SF Bay Area (2): Live-Work-Commute

SF Bay Area (2): Live-Work-Commute

Introduction:

This is part 2 of an exploration of jobs, commuting, and housing in the San Francisco Bay area.
Link to this post (for sharing): https://meetingthetwain.blogspot.com/2018/03/sf-bay-area-2-live-work-commute.html
Part 1 is here: http://meetingthetwain.blogspot.com/2018/03/sf-bay-area-1-live-work-commute.html

In part 1, we saw that San Francisco and Santa Clara Counties had added the most jobs resulting in a "housing deficit" in the two counties while other counties were either in approximate balance or had a "housing surplus" (i.e., bedroom communities).

Now, in part 2, we look at how this changed over the period 2002 to 2015.    We find that most counties had a pretty consistent "housing deficit" or "surplus" over the 14 years looked at.  The big exception was San Francisco.  San Francisco started out with a large housing deficit which then got much worse very quickly as it added a lot of jobs and very little housing over most of the period examined.

Then we explore the effect on commuting.  We see a 16% growth in workers in the 14 years since 2002 but a much bigger growth in long distance commuters.  Those commuting 25-to-50 miles and those commuting over 50 miles grew 26% and 49% respectively.  This has caused a massive impact on traffic.  Looking at the total Vehicle Miles Traveled (VMT) we find those long distance commuters - those who travel over 25 miles - are only 27% of workers but cause 66% of the traffic (VMT).

We are using the tool "OnTheMap" which I have described here:
http://meetingthetwain.blogspot.com/2016/12/how-to-use-onthemap.html

Job Increases and Decreases:

In the 9 county region, from 2002 to 2004, the number of full-time jobs decreased (aftermath of "dot-com" bubble), recovering to 2002 level in 2008, and then decreasing again after the housing bubble burst.  The number of jobs did not get back to 2002 levels until 2011.  Job growth did not really take off until 2012.  Click on graph to enlarge:

557,000 jobs added since 2009 = 19% increase = avg. of 3% per year.

467,000 since 2002 = 16% increase over 14 years = avg of 1.1% per year

Same thing - less dramatic - based at zero instead of 2,700,000:

Number of jobs remarkably steady for 10 years then (barely) exceeds 2002 level in 2012.

You often hear "NNN thousand jobs have been created since 2010 and not enough housing has been built."  Several reasons for that.  Reason 1: In 2010, the number of full time workers in the SF Bay areas was below 2002 levels (see above chart).  If there was enough housing in 2002, there was no need to build any more in 2010.  Reason 2: Housing prices were declining from 2007 through 2012.  No one will build new houses when prices for existing houses are going down.  See the following graph:

https://www.paragon-re.com/trend/3-recessions-2-bubbles-and-a-baby

Declining housing prices is a clear message there are more houses than buyers - too many houses for the existing market - no more needed.

It takes 2-years from breaking ground to finishing construction of an apartment.  It takes 5 minutes for a CEO to decide to hire more employees.  The housing supply will not catch up with demand until the increase in demand (i.e., increase in workers) pauses for a few years.  We discussed this in more detail here: http://meetingthetwain.blogspot.com/2018/01/housing-jan-2017.html

Job Growth:

We show job growth by county in the following chart (click chart to enlarge).  To compare the counties, we start each county at 100% in 2002 and chart it's growth.  San Francisco had by far the greatest job growth in percentage terms from 2002 at 35%.  Santa Clara is number 2 at 16% with the other counties at smaller percentages down to Alameda at 7%. As before we group the 4 North Bay counties of Marin, Sonoma, Napa, and Solano as one entity.  (click chart to enlarge)

San Francisco stands out as most job growth at 35% higher than 2002

Alameda County had least at 7%

Several counties with 9-10 years negative job growth from 2002 until 2011-2013

In the above chart we see that all counties had negative job growth in the aftermath of the 1990's Dot-Com bubble.  Most counties had dramatic downturns after the "dot-com" bubble burst in 2000.  They had barely recovered in 2007 when the "housing bubble" burst.  San Mateo, Santa Clara, and Alameda counties had extended negative job growth and didn't get back to 2002 job levels for 10 or 11 years (2012 and 2013).

Some counties recovered more quickly, notably San Francisco.  San Francisco stands out further in having only a very minor 1 year decrease (2009) in job growth due to the 2007-2012 "Great Recession".  Other cities in the SF Bay area suffered much longer declines in jobs lasting 3-5 years.  San Francisco exceeded 2002 job levels by 2007 and never looked back.

Creating jobs is all well and good, but the question of the day is housing.  Did housing growth keep up with job growth (where there was growth)?  We look at each county from 2002 to 2015 in terms of "housing deficit".  If a county has a 20% more jobs than resident workers it has a "housing deficit" of -20%.  If there are 10% more jobs than resident workers, the county has a "housing surplus" of 10%.

