Monday, October 31, 2011

Brown Town 2.0

Missed this in what is now last week's (October 22-28) edition of the Economist magazine.  See the article: Smaller is more beautiful, subtitled Many other cities are battling problems almost as acute as Detroit’s.

If you are waiting for a smackdown on Steeltown, you will be left hanging.  The quotes about Pittsburgh just start with this accolade:
Pittsburgh is often pointed to as a model for other shrinking cities. Its revival since its steel industry collapsed in the early 1980s is partly thanks to good long-term planning. Under the leadership of Tom Murphy, a three-term mayor, more than 1,000 acres of abandoned, blighted industrial land in Pittsburgh was cleaned up and is now thriving commercial, retail, residential and public space. Once lined with factories the city’s waterfront has been given over to parks. Mr Murphy oversaw the development of more than 25 miles of new trails alongside the river and urban green space. He helped develop public-private partnerships which leveraged $4.8 billion in economic development.
It even ends on this note:
Pittsburgh, for instance, reinvented itself as a successful tech and health hub, even as its population continues to fall. As Aristotle put it, “a great city should not be confounded with a populous one.”
Which I guess makes Pittsburgh akin to the æther.  In modern times it all seems to flow though.  The Economist has had a positive beat on Pittsburgh before most anyone else from the outside word. Remember:  How Now Brown Town?


Blogger Infinonymous said...

If you ignore the poorly administered subsidization that accompanies every city development project larger than a lemonade stand, disregard the insularity, forget the insolvency, and avert your eyes with respect to the civic and elected leadership vacuum, Pittsburgh is a feel-good story! (The recent cinematic equivalent might be "How Do You Know?")

Why the Economist's reporters and editors ignore the context is a puzzler, unless these reports are result-driven.

Monday, October 31, 2011 10:46:00 AM  
Anonymous MSL said...

A great city may not be a populous one, but I think at some point, a great city needs to stop losing thousands of people every year.

Monday, October 31, 2011 12:24:00 PM  
Anonymous Anonymous said...

It already has. Domestic migration is no longer negative. We're only losing population now due to natural change, ie more deaths than births, which incidentally, will work to make the city "younger."

Monday, October 31, 2011 1:25:00 PM  
Blogger C. Briem said...

That refers to the MSA.. not the city.

Monday, October 31, 2011 1:27:00 PM  
Anonymous MSL said...

@ Anonymous.

The census estimated the 2009 population of Pittsburgh as about 312,000. The 2010 population was about 306,000. That's a hell of a lot of natural population loss for a single year.

Monday, October 31, 2011 2:29:00 PM  
Anonymous MH said...

It is the place that invented putting fries on top of everything, so maybe not.

Monday, October 31, 2011 3:55:00 PM  
Anonymous Anonymous said...

Ok, ok. You're right. That was the region, not the city. Anyone know what the natural change for the city is? Chris?

Monday, October 31, 2011 4:28:00 PM  
Anonymous Alethiometer said...

There you go with the Murphy cheerleading again. Can't wait til you cut the ribbon on the Murphy Bridge.

BTW, how'd you type that "a" and "e" together?

Tuesday, November 01, 2011 9:22:00 AM  
Blogger C. Briem said...

am just the messenger of Pittsburgh in a major international publication ok...

there will never be any ribbon cutting we all know.

cut and paste. or html æ

Tuesday, November 01, 2011 9:28:00 AM  
Anonymous MSL said...

@ Alethiometer

It's funny that these articles never mention Murphy's plan to turn downtown into a mall or the millions of public dollars he spent to lure Lord & Taylor and Lazarus to the city as part of his master plan to revitalize Pittsburgh.

Tuesday, November 01, 2011 10:09:00 AM  
Anonymous BrianTH said...

Methodology note: you can't use the Census annual estimates and an actual Census count to estimate year over year changes. There is plenty of known error in the estimates methodology, and in fact it is non-random error with respect to actual Census counts.

Generally, the available evidence indicates the City's population losses at least slowed down late in the decade, and may in fact have reversed.

Tuesday, November 01, 2011 2:07:00 PM  
Anonymous MSL said...

Well, the ACS also showed a loss. The 2009 estimates had the population at anywhere from 311,640 to 313,118 depending on if you used the 1, 3, or 5 year estimate versus a 2010 population of 305,704. I'm not sure what other data sources there are for year-to-year population estimates, so I'd be interested to see the available evidence that shows the losses may have reversed.

Tuesday, November 01, 2011 3:09:00 PM  
Anonymous MH said...

Maybe school enrollment shows a positive sign?

Tuesday, November 01, 2011 3:48:00 PM  
Anonymous Anonymous said...

I must say, it often seems like it's Pittsburgh visitors who rave about the place, not actual residents, which kind of runs counter to the old adage: "It's a nice place to live, but I wouldn't want to visit."

Tuesday, November 01, 2011 4:11:00 PM  
Anonymous BrianTH said...

Here is one-year ACS back to 2006 (before that you run into a problem with group quarters not being included):

2006 297061
2007 290918
2008 297187
2009 311640
2010 305759

If you fit a line to that data you get a positive slope of 3812.

