

I'd love to help you understand how to use data visualization to communicate clearly, and to understand it well for business, and in particular, for the relevant section of the GMAT.  Please don't be a stranger!  Get in touch with me here.

Headed for a Crash? The Buffet Ratio.

This is an incredibly elegant graph from https://www.gurufocus.com/stock-market-valuations.php.  It shows the ratio of US total stock market capitalization to US GDP.  They attribute this ratio to Warren Buffett.  The ratio is higher than any time since the height of the last tech bubble.  And we know what happened then.

I have some questions about the economics behind using this ratio.  For example, there's been a huge move towards privatization of public companies and an increase in their debt, which means that the stock market capitalization in one way measures fewer companies than in the past, but on the other hand it is understating how large corporate assets (the sum of equity and debt) is relative to GDP.  So don't use this by itself to decide anything.  However, I think they get points for making a troubling pattern apparent with very accessible data and graphics.

Normally I'm not a fan of colored or shaded backgrounds, but in this case it's pretty restrained and pretty attractive so they get a pass.


I'd love to help you understand how to use data visualization to communicate clearly, and to understand it well for business, and in particular, for the relevant section of the GMAT.  Please don't be a stranger!  Get in touch with me here.

JobsNotMobs

It's the election tomorrow, and I'd lose my Blogger's license if I didn't write something about it.

The NY Times has a very cool infographic here:

https://www.nytimes.com/interactive/2018/11/04/technology/jobs-not-mobs.html

It is a long horizontal line that tracks usage of this really horrendous video clip, associated with the tag JobsNotMobs.  It tracks how often this meme is used, and highlights events that were associated with changes in the pattern.  It has a nice use of color, with  orange and blue circles representing how often it was posted on two platforms:  Reddit and everything else.

It reminds me of the famous Minard map of Napoleon's march to Moscow.  https://youtu.be/laXh2cgE2g0

Other than 'bravo,' I have one suggestion.  It's hard to see the whole time series at once.  It would have been cool if one could zoom in or zoom out of the graphic.

I also feel obligated to say that the message this video clip sends is horrifying.  My grandparents and great-grandparents immigrated from this country.  I'm so grateful to them for having been brave and given me the opportunities that came with being born here.  They also faced prejudice, the same prejudice that led to the US turning back boatloads of Jews fleeing Hitler.

See https://youtu.be/ygVX1z6tDGI for a more eloquent and entertaining perspective on immigration policy.


I'd love to help you understand how to use data visualization to communicate clearly, and to understand it well for business, and in particular, for the relevant section of the GMAT.  Please don't be a stranger!  Get in touch with me here.

'Chart Junk'

Here's something tweeted from FT's Alphaville.  https://ftalphaville.ft.com/

It was created by Standard Chartered Bank.  It's a bar chart, where the top or bottom of each rectangle represents a value.  It's fine, but as is often in graphics, 'less is more.'  There's a lot of ink for each data point, where a simple line, perhaps with points to highlight that these are discrete observations, would have done the job.  One of my teachers called such unnecessary ink 'chart junk.' 

I like bar charts better when the bars actually mean something such as components of the overall number you are tracking.  Bloomberg did this with the chart in my prior blog post https://betterdatavisualization.blogspot.com/2018/10/bloomberg-bar-charts-shows-more-is-less.html.  Unfortunately, they did some other things to that ruined it.