Saturday, January 23, 2016

Atlanta Home Prices

My previous post on Atlanta housing shows rental prices by neighborhood and their five-year trend. This post uses home price data from Zillow to look at an eighteen-year trend. I'll also explore the divide between southwest and northeast Atlanta and racial income gaps.

First, a map of home prices in Atlanta, as of November 2015. This map was created using price per square foot data from Zillow and multiplying to estimate the average price for a 2000 square foot home. Blank neighborhoods do not have data in the Zillow dataset.

As expected, the map looks similar to the apartment price map- neighborhoods with expensive homes tend to have expensive apartments. The graph below demonstrates this by graphing home prices against apartment prices. 

This graph is helpful for understanding the data set. Neighborhoods above the trend line, like Midtown and Downtown, have high rental prices relative to home prices. This could be because rental stock that is nicer than home stock, or because many homes in these neighborhoods are condominiums, and are less expensive on a price per square foot basis because they have condo fees and less land per resident than a traditional home.

The home price map also shows a striking divide between Northeast and Southwest Atlanta. The graph below investigates this by graphing the home price for the two halves of Atlanta. All prices are adjusted to 2015 prices using the CPI.

The Southwest rose by more on a percentage basis from 1998 to 2006- 62% compared to the Northeast's 38%. But home values in the Southwest then fell by 59% during and after the recession, while homes in the Northeast only fell by 31%. The Northeast homes are now only 8% below their pre-recession high, while Southwest homes are 57% below their pre-recession high.

The larger price swings in Southwest Atlanta were caused at least in part by federal programs. Both Presidents Clinton and Bush led programs designed to increase home ownership among low to middle income households, and Bush specifically targeted his program towards minorities. These programs increased prices in low-income neighborhoods by creating increased demand, and the lax lending rules put the neighborhoods at a greater risk during a recession. The less educated were also more likely to lose their jobs during the recession, compounding the problem.

The unintended consequences of the home-ownership programs is similar to the problems with affordable housing programs described in my previous post- they are well-intentioned programs designed to help the poor and middle class, but end up making the average person worse-off because they distort the market.

Back to Atlanta: The video below shows how each individual neighborhood changed from 1998 to 2015.

For those interested: click here for the interactive graph version of the above video. Or click here for the video (or graph) that shows growth to 2015 over time (instead of growth from 1998), which I think is less intuitive but more relevant.

Comparing Northeast and Southwest Atlanta raises the question of race- the Northeast half is mostly white, and the Southwest half is mostly black. The graph below shows median household income for blacks and whites in the city of Atlanta, as well as the Metro area and the country.

The income gap for the city of Atlanta is depressing. Blacks in Atlanta have noticeably lower incomes than the national average for blacks, while whites have higher incomes than the national average. This situation is unique to the city proper- in Atlanta's metropolitan statistical area, blacks have a higher income than the national average.

My last post led to interesting discussion on Atlanta Reddit and Curbed Atlanta. I also received an e-mail from Jarod Apperson on the relationship between gentrification and luxury construction:

