« previous   misc   next »

San Francisco rents are not rising


By Patrick   Follow   Thu, 19 Apr 2012, 11:18am PDT   5,683 views   35 comments
Watch (0)   Share   Quote   Permalink   Like   Dislike  

Comments 1-35 of 35     Last »

Patrick   Thu, 19 Apr 2012, 1:01pm PDT   Share   Quote   Permalink   Like   Dislike     Comment 1

BTW, you can generate these graphs for any city you want, here:

http://patrick.net/housing/graphs.php

It's Craigslist data.

dublin hillz   Fri, 20 Apr 2012, 2:14am PDT   Share   Quote   Permalink   Like   Dislike     Comment 2

They may not be rising, but to pay close to $3,000 a month for rent plain sucks.

RentingForHalfTheCost   Sat, 21 Apr 2012, 4:19am PDT   Share   Quote   Permalink   Like   Dislike     Comment 3

dublin hillz says

They may not be rising, but to pay close to $3,000 a month for rent plain sucks.

Comparing 3k/mth to buying at 1mil is an easy decision if you don't buy into house price appreciation.

Patrick   Sat, 21 Apr 2012, 4:37am PDT   Share   Quote   Permalink   Like   Dislike     Comment 4

SFace says

Craigslist data again?

What is the mix of shared apartments vs. Condo vs. Sfh?

Yup. It's everything on Craigslist. In aggregate, the mix does not change.

So rents in SF really are flat overall.

Patrick   Sat, 21 Apr 2012, 5:24am PDT   Share   Quote   Permalink   Like   Dislike     Comment 5

What is "dead space"?

dunnross   Sat, 21 Apr 2012, 5:31am PDT   Share   Quote   Permalink   Like (2)   Dislike     Comment 6

Patrick says

What is "dead space"?

It's the space between SFAce's ears.

tatupu70   Sat, 21 Apr 2012, 5:33am PDT   Share   Quote   Permalink   Like   Dislike     Comment 7

Patrick--

He's saying to redraw the y axis in the graph and zoom in on 2500 - 3000. Any graph looks flat if you zoom out too far...

bubblesitter   Sat, 21 Apr 2012, 5:51am PDT   Share   Quote   Permalink   Like   Dislike     Comment 8

RentingForHalfTheCost says

Comparing 3k/mth to buying at 1mil is an easy decision if you don't buy into house price appreciation.

I'll pay 3K month rent rather then putting 200K down and taking out 800K mortgage. As always I say, DO YOUR MATH.

Patrick   Sat, 21 Apr 2012, 6:44am PDT   Share   Quote   Permalink   Like (1)   Dislike     Comment 9

tatupu70 says

He's saying to redraw the y axis in the graph and zoom in on 2500 - 3000. Any graph looks flat if you zoom out too far...

The one lesson I learned from "How To Lie With Statistics" is never to do that. Quote from p. 62 of the 39th printing:

(Basing a graph at zero) is very well if all you want to do is convey information. But suppose you wish to win an argument, shock a reader, move him into action, sell him something. For that, this chart lacks schmaltz. Chop off the bottom.

Now that's more like it. ... It is the same graph. Nothing has been falsified -- except the impression that it gives. ... Now that you have practiced to deceive, why stop with truncating? You have a further trick available ... Simply change the proportion...

Anyway, it goes on about how to make the graph as deceptive as possible. I don't want to deceive at all, so I do the honest thing and base the graph at zero.

justwantaniceplacetostay   Sat, 21 Apr 2012, 6:47am PDT   Share   Quote   Permalink   Like   Dislike     Comment 10

from my personal experience in some popular places that are livable south bay rents are definitely up since 2009. Average 2br used to be 1700 now its 2400. 3 brs are close to 3000 they used to be 2400.

still prefer to rent though..

RentingForHalfTheCost   Sat, 21 Apr 2012, 6:51am PDT   Share   Quote   Permalink   Like (1)   Dislike     Comment 11

Amazing. First people try to throw away the data. Then they complain about how it is represented, then they start throwing punches at the defenders. All that negative energy, rather than just taking the time to readjust their misconceptions about the market. "Rents are on fire! They are going up all over the BA! Renting now is crazy". You people are borderline insane. I've seen kids do better at realizing they are wrong. These are kids from Cupertino as well. :)

justme   Sat, 21 Apr 2012, 7:09am PDT   Share   Quote   Permalink   Like   Dislike     Comment 12

Right on, Patrick.

Patrick says

I don't want to deceive at all, so I do the honest thing and base the graph at zero.

I think however that the graphs would be easier to read if they were more quadratic (had a more even aspect ratio). Maybe even some grid lines that go all across instead of just tick marks. I dunno, just an idea.

tatupu70   Sat, 21 Apr 2012, 7:32am PDT   Share   Quote   Permalink   Like   Dislike (1)     Comment 13

Patrick says

Anyway, it goes on about how to make the graph as deceptive as possible. I don't want to deceive at all, so I do the honest thing and base the graph at zero.