Or you could believe the population increased 14,000 from 2008 to 2009 and then dropped 6,000 from 2009 to 2010, but I kinda doubt that.

3-year is:

2006-08 295,988
2007-09 312,042
2008-10 306,668

Again, I kinda doubt it is actually bouncing around like that. For slope I get 5340.

I'd suggest that is the best single source. But another thing I am watching is labor force--obviously labor force doesn't have a constant relationship to population, but it tends to be in a relatively narrow range, and if anything you would think labor force participation would have trended down in 2009 and 2010 thanks to the recession finally hitting in the Pittsburgh area. Anyway, here is labor force for the City:

2001 157129
2002 156710
2003 154430
2004 153101
2005 151743
2006 151139
2007 151642
2008 153996
2009 153136
2010 153014

It tells a fairly similar story to the ACS data, as it turns out. If you look just at 2006-2010, you get a slope of 524. And so far 2011 is trending up quite a bit from 2010--I'd suggest again that in 2009 and 2010 you may be looking at a reduction in the participation rate layered on top of underlying population growth, with participation snapping back in 2011.

Tuesday, November 01, 2011 10:59:00 PM  
Anonymous MH said...

Or you could look at the -8.6% change from 2000 to 2010 and wonder if the ACS maybe comes with 95% confidence interval and, if so, how big it is.

Tuesday, November 01, 2011 11:40:00 PM  
Anonymous BrianTH said...

It is important to keep in mind there are many possible paths between two points ten years apart.

In this case, the year to year data available suggests the ten-year loss was mostly or entirely caused by large losses in the first part of the decade, which were more than sufficient to outweigh any flat or gaining periods in the second part of the decade.

Consider, for example, the labor force data. You've got a loss of 4115 between 2001 and 2010, but you get there with a loss of 5990 between 2001 and 2006, then a gain of 1875 between 2006 and 2010.

This lop-sided U pattern is also suggested by the ACS data, although you have to do the work of backing out group quarters 2006 and later to show that. It is also suggested by the annual Census estimates, once you apply the known nature of the error in those estimates.

Wednesday, November 02, 2011 7:00:00 AM  
Blogger C. Briem said...

The census folks make it as clear as they can that the ACS should not be used for count data.. especially for trends like that. The estimates data are what they say should be referenced for intercesnal population trends. and even for those you have to be careful with Pittsburgh. They made some adjustments over last couple of years because they were not fully capturing the group quarters population here that will skew any interpretation of the trend. Note also the Western Penn reopened which is captured in the numbers I believe and is a nontrivial number for the city. So for that in itself you have to be awfully careful about extrapolating. On top of all of that, the ACS data skipped from a census 2000 baseline to a census 2010 baseline with the 2010 ACS. So one really really really shouldn't overinterpret any differences in 2009 vs. 2010 ACS data.

I won't restart the argument more than I have to, but for the city proper all or I can't believe anyone would dispute most, of the gain in population among younger adults is coming from higher enrollment at local post-secondary institutions. Whether even that is sustainable will have to be seen. If nothing else, I sure hope the incarcerated population is not an upward trend that will continue.

Wednesday, November 02, 2011 7:09:00 AM  
Anonymous BrianTH said...

To be precise, the official Census estimates are what must be used for various legal purposes. They are also quite clear that certain sorts of error can occur with their estimates for subcounty units, which is part of why I think it is useful to look at the ACS as well when it comes to subcounty units, and labor force data, and any other independent source of data. Of course, understanding the possible errors with any given data source and correcting for those errors to the extent possible is important, but that doesn't mean you should simply throw out potentially useful data.

And in fact once you do all that, there is a consistent picture that emerges: more rapid losses at the beginning of the decade, followed by either much less rapid losses or actual gains at the end of the decade. And I'm honestly not quite sure why so many people seem to think that is implausible.

As for the young adult increase in the City--more students is definitely not "all of" that increase. Whether more students even account for a majority of that increase is harder to say, but the reported increases in student enrollment combined with the observed age distribution of the student population versus the age distribution of the increases in young adult population suggest more students may well account for substantially less than a majority. That's where we left the issue as I recall--serious attempts to estimate the number of new students living in the City weren't adding up to anything close to a majority of the observed young adult increase. Finally, I might also note that the labor force increases seem to undermine the "it is all just students" hypothesis as well.

That's a review--I might take a look at the ACS data and see what it says.

Wednesday, November 02, 2011 9:14:00 AM  
Anonymous MH said...

And I'm honestly not quite sure why so many people seem to think that is implausible.

Because of the trend since 1950. Comment boxes are usually not very good places to display a graph, but by compressing the horizontal axis I can reproduce the graph of Pittsburgh's population from 1950 to 2010:


Wednesday, November 02, 2011 9:24:00 AM  
Anonymous BrianTH said...

OK, I still can't get the age distributions I want. But the increase in 20-29 was 12,511. The increase in college and grad student population 18-34 was 10,835 (this is 2010 ACS minus 2000 Census Long Form, and it roughly agrees with our prior enrollment increase estimates). So we are short already, and then you also have to take out the student increases in 18-19 and 30-34, which are not trivial chunks of the student distribution. It looks to me like it still going to end up right around half of the 12,511, whether above or below I don't know.

