Saturday, January 28, 2012

When a challenge isn't

Well, this is curious semantics.  A news article just out says: Many city homeowners may find property tax payments reduced.  In it one of the first lines is:
"This estimate was released today by RealSTATS, which challenged University of Pittsburgh economist Chrisotpher Briem who earlier this month the new assessments would reduce property taxes for two out of three homeowners."
The article has data saying 54% of homeowners will likely see tax bills go down as a result of new assessment numbers.  I had said here that it is likely 65%.    Strange semantics to say there is any 'challenge' going on. Seems to me they are agreeing with me more than anything else.  I don't get it.  But like I said, nobody asked me on this.

Anyway, just to avoid any confusion.  I have put the full distribution of values I scraped from the county assessment web site online and explained in that previous post. I put my full dataset online as well exactly where the distribution comes from for those who want to look at it.  Feel free to go and use it to do your own calculations. Remember, the old article being references was done at a time when the county was not releasing the assessment data even to the media even though it was central to the entire public debate at the time.  The real issue here is that it is about open data, transparency and what the public should not be guessing about.  There need not be any confusion or debating numbers here.  In itself, figuring if your taxes go up or down need not be proverbial rocket science.

Note I used data directly from the assessment which is the basis for tax bills.  I think the Realstats data is great, but I dont think they have access to the actual assessment data yet do they? (see update below)  If they do I really want to know how they got it since to my knowledge the county has not released it yet. I also can not tell from the article if the realstats estimate of the tax rate change is being based entirely on residental values, or if it includes the reported commercial value increases as well.  The higher increase in commerical values would create a bigger drop in the revenue neutral rate and thus a bigger percentage of homeowners that will face tax bills. 

But I don't quite see the 'challenge' part anywhere. This news item says there is an estimate that 54% of homeowners will see a price drop. I came up with 65%.   We can argue over the 10% there, but the point is that there is not additional analysis saying that a majority of homeowners will see a price drop in the new assessment. Something that I suspect a majority of people still find suprising given the political rhetoric.  I kind of see it as the opposite of any 'challenge'.  I don't get it??  

In fact, they really agree with the big point I made earlier.  Per the article summarizing the realstats data "The areas in which taxes are going up are Lawrenceville, Stanton Heights, Morningside, South Side, Bloomfield, Highland Park and East Liberty".   I think that comes awfully close to what I said and was quoted in the PG:
The neighborhoods where assessment increases are most concentrated include the South Side Flats, Lawrenceville, the Mexican War streets and portions of Highland Park and East Liberty. Many city and county officials have pointed to those neighborhoods as redevelopment success stories where new residents are moving in and real estate prices are rising, Mr. Briem wrote.
What am I missing?  I wish everyone who 'challenged' me would so agree with me?  I can only speculate a bit on where the difference is from 55%, but as I said.. I based my ratio on the actual assessment data and not on an analysis of recent sales data which I suspect is what Relastats has since that is what they distribute regularly. Also as I explained, I took out parcels that were likely not habitable parcels which are a big chunk of all the parcels coded as 'residential' in the assessment data for the county.

You know what would be fun?  The county distributes an actual dataset that the PG could then use to come up with a number themselves and then not get caught up in this.  Why argue over a number that clearly can be calculated without too much speculation needed. Of course, it is unknown to me if the PG has yet to be able to acquire the actual data which I know they have been trying to get since December to little avail.

Update:  So I've seen the Realstats report though it still isn't on their web site yet.  From what I can tell he has somehow acquired the actual property assessment data, which is a big mystery since everyone I know including public officials who needed it did not have it. Though they may have it now.  I think, and this is speculation a bit that Realstats is talking about a different set of issues. I was looking at City of Pittsburgh residents which at the time was clearly the only thing any of us could have been talking about since the county had only released the residential values.   I think the count of residential properties in the city of Pittsburgh and Mount Oliver is close 119,100.  There are another 30K or so commercial, or government parcels as well.  So in the Realstats report I see he references 128 thousand parcels which I am pretty sure is talking about a lot more pacels than matters if you are talking about Pittsburgh residents and residential owners. I bet it is including commercial properties or a lot of properties that are clearly vacant in some form.  So the numbers they are compiling are just not comparable to what I did and thus there are some difference between them. 

