Calculated Risk: New Resource for Tracking Home Sales

Calculated Risk: New Resource for Tracking Home Sales

Housing starts are a non starter

The July 2011 housing starts at around 600,000 suggest that the US housing market is still in critical care. Down from the highs of 2.2 million annualised starts in 2066, the new housing market in the US continues to struggle at annualised 600,000 starts.

The foreclosed homes in the US have dented the housing market big time. With a lackluster economy that refuses to generate jobs for the millions who are now unemployed, the morale of the American consumer is low, thus keeping the housing markets down.

There were some celebrations by speculators since the starts were down by only 1.5% over the last month.  Thus a less than expected decline in housing starts prove to be a good news!

The reality remains as unemployment rate remains consistently high over the years, the housing market continues to perform significantly lower than its historical highs.

“A bloated inventory of unsold homes and a weak economy are weighing down on the housing market, whose collapse was the main catalyst of the 2007-2009 recession. A large foreclosure pipeline also is not helping, leaving builders with little incentive to break ground on new projects.”

London riots and Toronto’s suburbs

As violence spread through the immigrant dominated, low income, suburban neighbourhoods in UK, it has raised concerns about a similar violent breakdown in low-income neighbourhoods in Canada.

A recent report by Prof. David Hulchanski of the University of Toronto has pointed out the growing disparities between low- and high-income households in Toronto. In this post I argue that while income disparities have increased in central parts of Toronto, concerns about violent breakdown are largely misplaced for most Canadian cities, including Toronto.

In an earlier post I have argued that Professor’s Hulchanski’s results were influenced by the choice of spatial limits used to define ‘Toronto’. It is true that income disparities have worsened in Metropolitan Toronto, which is home to 2.5 million people and covers an area of 250 square miles. However, the increase in income disparities has resulted from outward migration of Toronto’s middle class to suburban municipalities that constitute the greater Toronto area (GTA).

For details see my earlier postings below:

The good, the bad, and the ugly in Toronto

Where is Toronto’s missing middle class? It has suburbanized out of Toronto

A quick review of the 2006 Census data suggest that neighbourhoods in Toronto’s outer suburbs (municipalities that constitute the GTA along with with Metropolitan Toronto) are more egalitarian than the ones in Metropolitan Toronto. The outer suburbs are now the choice digs for Toronto’s middle class, which is lured by abundant supply of affordable housing and other amenities required by households with children. A large number of recent immigrants have also flocked to the suburban parts of the GTA in pursuit of affordable shelter.

One way of comparing Toronto to its neighbouring suburbs is to deploy the same income typology as was used by Professor Hulchanski to categorise neighbourhoods in Toronto. He categorised neighbourhoods (Census Tracts) as follows:

Income quintiles	Neighbourhoods	Percent
Less than 40% average CMA income	134	13.36
Between 20% and 40% below average CMA income	276	27.52
Between 20% above and 20% below average CMA income	398	39.68
Between 20% and 40% above average CMA income	75	7.48
More than 40% above average CMA income	120	11.96
Total	1,003	100

 

CMA in the above table stands for the Census Metropolitan Area that comprises Toronto and most of its suburban municipalities. Toronto CMA has  a population of about 5 million. The above table suggests that in 2006 over 40% neighbourhoods in Toronto region reported average household income below 20% of the regional average.

I produce below tabulations between immigrant population and income quintiles. To account for the impact of recent immigrant concentration on the socio-economics of a neighbourhood, I categorise neighbourhoods based on what percentage of immigrants within a neighbourhood arrived in Canada between 1995 and 2006. Furthermore, I produce two tabulations, one for Metropolitan Toronto, and one for all other municipalities that are part of the Toronto CMA.

The tabulation for Metropolitan Toronto is presented below. Similar to what Professor Hulchanski found, I report that 46% neighbourhoods falling under the highest recent immigrant concentration (category 4) belonged to lowest income category. In comparison, only 4% of the neighbourhoods that were categorised as having the lowest concentration of recent immigrants belonged to the lowest income category. Furthermore, the highest income neighbourhoods were predominantly the ones with lowest concentration of recent immigrants.

However, the situation is quite reversed when the same tabulation is generated for outer suburbs in the Toronto region. Consider the following table where 48% neighbourhoods with the highest recent immigrant concentration corresponded to middle or higher than middle income categories. Only 8% of the highest recent immigrant concentration neighbourhoods fell in the lowest income category in suburban Toronto.

Furthermore, across all concentrations of recent immigrants, i.e., from lowest to the highest, the most frequent income category was the mid income range, i.e., between 20% above and 20% below average CMA income.

In summary, a quick review of Census data from 2006 reveals that Toronto’s outer suburbs are home to the middle class where recent immigrants are enjoying the benefits of affordable housing and other amenities that are not available in Metropolitan Toronto. In addition, Metropolitan Toronto continues to experience higher income disparities that adversely affects recent immigrants at an increasing rate.

Are students' teaching evaluations influenced by instructors' looks?

