This dissertation consists of three essays in which I study the consumption and investment decisions of households in the presence of two major asset classes: housing and human capital.

In the first essay, I analyze how the dual consumption-investment nature of housing affects the consumption dynamics of households. A key feature of the housing market is that for most households, the consumption and investment benefits of housing are non-separable. I propose a tractable theoretical framework to understand the impact of this constraint on the consumption allocation of homeowners who would ideally like to own just a fraction of their home. For these homeowners, the relative cost of living in their home is not just the imputed rental cost. It also includes an opportunity cost of having an unbalanced financial portfolio. This cost varies substantially over time, and it is especially high in good times, when available investment opportunities yield high returns and homeowners allocate a high fraction of their wealth to current consumption. As a result, this cost dampens variations in the level of their housing consumption, and it amplifies variations in both their level of non-housing consumption and the composition of their consumption baskets.

I then test empirically this theory in the second essay. Using household-level data from the Panel Study of Income Dynamics (PSID), I test the hypothesis that homeowners who face a high opportunity cost choose ceteris paribus a low housing consumption volatility. I also develop a method to identify these constrained homeowners by comparing their characteristics to those of a subset of unconstrained homeowners: the landlords. The results are consistent with the predictions of the model. First, the characteristics of homeowners that determine how constrained they are in the model are strong predictors of those homeowners who choose to be landlords in the data. For example, homeowners with a low level of risk aversion, little value for housing consumption, and a long horizon are relatively more likely to be landlords. Second, I find evidence that the more constrained homeowners adjust their level of housing consumption much less over time.

In the third essay, which was developed in collaboration with Thomas Jansson, Christine Parlour, and Johan Walden, we investigate the relationship between workers' labor income and their investment decisions. Using a detailed Swedish data set on employment and portfolio holdings we estimate wage volatility and labor productivity for Swedish industries and, motivated by theory, we show that highly labor productive industries are more likely to pay workers variable wages. We also find that both levels and changes in wage volatility are significant in explaining changes in household investment portfolios. A household going from an industry with low wage volatility to one with high volatility will ceteris paribus decrease its portfolio share of risky assets by 25%, i.e., 7,750 USD. Similarly, a household that switches from a low labor productivity industry to one with high labor productivity decreases its risky asset share by 20%. Our results suggest that human capital risk is an important determinant of household portfolio holdings.