Again, my data, explanation of how it was acquired, distribution of results, presumptions and explantion are all ONLINE HERE.  Everyone can go poke a it and make their own conclusions. It might be good if Realstats put their complete dataset online as I have done for comparison.

If there is any difference in the final results it is because the county was not putting any of this data out at the time.. Right when everyone was getting their assessments and jumping to a lot of conclusions without much clarification from anyone out there.  So it would have been nice if Realstats had put out this analysis of the actual assessment data when it really mattered, but it's great they are in large agreement now that folks have started to calm down.  I would hope that with the actual assessment data we might reach a more definitive answer.  It just isn't comparable to say "my numbers" are different from what comes out later.  It isn't quite how this all works.


Blogger Infinonymous said...

Don't snag on this, Mr. Briem. The Post-Gazette carried a poor report, poorly written. There does not seem to be much more to this than that.

Here's one thing the Post-Gazette (like others who rely on information provided by RealSTATS) has not reported:

Who are RealSTATs' clients?

Saturday, January 28, 2012 12:57:00 PM  
Anonymous BrianTH said...

Seriously, no need to be concerned. You are right this is far more vindication than challenge, and your estimate was way more timely.

And incidentally--eastern suburbs have mailed (we got our Wilkinsburg assessment today). If you wanted to do similar estimates for those jurisdictions that would be fantastic--I promise I won't complain if you are later "challenged" in this fashion.

Saturday, January 28, 2012 2:53:00 PM  
Blogger Tom Moertel said...


For what it's worth, if I take your data set and do a straightforward calculation on all the properties – no Homestead adjustments, no removal of sub-$3000 properties, and just a naive anti-windfall adjustment – I get the 54% figure, too. Here's a plot of the corresponding distribution (cumulative).

If I remove the properties having a 2011 assessment of less than $3000 and then recompute, I get 57%.

So do the Homestead adjustments account for the remaining difference?

I'll throw the analysis online if you're curious. (Okay, I dropped it on my GitHub account: the analysis.)


Saturday, January 28, 2012 3:29:00 PM  
Blogger C. Briem said...

That's great. I think we are in agreement on the numbers, but are just talking about some differnt things. Lets be sure though.

In my data I have 118K records. Taking out everything under 3000 its 105,238 parcels. If just doing the distribution of the change in that data I think it is indeed as you say 54% below median.

But that isn't the question by itself(or wasnt my question). These are residential parcels since that is all the county had put online at the time and is really the universe that seems to be at issue if you are talking about city residents. Thus the distribution picture I put up has 3 points, the revenue neutral lines for residential, commercial and a composite.

So of the 105K parcels below the revenue neutral point (1.453 is what I am getting now) just among residential parcels then I do get 57% or so. But the law is about revenue neutrality overall which I think is the 1.58 point. Take the proportion of these parcels under 1.58 to get 65%. I just now am getting 64.3% and will go back and see what is causing the difference unless I was just overly optimistic in the rounding.

I think this is the only issue with the realstats data which I think is mixing the commercial and residential properties. My posts here have alway only been for the residential properties.. which if that was not clear it had to have been the case since that was all that the county had put online at the time. It would be quite sime time later before the county put the commercial property values up there.

I would be interested if you concur with the numbers just for my own sanity... anyone posting R output I will trust no matter.

Saturday, January 28, 2012 3:57:00 PM  
Blogger C. Briem said...

and for BrianTH.. I have to believe the media has some aggregate numbers from the county to report in this round.. if not then I have failed completely I guess. I would not be surpised if some of that is not in the paper tomorrow.. But I will see if I can do it just for Wilkinsburg. I can tell by calls I have been getting that is a particular concern which it may indeed be since Wilkinsburg is more complex than people think in terms of how differnet neighborhoods are doing.

Saturday, January 28, 2012 4:06:00 PM  
Blogger C. Briem said...

and I will say I trust all electrical engineers after having spent 4 years integrating imaginary numbers doing Fourier analysis.

gives me shivers just remembering it.


Saturday, January 28, 2012 4:09:00 PM  
Blogger Tom Moertel said...


Indeed, my calculations match yours. For the initial 54% figure, my "anti-windfall" adjustment corresponds to your original 1.46 point, representing residential properties:

> 1 / anti_windfall_scale_factor
[1] 1.46403

And, when I use the all-inclusive revenue adjustment of 1.58, I get 64.3% of properties having lower taxes, as you reported. (Updated calculations).

