Date of Award
Doctor of Philosophy
Narayan KUNDAN KISHOR
REBECCA NEUMANN, Jangsu Yoon, Filip Vesely
Applied Macroeconomics, Econometrics, Time-series
My dissertation studies different aspects and phenomena of the housing market in the United States. In the first chapter, I explore the long-run relationship between housing starts and building permits along with its short-run deviations. This paper fits a pre-specified cointegration model to verify the long-run co-movement property of building permits and housing starts in the U.S., and its census regions to improve the predictability of housing starts at different forecast horizons. The out-of-sample forecasting performance of housing starts is derived from the feature of the short-run vector error correction model that suggests that only housing starts adjusts to correct for any disequilibrium in the equilibrating relationship between housing starts and building permits. This result is robust to structural breaks, as well as the inclusion of additional controls in the short-run dynamics.
During the first decade of the 21st century, the housing market in the U.S. was not only going through an episode of exuberance, but one of the major causes behind the Great Recession of 2008 was the crash of the housing market. Against this background, the second chapter of my dissertation proposes a method to decompose housing demand into Consumption and Investment motives. For this purpose, housing is allowed to enter both the utility function as well as the budget constraint. Since the two motives of the housing are not separately identifiable in the resulting Euler equations, an unobserved component model is proposed to estimate them. Using data from 1987 through 2019, it has been found that the share of consumption motive in total housing demand is 83%. The results also suggest that the investment motive is much more volatile than the consumption motive and witnessed a big increase before the 2008 financial crisis.
In the third chapter, I explore the time-varying response of the tradable and non-tradable employment if a shock appears in the house prices. I use monthly state-level data from 2001 through 2020 and compare the time-varying impulse responses of tradable and non-tradable employments over different horizons. The methodology I use is a time-varying parameter vector autoregressive model with stochastic volatility. The results show that for 16 out of 45 states, the response to non-tradable employment is higher than that to tradable employment.
De, Akash, "Essays on Housing Market" (2022). Theses and Dissertations. 2882.