Are students' teaching evaluations influenced by instructors' looks? ggplot2 may help find the answer.

The recent release of RcmdrPlugin.KMggplot2 has made ggplot2 available to those who prefer GUI to the command line interface. With the new plugin for Rcmdr, one can simply point and click and within seconds can have an amazing publish-quality graph.

I test drove the plugin and liked it very much accept the glaring omission: it does not provide for faceting for scatter plots, which is a key strength of ggplot2. Let me illustrate this with an example.

I use the Teaching Ratings data from the AER package. The data set was used in a study to determine if students were influenced by the instructor's good looks or otherwise while evaluating the instructor's teaching performance.

Using Rcmdr and the ggplot2 plug-in I quickly draw the graph, which is presented to the right. The blue triangles represent women and the red dots represent male instructors. The Y-axis is a normalized measure of instructor's looks and teaching evaluations are presented on the x-axis.

It appears from the scatter plot that  good looks are correlated with higher teaching evaluations.

Now let's add two more dimensions to account for the following. First, is there a difference in teaching evaluations for tenured instructors versus the non-tenured instructors. And second, do students evaluate instructors belonging to visible minorities groups any different from the rest.

Using faceting in ggplot2, I draw the new graph that is presented to the right. Notice that the graph has four quadrants capturing the 2 x 2 possible combination for tenured (yes/no) and minority group (yes/no).

Let's focus on the top right quadrant, which shows that non-tenured, visible minority faculty members happen to be exclusively males receiving low beauty evaluations, but higher teaching evaluations.

The bottom right quadrant however shows that most tenured visible minority instructors are females who receive a higher teaching and beauty evaluation.

The second graph was produced by tweaking the command generated by the plug-in to illustrate that with faceting one can quickly plot multi-dimensional graphs. I hope that the next version of the plug-in  incorporates faceting for scatter plots.

Below is the code used to generate the second graph.

data(TeachingRatings, package="AER")
tr<-TeachingRatings

.df <- data.frame(x = tr$eval,y = tr$beauty,z = tr$gender, z1=tr$tenure, z2=tr$minority)

.plot1 <- ggplot(data = .df, aes(x = x, y = y, colour = z, shape = z)) + geom_point() +
  scale_colour_brewer(palette="Set1") +facet_wrap(z1~z2) + xlab("eval") + ylab("beauty") + labs(colour = "gender", shape = "gender") + kmg2_theme_gray(10, "sans") + opts(legend.position =
  "right")
print(.plot1)
rm(.df, .plot1)

Why geography challenges Americans?

As citizens of a country that ends up in armed conflicts all over the world, Americans should know a thing or two about geography. Recent polls over several years however have shown that the American youth cannot locate even those countries on a map where the US has been involved in an armed conflict. A May 2006 poll revealed the following:

Take Iraq, for example. Despite nearly constant news coverage since the war there began in 2003, 63 percent of Americans aged 18 to 24 failed to correctly locate the country on a map of the Middle East. Seventy percent could not find Iran or Israel.

Nine in ten couldn't find Afghanistan on a map of Asia.

And 54 percent were unaware that Sudan is a country in Africa.

Remember the December 2004 tsunami and the widespread images of devastation in Indonesia?

Three-quarters of respondents failed to find that country on a map. And three-quarters were unaware that a majority of Indonesia's population is Muslim, making it the largest Muslim country in the world.

And then there is Miss South Carolina in Miss Teen USA 2007 pageant with her illustrating answer about the causes of such ignorance amongst the “US Americans”. Watch the 49 second clip below and you will see how deep-rooted the problem is as she explores the reasons behind why 20% of the Americans couldn’t locate  their own country on a map.

 

Causes of geographic illiteracy amongst “US Americans”

Winehouse, Cobain, Hendrix: What is it about age 27? - The Globe and Mail

Image by NRK P3 via Flickr
The Globe and Mail wonders why some performers have died at the young age of 27. While there may not be anything peculiar or special about some performers dying at age 27, there is however enough statistical evidence to suggest that pop stars have a higher propensity to die early than the general population.

Mark Bellis and others studied the mortality rates of 1,064 British and North American pop stars and concluded that two years after becoming famous, pop stars experienced a two to three times higher risk of mortality than those in the general population who shared similar demographic traits.

Writing in the Journal of Epidemiol Community Health, the authors noted that drug and alcohol were associated with most deaths followed by cancer, accidents, and other cardio-vascular ailments.

Winehouse, Cobain, Hendrix: What is it about age 27? - The Globe and Mail

Forklift mishap destroys $1m of shiraz wine

The birth of containerized logistics is partly owed to the fact that shipments were damaged during transport. However, there were certain shipments that had a higher incidence of damage than the rest. For instance, alcohol often ended up lost when bottles broke during shipments. No one really new if the ship's crew helped themselves to some fine while being on sea for weeks. Containerization put an end to this until a forklift destroyed $1m of fine wine.

BBC News - Australia: Forklift mishap destroys $1m of shiraz wine