I might note I reviewed the old thread, and I believe Chris was suggesting he could come up with numbers for non-four-year professional schools--I'd love to see those if that is is true.

Wednesday, November 02, 2011 10:02:00 AM  
Anonymous BrianTH said...


But there are specific reasons for that history. Lots of roughly comparable U.S. cities lost population from 1950 to about 1980, through a combination of well-studied factors (suburbanization, white flight, lowering people per household, and so on). A bunch of those cities started gaining population in the 1980s or 1990s, but Pittsburgh did not, thanks to the steel bust. The steel bust had predictable long-term population effects, but it is also predictable that those effects would gradually fade out.

So many informed people have always been predicting that at some point the City's population would likely level out and start growing again. Of course most people expected that point to come later than now, but you don't need to look around long for an explanation of why that point might have moved up--just look at all of Chris's posts about how long the Pittsburgh region has been outperforming the nation in terms of its labor market.

And that's the story I see in the data: the already-expected turnaround point came a bit sooner than expected thanks to some rather dramatic macroeconomic events. I just don't see why that is an implausible story, such that people would close their eyes to what the data is telling us.

Wednesday, November 02, 2011 10:10:00 AM  
Anonymous Anonymous said...

Shorter BrianTH:

If we ignore all the data that contradicts my view, and look only at the data which supports it, you can clearly see that the population is surely rising.

Wednesday, November 02, 2011 10:19:00 AM  
Blogger C. Briem said...

If you want a more recent factoid.. and if you look at our neighborhood profiles based on ACS data.. more than a third of all city neighborhoods (32 of 90) showed literally zero new residents who moved in from out of state in the previous year. Probably another third have virtually inconsequential numbers of new migrants. Hard for there to be enough net migration into the city in the remaining neighborhoods to really offset what appears to be virtually zero new migrants in most city neighborhoods... and again what you see in the rest have a lot of college matriculants.. who count and are important, but you have to think about differently in demographic projections. they come and they go, which is normal.

enrollment data by institution is in IPEDS.. I just have not had time to poke at it beyond that report last year.

Wednesday, November 02, 2011 10:24:00 AM  
Anonymous MSL said...

Keep in mind that many grad students in the 25+ demographic are married or have pretty serious partners, so they are adding two people to the city's population. That also needs to be counted when looking at the effect of rising enrollment on the 20-29 range.

Wednesday, November 02, 2011 10:35:00 AM  
Anonymous BrianTH said...


What data am I ignoring?

Chris says:

"Hard for there to be enough net migration into the city in the remaining neighborhoods to really offset what appears to be virtually zero new migrants in most city neighborhoods."

Why is that hard? In fact that's how it has worked in a lot of other cities too--at any given point there are only a few "hot" neighborhoods where most of the net growth is occurring.

I do understand the point about students, but I would emphasize that not only do students count directly, they also spend money and therefore attract investment, and a few of them will end up sticking around. And unless we have reason to believe the percentage that sticks around is changing for the worse (I'd suggest if anything the opposite is likely true), then more students means more of them eventually sticking around. So they are a source of long-term growth, even if they are not as powerful a source of long-term growth as a young couple about to have kids.


Indeed, and it would be nice to put some numbers on that. But I would further note two things. First, such students may be a bit less likely to be living in the City in the first place, particularly if they do in fact have kids already. Second, I'd hypothesize such students are a bit more likely to stick around. In fact, as I noted last time, I have personal experience on both those counts (as a married grad student I was living just outside the City borders, and I eventually made a career change in part so that we could stay here where my wife had an excellent job).

Wednesday, November 02, 2011 11:31:00 AM  
Anonymous MH said...

The old UFO investigator's slogan of "extraordinary claims require extraordinary proof" seems applicable here. Not that Pittsburgh increasing in population is quite at the UFO level of extraordinary, but it is not expected. When you have a 50 year trend, your most reliable evidence (i.e. the decennial census) shows that trend is continuing, and the evidence for a reversal comes only from data that are known to be less than reliable, I’d think caution is in order.

Wednesday, November 02, 2011 12:20:00 PM  
Anonymous BrianTH said...


Again, a slowing down or turnaround was in fact expected--it was a matter of when. Moreover, when an area has a notably better labor market than much of the nation for an extended period of time in the midst of a very bad economic period, you would expect it to attract more job-seeking migrants and/or lose less job-seeking migrants than you would otherwise have expected. So I really don't see the violation of expectations here.

As for the Census--I fully agree it is the most reliable evidence on the population count at those points in time, but it cannot tell you the path that was taken between those points. I think people are simply assuming a straight line is the most likely path, but that is in fact an assumption, not something you are actually getting from the Census data itself. And when you look at all the other data available in this case, that assumption appears unwarranted.

So in that sense, there really isn't any data on the other side. There is just an assumption, which doesn't match the actual data, and we have a very clear and well-supported explanation as to why that assumption is not valid in this particular case.

Wednesday, November 02, 2011 1:40:00 PM  

Post a Comment

Links to this post:

Create a Link

<< Home