Introduction

In this project, we conduct a comprehensive analysis of regional disparities in U.S. residential real estate prices. Our investigation is driven by two primary questions:

What sort of regional disparities exist for residential real estate prices across the US?

What factors contribute to high housing prices?

We focus initially on housing prices, with a particular interest in the effect of Covid-19 on state-level housing price percent change. We calculate this percent change by focusing on the shifts seen from 2018 (solidly before the pandemic) to 2023 (the most recent year that had solid data for us to examine).

Through this analysis, we determine two states of particular interest: Idaho and Louisiana. Idaho saw far higher median housing price increases than the national average increase, while Louisiana was the only state in the union to actually experience a decrease in median housing prices.

We bolster this analysis with an examination of interstate migration in the US, under the assumption that more people moving into a state would entail higher housing prices (and vice versa for more people moving out of a state), hoping to identify a potential causal variable for the housing price trends we see.

We are particularly interested in the effect Covid-19 had on interstate migration, as we found many suggestions that the pandemic saw significant population shifting in the US. One source of particular interest for us was this article by the United States Census Bureau, which posited that this migration was caused by people moving to their state of birth, possibly to be closer to their families.

The following report will delve deeply into the above trends and themes, and explore potential explanatory variables for pandemic population shifts.

For this analysis, we integrate two key data sources:

Acquiring Data from Zillow:

Data was dowloaded from Zillow website using the following link: zillow housing data

Numerous entries in the dataset represent national averages; thus, we seperate the dataset into one for national entries and one for state-level entries, to avoid encountering too many NA values as we continue to focus on state-level data.

Zillow Data Inspection, Wrangling, and Visualization

We find that the only missing values in the Zillow dataset pertain to state-level pricing data. Because this is the variable we are interested in examining further, we must determine whether year or state (or an interaction between the two) is the biggest factor in whether an observation is or is not NA. Below is a heatmap exploring this potential NA interaction. For our purposes, an interaction would be an issue for state-level data analysis and comparison, since we wish to determine which states had comparatively significant pricing changes in the context of the whole country, and NA values could throw those calculations off.

Upon analyzing the heatmap, we observe that missing data (NA values) predominantly occurs in earlier years, although certain states exhibit consistently higher rates of NA presence. Notably, 2005 represents a significant reduction in missing values, with another marked improvement occurring after 2012, at which point missing data becomes minimal nationwide. Considering our research objective—“Are there distinct patterns or clusters in housing prices when analyzed across different regions or locations?”—maintaining comprehensive geographic coverage is essential. Thus, we have decided to exclude data before 2012 to maximize completeness while retaining all states in our analysis.

Because our objective is to examine changes from 2018 to 2023, though, this poses no significant issues.

Moving forward, regional disparities will be visualized using maps that illustrate the percentage change in housing prices between December 2018 and December 2023. We will create separate maps representing both average and median state-level changes to effectively capture and interpret trends across different price segments.

Key Takeaways:

Housing prices increased across all states when measured by average prices; however, the median price map indicates at least one state experienced a decline.

States exhibited notably larger median price changes compared to average price changes, indicating that properties with standard (less extreme) prices had greater relative increases during this period.

The median price metric is likely more informative and representative, as it better reflects typical housing market experiences, although including the average prices provides additional context regarding the nature of price changes, especially at the extremes.

To identify specific states for further analysis, we will now explore those with the most significant increases and decreases in median housing prices.

Top 5 States by Median % Increase
rank state pct_change_med
1 ME 81.16630
2 ID 74.98449
3 NH 72.23471
4 MT 68.47445
5 FL 66.97849
Bottom 5 States by Median % Increase
rank state pct_change_med
1 LA -2.569941
2 MS 14.511985
3 ND 17.376260
4 NM 22.242840
5 AK 25.669667

Identifying States for Further Analysis

To pinpoint specific states for deeper analysis, we will identify those that experienced the most significant increases and decreases in median housing prices between 2018 and 2023.

