Foreclosure Trends in Dane County

Introduction

            I was hired by a research group wanting analysis of foreclosure patterns in Dane County, Wisconsin. The timeframe I’d be looking at is short, only the years 2011 and 2012. This is not very much data to work with and it is hard to see trends in so few data points. With that said, the county sees a major problem beginning, an increase in foreclosures throughout the county. The research group is only looking for analysis, not a reason for this perceived trend.

            I looked at a number of factors in establishing a trend. These factors included the change in numbers of foreclosures per census tract and their distribution geographically. Finding a trend statistically or geographically may help county officials understand what is happening and use it to find the cause.

Methodology

            To analyze the data given, I used a number of statistical tools as well as mapping tools to look at the issue geographically. First, I was given the data for foreclosures in Dane County by Census Tract from the independent research group. While that data that was provided include many fields not related to the issue, the fields that were used were:

-       COUNT2011

-       COUNT2012

-       NAMELSAD10

I also created a field to show the change from the 2011 count to the 2012 count. It is labeled:

-       COUNT_CHAN

By creating this field, I can see clearly where some tracks increased and some decreased. This then was used to make Figure 4, the map to help visually represent the distribution of foreclosure rates.

The data in these fields offered all of the information needed to analyze the issue at hand. The statistical tools used were:

-       Z-Score

-       Probability

Z-Score is the value given to a data point to show its relation to the mean. Z-Scores shows how many standard deviations the data point is from the mean. It can be positive or negative.

Probability is the likelihood of an event happening. It is found by comparing a specific event to all events possible. The higher the probability the more likely it is to occur.

Results

Based upon the tools used to answer the research question, a trend does exist. First, by looking at the Z-Scores of particular census tracts shown in Figure 3, we see that in tracts 108 and 25, both lower from 2011 to 2012. Tract 108 drops closer to the mean in 2012, as 25 drops even farther from the mean in the same time frame. It is important to also acknowledge that the mean changes from 11.39 foreclosures to 12.29 and the standard deviation grows as well. Even with these changes, it appears that foreclosures are decreasing.

                                            Figure 1: Z-Score Calculations

                                Figure 2: Probability Calculations

Census Tract	2011 Z-Scores	2012 Z-Scores
108	2.0063	1.4840
25	-0.6144	-0.9387
120.01	1.7784	2.9982

 Figure 3: Z-Scores by Tract and Year

In contrast, what may have county officials sounding the alarm is the growth of the mean and the drastic jump in tract 120.01. 120.01 in 2012 is clearly an outlier as the Z-Score indicates that is nearly 3 standard deviations from the mean. This shows us that 120.01 is on the extreme high end of the scale with almost all of the other data points occurring below it.  

When bringing probability into the analysis, we see calculations for the entire county. To find out the number of foreclosures that occur 80% of the time, I used the probability chart to find a Z-Score of -0.84. Plugging that into the Z-Score equation, I found that 80% of the time, each census tract will exceed 3.9778 foreclosures. This figure should not be surprising as to the size of Dane County and its population. It is unlikely that the county will have less than that. Additionally, I found the number of foreclosures that are likely to be exceeded 10% of the time. Again using the probability chart and equation, I found that 10% of the time, each census tract will exceed 24.9792 foreclosures.

Figure 4: Foreclosures Changes by Census Tract, 2011-2012

When looking at the issue geographically, Figure 4 shows the changes in foreclosure numbers by census tracts. The blue tracts are ones that had fewer foreclosures in 2012 than in 2011 and the red tracts had more. Just by viewing Figure 4, you can see that the tracts that were the worst are primarily located on the outside of the county in comparison to the middle, around the city of Madison.

Conclusions

The results found during the analysis of the issue showed a geographical trend that the outsides of Dane County were experiencing higher rates of foreclosure, but the center around Madison did not vary by much. Additionally, when examining specific census tracts, it showed a lowering of foreclosure rates. While this sets a trend contrary to the one found by county officials, I also found indicators that could have led them there. For example, census tract 120.01 had a Z-Score way above the mean. While this is alarming, it is just one area. Similarly, the mean of foreclosures per census tract went up. These two figures alone would worry someone who had not looked at the statistical data.

            The implications of my findings can help county officials worry less about the housing market in Dane County. While foreclosures were up overall, and that this is an indicator of how the housing market is doing, it fails to take in other factors that play a role in determining foreclosures. Factors like location, median household income, and neighborhood crime rates are all involved with a foreclosure. A solution to some of these findings would be to locate areas of concern, like the dark red census tracts in figure 4, and localize efforts there. Additionally, further research can be done to see why foreclosure rates drop in some areas and not others. What is happening in those tracts that can be implemented in tracts that are struggling. Overall, the first step was to analyze the trends both statistically and geographically and use that information to proceed.