Related to this topic, one thing I think could use more clarity is the relationship between luxury construction and gentrification.  If gentrification is defined as an area's mean income, rents, etc. rising from wealthier residents moving in, that can happen either through displacement, new construction, or some mix of the two.  Often displacement is what folks get most worried about.  It is concerning to think that people who have lived in a community for some time are not able to continue doing so because of changes happening around them.  In lamenting that new construction is expensive, the articles you point to seem to be misinterpreting the reality of new (luxury!) construction's relationship with displacment, which you hint at in point 4 on your list of takeaways.  
Michell Eloy at WABE: "But low- and middle-income renters – renters like Huftalen – say they feel increasingly squeezed by the market, unable to afford the new luxury apartments and edged out by stiff competition and equally high prices for older units."
The point Eloy fails to make is that were it not for new luxury apartments, the high prices for older units would have risen even more (more on that in a second).  
Editorial Board at CL: "But left unchecked, this environment could create a frightening affordability crisis where, years down the line, Atlanta is only a place for the well to do. That's how affordability crunches happen — developments get rolled out, the city's population grows — and boom — you're suddenly pushed out by property taxes and house-hunting in Riverdale."
No, that's not how affordability crunches happen.  New residential developments rolling out, do not make housing less affordable.  In fact, they do the opposite.  Rising residential demand from any number of factors (preference for less time in traffic, proximity to new amenities like the Beltline, proximity to new restaurants/stores etc.), leads to affordability crunches.  Meanwhile as new housing developments get rolled out, they ameliorate such crunches.
Rather than lamenting luxury construction, these writers should be championing luxury construction as a deterrent from the displacement they are so worried about.  The reason is this: by decreasing demand for older units, new luxury construction lowers rents for the existing housing stock.  The theory is simple.  When a new building opens, it attracts some residents from older places nearby and because there is less demand for their units, those places charge rents that are lower than they would be in the absence of this new construction (note that this doesn't mean rents necessarily go down at older places, they might go down sometimes, but they certianly end up lower than they would in the absence of new residential construction).
So, given this wrinkle, what really matters is changes in the distribution of rents, not changes in the mean rent.  It is entirely possible for mean rents to rise substantially without any displacement occurring as long as housing options aren't lost on the left side of the distribution.  
Indeed there is evidence of this phenomenon is occurring here in Atlanta.  A recent BisNow article reports, "While rents shot up another 10 cents/SF on average to $1.76/SF...same-unit rents actually slipped nearly 4% year-over-year."  So what is happening is this: new developments are opening at above-average rents (luxury units, raising the mean), and attracting at least some tenants who would have otherwise chosen nearby existing multifamily units.  As a result, those existing multifamily units are getting cheaper, lowering their prices to stay full.  In other words, not only is luxury construction not leading to displacement, it is actually preventing displacement.  It is keeping our city affordable.
It would be a shame to see a plan like this one derail luxury construction's role in keeping Atlanta affordable.  I worry that a misunderstanding of housing markets on the part of both the press and politicians may end up leading Atlanta toward less affordable housing, despite hopes for more.
Update (1/24/16): commenters on reddit point out that Atlanta had high levels of mortgage fraud before the recession, which inflated home values. This happened in both low-income and high-income neighborhoods. Lax lending rules intended to increase home ownership rates were partially responsible for this problem too.

Tuesday, January 12, 2016

Atlanta Rental Prices: A Tale of Two Cities

Neighborhoods in Old Fourth Ward have gone from abandoned to beautiful in a few short years. A beautiful park, apartments, and Ponce City Market exist in a space formerly occupied by warehouses, parking lots, and abandoned buildings. As this space and others have improved, the rents have rose, and several recent articles voice this concern.

My neighborhood in Kirkwood is great and far cheaper than where my sisters live in New York and DC, so I wanted to better understand calls for city hall to take action on affordable rents. I downloaded data on average apartment prices by neighborhood from Zillow to make this visual:

Mouse over the map to get price per square foot for each neighborhood, and the five year price change, adjusted for inflation (CPI). Blank neighborhoods are missing in the Zillow data.

Its true that there are a couple neighborhoods in Buckhead with median rent over $2 per square foot, and Midtown is up to $1.85 per square foot. But half the city can still be rented for less than $1 per square foot. My own neighborhood has a median rental price of $1.19 per square foot.

To better show change over time, the visual below shows current price graphed against the five year percentage change in prices.

The visual is striking. Although a price divide existed between the Southwest and Northeast halves of the city five years ago, almost every neighborhood in the Southwest half of the city has gotten less expensive, and every neighborhood in the Northeast half has gotten more expensive.

I have several take-aways from these graphs.

1. When people complain that the city is getting too expensive, they only mean the most desirable neighborhoods where they want to live. Half the city is very cheap.

2. Price is a good measure of desirable neighborhoods. The divide between nice neighborhoods and bad neighborhoods in Atlanta is growing.

3. Plans for affordable housing should be very careful. Most affordable housing plans actually make housing more expensive. Consider a policy that requires XX% of affordable housing per new development. Less total developments will be built because developments are now less profitable. Then, in the developments that are built, less units will be on the open market. Prices are then higher due to the policy because the housing supply is smaller. Good for the lucky few who win the affordable housing lottery, but bad for everyone else.

4. Atlanta should instead be very generous in encouraging and approving development. Increased housing supply will help keep down prices and allow more people to live where they desire.

5. Increased property tax from development can then help our poor neighborhoods improve. Luxury buildings, scary to some, generate property taxes that can be spent on improving schools, safety, and transit, or on more targeted development plans. Atlanta needs to increase investment in poor neighborhoods as property taxes rise. Luxury buildings also have residents who contribute to the economy and benefit local low-wage earners.

I'll publish another post on home price data from Zillow later this week. There are some differences from the apartment data, and more years. For a notification, use the google plus or e-mail widgets on the right, or follow me on twitter

Wednesday, January 6, 2016

Using Data for My NBA All Star Vote

Starters in the NBA All Star game are decided by fan voting, which is open until January 18. I take my voting responsibility very seriously, like a presidential election or new potato chip flavors. For the All Star game I downloaded 2015-2016 data for player efficiency rating (PER) and real plus minus (RPM). I like using these two stats because they have fundamentally different sources- PER is a clever aggregation of box score statistics, while RPM measures how the score changes while a player is on the floor, controlling for the other players on the floor. The first graph shows the Western conference.