The following chart (click on chart to enlarge) shows by county the housing deficit or surplus over the 14 years from 2002 to 2015. It is information heavy so we discuss several counties below.  Click Chart to enlarge.

Percentage Housing Deficit

San Francisco 55% Housing Deficit

Contra Costa 25% Housing Surplus

Horizontal Red Line in Center is "Perfect Balance" of Jobs-Housing

The key point is that counties are pretty consistent over the 14 year period.  They are in a negative "housing deficit" (below the red line in the middle), "near balance" or positive "surplus" pretty consistently throughout the 14 years 2002 - 2015.

Looking at several counties in detail:

Contra Costa had a housing surplus of +25% in 2002, and exactly the same surplus in 2015.  In the intervening years the surplus bounced between a narrow range of +19% and +26% averaging +22.5% +/- 3.5%.  This means that on average, nearly 1 in 4 residents commuted out of  the county for a job in another county.  (Click on chart to enlarge:)

Santa Clara County, had an even narrower range (but a housing deficit) of -18% to -13% averaging -15.5% +/- 2.5%.  This means that, on average, for every 7 jobs there was housing for only 6 workers - another worker had to commute in from a neighboring county.

San Francisco is the exception - a big one!  It started bad with a -42% housing deficit and deteriorated further to -55%.  From very bad to even worse!  This means that on average, for every 3 jobs, there was housing for fewer than 2 workers - a third worker had to commute in from another county.

Commuting and VMT

As the jobs increase in one place but the housing increases in another place, more and more workers need to commute further and further.  While this is to be expected, the extent of this is astonishing.  The following graph shows a huge increase - about 50% - in the percentage of commuters who travel more than 50 miles to get to work (purple line).  The percentage commuting 25 to 50 miles (green line) increases over 25%.  The percentage commuting less than 10 miles (blue line) actually declined from 2002 until 2014 only going above the 2002 level in 2015. (click chart to enlarge).

The following graph shows the numbers of those commuting in each commute range:

The numbers of commuters in the longer ranges don't look too big compared to the number in the shorter ranges.  But note that there is almost a 50% increase in the longest range commuters (over 50 miles) from 315,000 to 469,000.  These long distance commuters have an out-sized effect on "Vehicle Miles Traveled" (VMT) which is the standard metric for traffic.  VMT is simply the sum of each commuter's miles traveled.

For example, if 10 people each commute 10 miles each, that gives us:
   10 vehicles x 10 miles traveled each = 100 VMT (Vehicle Miles Traveled)

If one more person is added who commutes 100 miles, that changes the calculation to
(10 x 10) + (1 x 100) = 200 VMT.  See graphic below:

Lots of Cars drive a little or One cars drives a lot.  Same problem.

In this example, a 10% increase in commuters resulted in double the VMT because of the longer distance traveled by that commuter.  That means double the gas consumed, twice as much commute infrastructure required, 100% more Green House Gases emitted.

Next chart is the VMT change by year for the nine-county SF Bay Area.  You notice a complete reversal in the VMT compared to the earlier commute range population chart.  There are over 3 times more people commuting less than 10 miles but the longest range commuters (greater than 50 miles) contribute nearly 4 times the VMT.

What does this mean from a practical point of view for you, the commuter?  Consider two situations where you are driving 50 miles on a very crowded freeway.

In case A, everyone else is driving the exact same distance as you, and it is stop-and-go all the way because everyone wants to drive 50 miles.

In case B it is just as crowded but now everyone (except you) enters the freeway, drives exactly 1 mile, and then exits the freeway to be replaced by another commuter who does the same.

From your point of view, there is no difference.  The road is just as crowded in either case.  However, in case B, lots of people go short distances - the freeway has accommodated 50 times more commuters in case B.

What we have in the SF Bay area is Case A.  A small proportion, 27%, of long distance commuters (greater than 25 miles) are clogging the roads.  The ever worsening traffic is due to the 30% increase in VMT which is mostly due to the long range commuters.  Let's compare 2002 to 2015 to see the change:

The orange and green segments (commutes "over-50" miles and "25-to-50" miles, respectively) are only 27% of the drivers.  That 27% of long distance commuters (over 25 miles) cause 45,000,000 of the 69,000,000 Vehicle Miles Traveled.  I.e., 27% of the drivers cause 66% of the total Vehicle Miles Traveled and therefore traffic and congestion.

(Note on VMT calculation: I took the midpoint of each range so 0-10 miles range was computed using 5 miles times the number of those commuting less than 10 miles.  Similarly for other ranges, 10-24 miles was 17 miles x number of commuters, 25-50 was 37.5 miles x number of commuters, for over 50-mile commutes I used 60 miles x number of commuters.)

Much more to say but this is getting too long.  Part 3 coming soon.