Many people ask questions such as “would market crash?”, “What would be the housing price trend in 2023”. Lots of organizations and YouTubers also have tried to predict the housing market. No matter how people predict, they use a thing called predictive model. A predictive model can be an empirical model, or statistical model. No matter what type of model, it uses predictors or we called influential factors to predict. For example, some people might say: “look, the inventory skyrockets! So the price will drop.” This is an empirical model, the predictor is housing sales inventory. Some other people say: “The unemployment rate is historical low; rate hiking will result in housing price adjustment, but not crash”, predictors are unemployment rate and interest rate. Or people could say: ”the increased rate has cooled down the market so much, if the Fed shrinks its balance sheet, the outcome will be devastating!”, in this case, the predictors are Fed’s prime rate and balance sheet total asset information. No matter what models used for prediction, we need to know this quote from the famous statistician George Box: “All models are wrong, but some are useful”. Historically, there are many factors influencing the housing market, some important ones are: population growth, income, employment, interest rate, housing inventory, the Fed’s monetary policy, et al. In this video series, I’ll use Tampa as an example, and show you how these factors impacted the Tampa housing price within the past 30 years. Understand the potential relationship between these factors and housing price. Then I will apply data science and predictive model approach to simulate the potential market scenarios and forecast the potential housing price trend for the 2023 and 2024. Lots of stuff, so I will split the series into two parts, this is Part 1, let’s first dig out insights regarding how those important factors impacted housing price in different historical situations. Have deep understanding of Part 1 will help understand the stuff coming in up Part 2. So please be patient and watch till the end!

I will analyze how the Tampa housing price change from the aspects of population growth, median household income, Hillsborough county unemployment rate, mortgage rate, inventory and Fed’s total asset. One thing I need to call out, these data have different granularities. For example, population data is annual data; unemployment rate is monthly data, and mortgage rate is daily. So, before the analysis, I prepared the data, and summarized the data into monthly level, so that I can have enough data points for analysis and could derive meaningful outcome. First, let’s take a look at the target variable: housing price index. The reference point of housing price index is January 2000, the index is 100%, if house price is greater than that of January 2000, then the index is greater than 100%, vice versa. We can easily tell, Tampa housing price index show an increasing trend in general since late 80s. But experienced some recessions and drastic changes due to 08 crisis. The index significantly hiked in COVID pandemic period, and shows slightly decline starting the second half of 2022. Next we will take a look at the influential factors one by one, overlay with housing price index, and see what we get.

Let’s take a look a population growth vs housing price index first. For easy comparison, I convert the Hillsborough county population count into population index, using the January 2000 as reference point for calculating the index. Divide each year’s population by that of January 2000 to get the index. Then we can overlay the housing price index curve and population growth index curve on the same chart. And we can compare slope of each curve. The steeper the slope, the faster the increase was. We can have a quick conclusion by looking at this chart: Hillsborough county population and Tampa housing price both demonstrate long term increase trend.  The population growth helped housing price increase. This makes sense. Generally speaking, to have a healthy and sustainable housing market, an area needs to have stable population growth and/or in-flow migration. But we can also see something else on this chart. Population growth and housing price show different increase speed for different time periods. For example, house price increased faster than population growth before and after 08 crisis. This indicates that there were other factors impacted the housing price during those timeframes.

Let’s also take a median household income vs housing price index.  Similarly, I convert the Hillsborough median household income into index, and overlay the curve with housing price index curve on the same chart. Similar conclusion here: increase in the income helped house price increase. But the speeds of income increase and housing price increase were different before and after 2008 crisis, which implies that other factors boosted housing price in those periods.

Now let’s explore the relationship between unemployment rate and housing price index. This chart overlays Hillsborough county unemployment rate curve and housing price index curve. It has two different vertical axis, the left one is for housing price index, and the right one is for unemployment rate. We can see that the unemployment rate skyrocketed in 08 crisis, went beyond 10%, negatively impacted housing price significantly. During that recession, lots of foreclosures and bank owned properties pop up, reflected on the spike in housing inventory, as shown by this chart. Subsequently, the supply significantly exceeded demand, resulted in huge price decline, and housing market crash. The short term negative effect brought by the crisis surpassed the long term positive effect of population growth. So the housing price decreased a lot in short term. After the crisis, inventory kept decreasing, and interest rate decreased as well, which brought positive effects on housing price in addition to the population growth and income increase. If we zoom in the curves to the 90s time frame, we can also tell the negative relationship between unemployment rate and housing price. But the housing market didn’t crash that time. So the quantitative effect of unemployment rate on housing price change would be different from that in 2008 recession. In 2020 COVID pandemic period, the unemployment rate also greatly increased, but the market didn’t crash either because of mortgage forbearance. So, unemployment rate cannot fully determine whether market could crash or not.

What about inventory? Let’s compare its pattern with housing price index. The definition of inventory is the ratio of active listing count over pending listing count. We can see, in the 2008 crisis, inventory increased fast and housing price decline fast. And after COVID started in 2020, inventory rapidly decline and housing price rapidly increased, very obvious negative relationship. But if we look more carefully, we can tell that the turning points on the inventory curve and housing price index curve do not overlap. There are some lags between inventory turning points and housing price turning points. It makes sense, as it takes some time for a property to be sold and sold price reflected.

Another factor that can influence housing sales: mortgage rate. This factor has close relationship with Fed’s control on interest rate. I plot the 30-year term mortgage rate with housing price index on the same chart. At the first glance, we can tell overall, mortgage rate change and housing price change demonstrate negative relationship, which makes sense. But for different specific time periods, the interest rate effects on housing price seem to be different. For some time frames, housing price and interest rate even changed in the same direction simultaneously. In those periods, other factors’ effects may be dominating the housing price change.

Last but not least, let’s take a look at another factor related to Fed: the total asset on the Fed’s balance sheet. Fed’s historical expanding and shrinking its balance sheet are reflected on this data. I plot the total asset curve with housing price index. We can tell that after 08 crisis. It shows similar but not exact same patterns between total asset changes and housing price index changes. From previous analysis, we understand that other factors can directly impact the housing price.

Based on all of these exploratory analyses, we can summarize that: (1) population and household income are the major long term drivers for housing price change; (2) unemployment rate, housing sales inventory, interest rate and Fed’s action can have short term impacts on housing price; (3) in particular, unemployment rate and inventory have the most obvious visual impact for short term, and these factors have inverse relationship with housing price index; (4) the impact from inventory and unemployment rate changes may have some lags; (5) Mortgage rate and housing price show inverse relationship in general, but in certain time frames the rate and housing price might show the same direction change when there are other factors influence housing market.