Well, I'd argue that you are deceiving with the y axis as it is.. You should make the span of the y axis as small as is needed to illustrate the magnitude of change you are trying to show. If you want to be able to show a 10% change, you need to scale the y axis so a 10% change is visible.

How about this--what is the 1st point for rent and the last point?
If they are the same, then you are correct, that rents haven't risen.

Patrick   Sat, 21 Apr 2012, 7:44am PDT   Share   Quote   Permalink   Like   Dislike     Comment 14

justme says

Maybe even some grid lines that go all across instead of just tick marks.

Greyed-out grid lines are a good idea, but don't know how to do that in gnuplot. Got any example code?

tatupu70 says

You should make the span of the y axis as small as is needed to illustrate the magnitude of change you are trying to show.

The only honest graph shows the entire scale so you can clearly see the magnitude of any change in relation to the entire scale. Anything else is spin and marketing.

I didn't zoom anything out. Every graph on my site is always from zero, since I don't want to "lie with statistics" as the book so nicely explains.

Any significant percentage change will be clearly visible on an honest graph starting from zero.

tatupu70   Sat, 21 Apr 2012, 7:47am PDT   Share   Quote   Permalink   Like   Dislike (1)     Comment 15

Patrick says

The only honest graph shows the entire scale so you can clearly see the magnitude of any change in relation to the entire scale. Anything else is spin and marketing.
I didn't zoom anything out. Every graph on my site is always from zero, since I don't want to "lie with statistics" as the book so nicely explains.

Zooming in to adequately show the magnitude of change isn't deceiving. It's using the graph to serve the purpose. The graph you posted is pretty much useless because it can't show what it's intended to show.

You can't even see a 10% rise. I really think it's more deceiving as it is...

Do you have the first and last points?

Patrick   Sat, 21 Apr 2012, 8:07am PDT   Share   Quote   Permalink   Like   Dislike     Comment 16

I could get the first and last points, but why choose those? Because they seem to show a slight rise? Why not pick the peak around October and then the latest point?

What I'm trying to say is that arbitrary selection of two points to can always be done to spin things, and you see that the initial point I have is a bit lower than the latest point. So you might seize on that and say, look, these two points I choose prove my point, rents are rising. But obviously they are not, overall. They are fluctuating a bit. Believe your eyes.

Patrick   Sat, 21 Apr 2012, 8:10am PDT   Share   Quote   Permalink   Like   Dislike     Comment 17

orbitron says

I guess that change in debt isn't significant either...

The percentage change will be clearly visible if and only if you graph from zero. You can always zoom in arbitraily far to make an insignificant change look huge, but that's a kind of lying.

If you can't see a change in the national debt on a graph starting from zero, how significant could that change really be?

tatupu70   Sat, 21 Apr 2012, 8:14am PDT   Share   Quote   Permalink   Like   Dislike (1)     Comment 18

Patrick says

I could get the first and last points, but why choose those? Because they seem to show a slight rise? Why not pick the peak around October and then the latest point?

Well, if you are trying to determine if rents have risen over the past year, then the starting and ending points are pretty important.

Futher, if the points are significantly different (say 10%), then it shows your graph is deceiving. And the book's practice of starting every graph at zero is simplistic at best.

Patrick   Sat, 21 Apr 2012, 8:40am PDT   Share   Quote   Permalink   Like   Dislike     Comment 19

orbitron says

How about you make the Y size of your graph 3x as large and the X size 1/2 as large as what you have above and let us compare the two? They would both still start at zero. Can you please try this?

Wouldn't fit nicely on http://patrick.net/housing/graphs.php?uaddr=san+francisco%2C+ca anymore.

But wait a few minutes and I'll make the data available to Patrick.net Premium members.

tatupu70   Sat, 21 Apr 2012, 8:47am PDT   Share   Quote   Permalink   Like   Dislike     Comment 20

How about this--can you do a best fit? The equation would tell you if rents were rising or not.

bmwman91   Sat, 21 Apr 2012, 9:05am PDT   Share   Quote   Permalink   Like   Dislike     Comment 21

Patrick, shoot me the raw data. We can determine if there is a STATISTICALLY SIGNIFICANT change in rents over the time span given in the OP. If you have data from a few years ago to add in, that would be cool too.

In short, I'd fit a line to the rents, add a confidence interval to the fit, and then see if the mean rent falls within that CI. If the mean lies within the CI bands, then there is no statistically significant change (it's just noise).

bmwman91   Sat, 21 Apr 2012, 9:06am PDT   Share   Quote   Permalink   Like   Dislike     Comment 22

tatupu70 says

How about this--can you do a best fit? The equation would tell you if rents were rising or not.

