The primary datasets used are repeated cross sections from the 1980-2018 Integrated Public Use Microdata (IPUMS), which is the world’s largest individual-level population database (IPUMS). I chose this data source because of its large volume of microdata samples from the United States Census records, as well as the inclusion of those internationally. IPUMS is instrumental in that it gathers geographic and externality data on people through the census questionnaire. Secondary data was obtained from census questionnaires and consisted of cohorts born after 1946 and had an ending birth date of December 31, 2001. As a result, my sample was limited to three generations: Generation X, Generation Z, and the millennials. My focus was on the millennial generation.
The first cohort is from Generation X. Generation X individuals were born between 1965 and 1980. This means that the current age range of this group is 40 to 55 years old. The second cohort is Generation Z, which is the most recent generation and consists of individuals born between 1997 and 2020. The current age range of this group is infancy to 23 years old. The third cohort is the millennial generation – the primary focus of this study – and were born between 1981 and 1996. Thus, the current age range of this group is 24 to 39 years old.
I avoided prior generations and, specifically those born prior to 1946 in the so-called silent generation, because data was not available. In order to be included, the individual had to be at least 18 years of age and, for simplicity, the maximum age was 80 years of age. The age range selected was to allow for improved accuracy in relation to homeownership. For example, if an individual under 18 years of age was selected, they would be less likely to be a homeowner. This would be counterproductive to the purpose of this study because I wanted to focus on a group with a higher chance of homeownership as compared to other generations of the same age, albeit in a different year.
Date of birth, age, and state of survey variables from the census questionnaire allow me the ability to assign all individuals to a state and group level according to the year of birth for generation identification. The time used for birth date is the same used for that within the census data. As such, I am relying upon the accuracy of the census questionnaire data. This information allowed for the definition of generations. Marital status was classified as 1 (married) or 2 (unmarried). Education level effects were identified by the number of years. Labor force indicators were estimated by the summation of all individuals employed or unemployed in the generations. Another consideration was farms and if the property was a farm, it was valued as a 1, whereas non-farms were valued as a 0, which will lead to a property type-related approach. Census data contained personal annual income, which was then multiplied by the consumer price index and linearized by logging it. Family was used, as opposed to household, to create a clearer definition of direct relatives, not others. The home value to annual income ratio was constructed by dividing individual personal income into the total personal income for the state of interest. Alaska, Hawaii, and the District of Columbia were excluded from the study because the population size was not large enough. This will reflect the percentage of home value as compared to the income of the individual.
Within the present study, the primary strategy was through linear regression models, where state, time, and birth year being fixed effects. States included all but Alaska, Hawaii, and the District of Columbia. Time was for census surveys from 1980 to 2017. Birth years was for the birth year of the individuals and is the starting point. Following the example of existing literature, showing the differences on homeownership by generation and group disparities, I estimated the following regression equation:
indicates an outcome for homeownership of an individual in state with year of census survey, and birth year. I also controlled for state , time and age fixed effects . The model has three different vectors that act as controls for generations. Millennials born in 1981 to 1996 are with coefficient . Boomers born in 1946 to 1964 are the omitted category group in the model. Generation X born 1965 to 1980 are with coefficient The last generation are individuals born in 1997 until now are with coefficient .
I initially included dummies for marital status, while education was measured in years at the time of the survey, labor force, type of property (with the latter two variables previously described), family size, and income per individual (as described previously). Age variables provided for the identification of how young an individual was when they became a homeowner and race provided another variable for analysis. Additionally, I included home value to income ratio per cohort in each state selected for the model. Both variable trends help account for differences between generations relative to home ownership. I am concerned about the possibility that my results may be driven by unobserved factors causing generations to have higher or lower homeownership. As a result, I added (time) to the model, which will help identify age and the relationship on owning or renting a home.
Regression (1) (seen in the first column) shows the impact to homeownership by Generation X, Generation M, and Generation Z. Based on the results, none of the generations were likely to engage in homeownership. Therefore, it is evident that as people age, they are more likely to be homeowners. This is based on no dummy variables impacting the generations. In fact, across the three generations – Generation Z, Generation M, and Generation X – the difference comes closer to zero with each generation aging. Regression (2) (seen in the second column) shows the inclusion of the fixed effects of race, marriage, gender, education, and family size to homeownership. When considering these effects, all of the generations were more likely to engage in homeownership. However, it was seen in Regression (2) that millennials were more likely to engage in homeownership when considering the fixed effects as a whole. This was not the case when considering individual outcomes. Race, for example, had no impact on homeownership. In fact, this was seen to be true across all regressions – race had no impact on homeownership in any generation. It was also interesting to note that gender was statistically significant in all cases, except for Regression 2. This indicates that gender influences homeownership. On the other hand, this does not appear to be the case in Regressions 3 and 4. Family size also impacted homeownership for Generation Z and Generation X, but not Generation M. This could be associated with the affordability of it. Regressions (3) and (4) (see columns 3 and 4 in Table 2 respectively) had very similar results. The analysis showed that the generations had similar relationships. For example, Generation X and Generation Z showed a negative relationship with homeownership, whereas Generation M had a positive relationship with homeownership. This positive relationship began when farmland was included in the regression. It can be assumed that this is due to the lowered cost of farmland. Thus, it may be seen that Generation M is more likely to engage in homeownership when it is affordable, and farmland provides opportunities for this to occur. Both additional externalities – farmland and income – were added to the regression in these elements and both had a positive relationship with homeownership (farmland at 8.08% and income at 4.42%).
Figure 3 showed homeownership by generation at the same age based on year of birth, beginning at 18 years old. I found that the younger the individual is, regardless of the generation to which they belong, there is a negative relationship with homeownership. This could be due to unemployment, being a student, having low wages, or insufficient savings. When individuals turn 25 years of age, regardless of the generation, the curve begins to shift upwards, which could mean that they have finished their education and are earning higher wages, have no more student loans to repay, have more savings, or have been promoted in their jobs (leading to increased wages). If you analyze the results between generations, there is a disadvantage for Generation M, as compared to Generation B, and Generation X because they have lived in different economic periods where housing affordability was higher for the latter two generations.
In conclusion, the study analyzed whether or not homeownership varied across generations, as well as how the factors of income, race, marital status, years of education, sex, income, and home value impacted home ownership. Through regression models, it was found that the younger the individual, the less likely they are to own a home. The factors of gender, race, and home value established a negative relationship with homeownership, meaning that no accurate result was determined. If these factors were to be analyzed in the future, I suggest that they be broken down into further detail. The factors of marriage and family size led to the largest positive relationship towards homeownership as compared to income and education. I can conclude that a married couple with access to two incomes may lead to increased affordability for homeownership. The independent variable of farmland had a positive relationship with homeownership, which may be because farmland is away from the city, leading to a lower cost.
At the start of the study, I believed that the independent variable of education was going to have a larger impact on homeownership. However, it was found in the results that education had a very small contribution towards homeownership. Income was the most obvious independent variable to have a positive impact on homeownership. Throughout the study, homeownership as a means to build wealth was discussed. If Generation M wishes to build wealth through homeownership, the results of this study conclude that it cannot be accomplished in the same way as other generations have because of the many obstacles faced due to living in a very different economic time period.