So I'm getting what you're getting. And your explanation for why RealSTATS is getting something else seems to match the available evidence.


Saturday, January 28, 2012 4:51:00 PM  
Blogger C. Briem said...

thanks much.. I can sleep now. Although I am a bit angry at the 64.3 vs. 65, but I will have to see how I did that.

Saturday, January 28, 2012 5:11:00 PM  
Anonymous BrianTH said...

Don't let it get around you are taking requests . . .

. . . although Wilkinsburg should actually be a pretty interesting and educational case (as you note, there should be a real range of outcomes in different areas, and the math may work out a bit differently than some are expecting).

Saturday, January 28, 2012 6:23:00 PM  
Blogger C. Briem said...

Just to summarize and cycle back. I am pretty sure the Realstats data is not looking at just the residential parcels as I was. I think they are lumping residential and commercial parcels together. Nothing wrong with that, but it is not what I was doing as I hope was clear with all graphs labeled "Residential parcels" and which I explained to everyone I talked to as well.

and the curious thing about the Post-Gazette version of that article is that for something so centered on the ad hominum you might think they might give a call for a reaction or just to make sure it makes sense. But nothing. No call, email, text or smoke signal. In other circumstances, say if it was another publication I might have fired off an unsolicited email just incase they were trying to get in touch, but couldn't. Since I suspect 3/4ths of the PG newsroom has me in their virtual Rolodex I figure there was no interest in a response. Strange, but it is what it is. My only real beef is that at the end of the day the public is now likely a bit more confused.. or maybe not if they take this more as I do, confirmation that the panic was not justified when the numbers first started coming out.

Saturday, January 28, 2012 6:31:00 PM  
Blogger C. Briem said...

for BrianTh who has earned a lot of frequent commenter points: Wilkinsburg reassessments by parcel

If you see anything that does not look right let me know.

but for those who want to check this first pass. For taxable properties in Wilkinsburgh I am getting aggregate valuation increase of +24.67%. Gratuitous significant digits thrown in for good measure.

Saturday, January 28, 2012 11:20:00 PM  
Blogger C. Briem said...

to be crystal.. that is for taxable property values. They actually did not post any of the exempt valuations anyway.

Saturday, January 28, 2012 11:25:00 PM  
Anonymous BrianTH said...

I can't thank you enough--that's extremely helpful. But thank you!

Incidentally, the headlines might as well read "Chris Briem was right." I really don't think that one misleading line will take away from that.

Sunday, January 29, 2012 12:07:00 AM  
Blogger C. Briem said...

scratch the 'h' of course..

Sunday, January 29, 2012 12:30:00 AM  
Blogger C. Briem said...

Well... I don't know if it will work, but I've started what I will call the 'big scrape' to see if I can do the whole county. Either because of inefficiency in my program or response time it looks like it could take 24+ hours even if it works to completion.

Ideally it won't be needed and it will just guilt the county into putting out the complete data that everyone keeps asking me for since there is about to be an immense confusion hitting the streets as suburbs get their values and need to try and figure out their notional taxes based on changes in BOTH their municipality and their school district.

Sunday, January 29, 2012 9:26:00 AM  
Anonymous BrianTH said...

Yeah, it is truly shameful the County isn't immediately releasing the summary numbers plus a dataset for each taxing jurisdiction, but it is regrettably predictable given the political dynamic.

So I personally hope you keep going with this project, even though in an ideal world it wouldn't be necessary.

By the way, I think the Wilkinsburg numbers were "surprising" to some people, although they were pretty much what I expected. This whole notion of the aggregate appreciating more than the median residence is still tough for some people to grasp intuitively (and this is among the people who get the basic concepts involved in the anti-windfall laws).

Sunday, January 29, 2012 11:56:00 AM  
Anonymous BrianTH said...

So did RealSTATs see you as invading their turf? Because anyone reading you in real time knew you were talking about taxable residential property owners, and that memo seems to deliberately attempt to confuse readers on that point.

Sunday, January 29, 2012 1:43:00 PM  
Blogger C. Briem said...

I am sure there is no deliberate attempt to confuse...

Like most things, the answer depends on what question is asked. I think we are asking slighly different questions.

Sunday, January 29, 2012 2:45:00 PM  

Post a Comment

<< Home