A fit line doesn't necessarily tell you anything. If the R^2 value is really close to 1, then it is a good indicator that the fit line correlates to some underlying trend. You really want a confidence interval for the line though (95%).

Patrick   Sat, 21 Apr 2012, 9:35am PDT   Share   Quote   Permalink   Like   Dislike     Comment 23

OK, the data is now available via the link under each graph on the graphing page:

http://patrick.net/housing/graphs.php

The only catch is that you have to be subscribed to Patrick.net Premium. You don't have to pay anything for that, just write a short review of some property in the Open House Reviews forum and you automatically get a free week.

curious2   Sat, 21 Apr 2012, 11:05am PDT   Share   Quote   Permalink   Like   Dislike (1)     Comment 24

[...]

RentingForHalfTheCost   Sun, 22 Apr 2012, 9:17am PDT   Share   Quote   Permalink   Like   Dislike     Comment 25

curious2 says

The large empty space (dead space)

Hey it is one things to call SFace's brain area "dead space", but saying it is "large empty space" is just insulting. ;)

curious2   Mon, 23 Apr 2012, 5:47am PDT   Share   Quote   Permalink   Like   Dislike (1)     Comment 26

[...]

justme   Wed, 25 Apr 2012, 11:50am PDT   Share   Quote   Permalink   Like   Dislike     Comment 27

Patrick,

CR is repeating the SF Chronicle propaganda that rents in SF are rising.

http://www.calculatedriskblog.com/2012/04/san-francisco-rents-on-tear.html

You should have a word with him.

citizen jpp   Wed, 25 Apr 2012, 12:13pm PDT   Share   Quote   Permalink   Like   Dislike     Comment 28

Nice trend line for Ontario CA.

thomas.wong1986   Wed, 25 Apr 2012, 12:19pm PDT   Share   Quote   Permalink   Like   Dislike     Comment 29

justme says

You should have a word with him.

What landlords do and how business reacts to this are often found going in opposite directions. The word would be "business-killer". We dont have the fundemental basis for such a move.

We do have lots of propaganda regards to Twitter doing business, but not even when Bank of America was in SF didnt have such an influence on rents.

Patrick   Fri, 27 Apr 2012, 10:09am PDT   Share   Quote   Permalink   Like   Dislike     Comment 30

justme says

Patrick,

CR is repeating the SF Chronicle propaganda that rents in SF are rising.

http://www.calculatedriskblog.com/2012/04/san-francisco-rents-on-tear.html

You should have a word with him.

Thanks! Finally mentioned it to him, here:

http://www.hoocoodanode.org/node/15727

Daytona   Wed, 2 May 2012, 4:16am PDT   Share   Quote   Permalink   Like   Dislike     Comment 31

According to Avalon Bay, rents are rising around 7-8% YOY in NorCal.

http://seekingalpha.com/article/533781-avalonbay-communities-ceo-discusses-q1-2012-results-earnings-call-transcript

"renters should have the capacity to pay higher rents over the next 2 to 3 years. In fact, Witten Advisers estimates that nationally, apartment market rents will grow faster than their long term trend by 150 basis points through 2014.

During the first quarter, year-over-year rental rate performance in our portfolio accelerated from Q4, reflecting the pickup in economic and job growth in late 2011 and early 2012. The average rent change improved by 100 to 150 basis points from Q4, averaging over 4.5% in Q1, with renewals up by 6% to 6.5% and new move-ins by 2% to 3%. Looking forward, these favorable trends should continue to Q2 as offers for renewal for the April to June period are up by 6% to 6.5%, and new move-ins in April are projected to be right around 4%. The West Coast should continue to outperform as offers for renewals are generally in the 7% to 8% range, while offers on the East Coast are in the 5% to 6% range."

tr9500   Wed, 2 May 2012, 4:27am PDT   Share   Quote   Permalink   Like   Dislike     Comment 32

This isn't "rent", this is "asking" price... please label your data accordingly...

tr9500   Wed, 2 May 2012, 4:30am PDT   Share   Quote   Permalink   Like   Dislike     Comment 33

Also, please be mindful of your data sources. Not sure who 'avalon bay' is, but if they are broker/apartment owner/etc, they have strong vested interest in saying rents are rising.

toothfairy   Fri, 11 May 2012, 6:11am PDT   Share   Quote   Permalink   Like   Dislike     Comment 34

maybe my eyes are bad but it looks like your graph is rising from 2800 to about 3100

That's a 10% increase

supersunken   Sat, 12 May 2012, 12:22pm PDT   Share   Quote   Permalink   Like   Dislike     Comment 35

SF has rent control. It's more difficult for a landlord to kick out a renter not paying in SF then other counties. Remember, SF is it's own county.

Patrick is moderator of this thread.

Email

Username

Watch comments by email

home   top   share   link sharer   users   register   best of   about   questions or suggestions? write p@patrick.net