Top 5 States with the Highest Median Price Increases:

  • Maine, Idaho, New Hampshire, Montana, Florida
    These states saw the most substantial growth in median home prices, suggesting strong market demand, possible migration influx, or other economic drivers.

Bottom 5 States with the Lowest Median Price Increases (or Declines):

  • Louisiana, Mississippi, North Dakota, New Mexico, Alaska
    Notably, Louisiana is the only state where median home prices actually declined, making it a particularly interesting case for further investigation.

Moving forward, we examined Idaho and Lousiana more closely, hoping to identify city-level trends. Again, we will use median prices.

Cities with the Largest Declines: DeRidder and Lake Charles, Louisiana

City with the Highest Increase: Mountain Home, Idaho (+95%)

Here, we see no significant divergence from the state-level maps, which trends staying consistent across cities.

ACS Data Wrangling and Investigation

To acquire ACS data (Census data), we had to use an API, identify our variables of interest, and do significant wrangling to get the data in a form ready to use in map creation.

Below, we generate state-level graphs to analyze migration patterns and assess whether pandemic-driven relocation and population shifts can explain the observed changes in median home prices.

Idaho exhibits a significantly higher migration rate compared to the national median, whereas Louisiana falls slightly below it. This is largely in line with out findings for pricing changes, bolstering the viability of our aforementioned hypothesis.

As previously referenced, moving back to one’s birth place could be a viable explanation for this finding. However, we wish to explore other explanations that more traditionally would be used to explain people’s incentives to migrate to another state. In this way, we can explore alternative explanations, and see if the above theory really is the most solid or if it’s just a better sounding story than the truth provided by data.

Below we show a similar map to the one above, this time examining median changes to income at a state level.

In this case, Idaho once again stands out with a median income increase well above the national average, whereas Louisiana shows a less pronounced change. The factors driving Idaho’s income growth warrant further investigation, specifically whether it is primarily due to an influx of higher income individuals relocating to the state or a broader increase in wages attracting new residents. Since we are analyzing median income rather than the mean, the impact of extreme values is minimized, making this distinction particularly important to consider.

The fact that Idaho saw such a pronounced increase in income, while Louisiana’s income decreased, could be a potential explanation for the migration patterns we’ve seen; perhaps migration was incentivized by moving to states with higher median incomes in hopes of making more money. However, we are hazard to make such a claim, since this trend could just as well be explained by people who already had higher incomes moving away from places like Louisiana and into places like Idaho during the pandemic for other reasons, such as being closer to family. This is a distinction that warrants further exploration at a later date.

Another important variable to look at is that of commute time changes, explored in the map below.

Now we see Louisiana actually also had an increase here in median commute time, which seems to be at odds with our findings of fewer people living there. Idaho shows a more expected trend. This variable is flawed, though, in that many people moving did so because of access to remote work, which is not captured by this variable.

Moving forward, we will create city-level maps to further try to parse regional trends. This, too, entails significant wrangling, as well as switching from 1 year to 5 year estimates for ACS data, as city-level data is less available for our cities of interest at the 1 year level.

Visualizations

In Idaho, we see:

In Louisiana, we see: Louisiana presents a more mixed trend: Some cities exhibit strong red dots, signifying a high increase. Other cities show blue and gray dots, suggesting little to no growth or a decrease in median income. The distribution appears more scattered compared to Idaho.

The plot illustrates the City-Level Percent Change in Mobility (2018-2023) for Idaho and Louisiana. The color gradient represents the percentage change in mobility, with red indicating an increase, blue representing a decrease, and white/gray suggesting minimal change.

In Idaho, most cities exhibit a slight increase in mobility, with a few cities showing moderate red shades, indicating a noticeable rise in mobility. One city in the southeastern part of Idaho stands out with a bright red dot, reflecting a significant increase in mobility. Meanwhile, some cities have lighter shades of red or white, indicating smaller or negligible changes.