Clearly my vote is Kobe x 5. (Kidding) The first four starters are clear- Curry, Westbrook, Durant, and Kawhi. This leaves one more front court player. I choose Draymond- if you draw a 45 degree line perpendicular to the trend, he's ahead of Anthony Davis, Blake, or Cousins. He also plays for a better team, which I think is fair to consider. But even more convincing to me, I think Real Plus Minus is a better stat than PER. (More on this below.)

Next, consider the Eastern Conference:

The top players are less talented in the East. Lebron is an outlier, even though his stats put him below Kawhi in the West. Milsap and Lowry are next in the East, and Bosh is slightly ahead of Drummond for the third forward spot, especially given my preference for RPM. Jimmy Butler is the only guard in the next tier of players (Butler, Monroe, Gasol, George, Love), so he gets the fifth spot.

A few other observations as I was putting this together.

1. Real Plus Minus is getting really good. RPM has more promise as a stat because it captures the effect of things that don't show up in the box score- setting screens, boxing out, and most aspects of defense. But early estimates weren't very good- it's difficult to estimate because some players are more likely to play at the same time as other certain players and it is hard to disentangle their effects (multicolinearity). But recent models use more advanced methods to control for this. (These models can be difficult to estimate, and its a really cool service for ESPN to provide them.) Anyway, the results do much better against the eye test than they used to, and I found this graph very convincing:

This graph once again shows PER vs RPM, but is now color-coded by defensive RPM. PER is much better at measuring offense than defense. Notice that almost any case where a player has a better PER than RPM (they're on the upper left half of the graph), they are bad at defense. RPM is capturing this, but PER is not.

2. Using a filter to a specific team is interesting. Look at the Warriors, Spurs, or Thunder to see interesting graphs. Or see the Cavs example below. Lebron and Love are doing most of the work, with positive contributions from Tristan and Delly. Mo and Jefferson were not great acquisitions. Kyrie (RPM = 3.4 last year) and Shump's (2.3) returns will help, especially when they take minutes from Mo and Jefferson.

3. The Sixers, Nets, and Lakers are interesting, for other reasons. For the Sixers, Stauskas has the second worst RPM in the East, and the third worse PER (Players with at least 500 minutes). Also on the Sixers, Okafor has the the worst RPM in the league by a large margin, even though he has an above-average PER. I interpret this as, "he puts up good stats, but everything else he does is terrible."

4. Unrelated, but I came across this while looking up the other data: the Spurs are on track to have the highest margin of victory of all time. It probably won't last, but I was surprised by the magnitude.

Friday, January 1, 2016

Murder Rates: Depends on where You Measure

When I hear about city murder rates, I always wonder whether that refers to the city proper, or the greater metro-area. City murder rates are usually higher because suburbs tend to be safer. But city borders can be arbitrary- when considering two similar metro-areas, one city could have larger borders, which then makes it appear safer because it includes more suburban-type neighborhoods. To help clear this up, I used FBI data to graph the city murder rate versus the murder rate for the entire metropolitan statistical area (MSA):

New Orleans is the clear outlier. Detroit, Baltimore, and St. Louis have similarly high murder rates in the city, but New Orleans has a uniquely high murder rate for the MSA. The murder rate for the greater metro area of New Orleans is about as high as the city of Memphis or Atlanta (unlabeled).

The trend line shows the average relationship between the city rate and the MSA rate. Virgina Beach is noticeably above the trend line, but that is because the Virginia Beach MSA is a combination of multiple cities, including the less-safe Norfolk and Newport News, unlike most other MSAs which tend to be a single large city and surrounding suburbs. 

In each case, the largest city by population in the MSA is graphed, and only MSAs with at least a million residents are included.

The same graph is reproduced below as an interactive, which provides hover text for unlabeled cities:

The relationship between city murder rate and MSA murder rate is fairly strong (R square = 0.57). But this is at least partly driven by the inclusion of city rates in the MSA rate. The graph below shows city rate and the non-city rate- the MSA rate minus the murders and population from the city.

Notice there is almost no relationship between city murder rate and suburb murder rate (R Square = 0.02).

Cities lowest on the graph and to the left have the safest suburbs relative to the city. Milwaukee's murder rate is the most segregated along city borders- a person in the city is 28 times more likely to be murdered than a person in the metropolitan area outside of the city.

*Some MSAs did not have all member cities/municipalities report crime data. In these cases, I used the FBI's estimated totals. All cities had reporting for at least 94% of their population, with the exception of Indianapolis (86%).