In Louisiana, mobility trends appear different, with a prevalence of purple and blue dots, indicating a decrease in mobility in several cities. Many locations show light purple or gray dots, signifying only minor changes. A few cities have slightly darker purple dots, suggesting a more pronounced decline in mobility.

Comparing the two states, Idaho generally experiences an increase in mobility, whereas Louisiana exhibits a decline in mobility across most cities. This visualization highlights potential regional differences in commuting patterns and movement trends over time.

In Idaho, we can see:

In Louisiana, the pattern is notably different:

The differences in commute time changes could reflect varying patterns of urbanization, population movement, or shifts in work habits following the pandemic.

Finally, we examined property taxes as a variable to indicate whether a change in taxes affects house prices across states. We found this variable to be worth looking at since there’s often a theoretical relationship between property tax levels and housing affordability. Areas with rapidly increasing housing prices may see corresponding increases in property tax assessments, potentially creating additional financial pressure on homeowners. Our analysis revealed variable patterns across states, with some high-growth housing markets also experiencing significant property tax increases. Interestingly, the relationship was not uniform across all regions, suggesting that local tax policies and assessment practices play an important role in moderating this relationship. In our comparative analysis between Idaho and Louisiana, we noted that Idaho’s property tax increases, while significant, did not appear to dampen the strong housing price growth in that market, potentially indicating that other factors like migration and income growth were more dominant drivers of price changes during this period.

Overall, we can see that the change in property taxes between 2018 vs. 2023 is not significant especially when looking at states like Louisiana and Idaho. This would indicate that there is likely not a strong causal relationship between property tax changes and housing price variations during this period. Despite Idaho experiencing a substantial 75% median housing price increase, and Louisiana seeing a decline of 2.6%, both states showed relatively moderate changes in their property tax rates. This finding suggests that other factors we’ve examined - particularly migration patterns and income growth - played a more significant role in driving regional housing price disparities. The property tax data further supports our conclusion that housing markets during this period were primarily influenced by demographic shifts and economic factors rather than tax policy changes, at least in our key comparison states. This pattern is consistent with post-pandemic housing trends where remote work opportunities and lifestyle preferences appeared to have a stronger impact on housing demand than local tax considerations.

Conclusion

Our analysis provides a detailed exploration of housing price changes across U.S. regions, with a specific focus on the post-pandemic period from 2018 to 2023. By integrating Zillow housing price data with U.S. Census migration and socioeconomic data, we identified key trends and potential explanatory factors driving regional disparities.

Idaho and Louisiana emerged as particularly notable case studies, with Idaho experiencing the highest median housing price increase (+75%) and Louisiana being the only state to witness a decline (-2.6%). Our investigation into potential causes revealed that migration patterns strongly correlated with housing price changes. Idaho saw a significant influx of new residents, aligning with its rapid price growth, while Louisiana experienced below-average migration levels, consistent with its housing price stagnation.

Further, we examined income growth and commute time changes as alternative explanations for these migration patterns. Idaho displayed a substantial increase in median income, suggesting that higher earnings may have contributed to the state’s housing demand. Louisiana, on the other hand, showed minimal income growth and an increase in commute times, which could have deterred potential residents. However, the precise causal direction of these relationships remains an open question.

Additionally, we evaluated the impact of property taxes on housing price changes but found no strong correlation. Despite Idaho’s significant housing price appreciation, its property tax levels remained relatively stable, suggesting that taxation was not a primary driver of regional disparities.

Overall, our findings indicate that migration, income growth, and post-pandemic relocation trends played a more substantial role in shaping housing prices than traditional factors like property taxes. The broader shifts in housing markets during this period appear to have been driven by demographic preferences, employment mobility, and economic opportunities, reinforcing the idea that the pandemic fundamentally reshaped regional housing dynamics in the U.S.

Future research could expand on this work by incorporating additional factors such as housing supply constraints, remote work trends, and detailed demographic shifts to refine our understanding of these evolving market forces.

We have also produced an interactive shiny app to allow users to explore these data on their own terms, as we think that would be particularly enlightening.