This thesis advances spatial and urban economics by quantifying how local policies, heterogeneity in preferences and supply composition are interconnected in housing markets. These chapters demonstrate that treating housing markets as a uniform and spatially isolated market overlooks critical spillovers, segmentation and misallocation. Chapter 1 exploits the data for 228 English districts for 1996Q1-2023Q4 to test for spatial autocorrelation in house prices and to estimate spatial spillovers for planning refusal density on prices. We constructed a three year rolling mean of the number of refusals per km2 of a district for major housing applications and estimates baseline IV FE model and Spatial Durbin Model. Our results depict that there is a signicant amount of spatial autocorrelation (around ρ = 0.8) which proves that district-wise the housing markets are well interconnected with a well division between the North and the South of England. For every 1% increase in planning refusal density will lead to a 0.4−0.8 increase in standard deviation of house prices. There is also presence spatial spillovers as seen from the SDM with the indirect eects lining up to be 3 times stronger as compared to its direct counterpart. This implies that a planning is not local with consequences that lead to supply issues to the neighboring districts using lagged planning delays as an IV. Next, in Chapter 2, we test for the hypothesis to what extent are the housing markets segmented by its type and household groups and does the relative segment-specic sup- ply matter for prices. For this, we used the data from the Woon suvrey and the Dutch NVM housing transaction data. It documents dierences in oorspace consumption by demographic groups and denes housing types on the basis of oorspace and demo- graphic groups. Here we considered three demographic proles; households with kids, households without kids and retirees along with 3 housing types (based on oorspace); < 150m2,nogarden, < 150m2,garden and > 150m2. Our results documented for the three demographic groups, they have dierent preferences for oorspace and gardens. There is clear evidence of segmentation; households with kids occupy housing with dwellings with garden while while households with no kids choose oorspace less than 100m2. We then also employed an structural expenditure regression in which helped us identify the substitutability parameter (ρ). This helped us identify ρ= 1.11 which makes substitutability dicult for demographic groups in housing types with own-segment sup- ply elasticities of −0.14. ix Finally in Chapter 3 we progress towards how does the composition of dierent hous- ing types across locations eect equilibrium sorting, prices, wages and welfare in a spatial setting. The model is calibrated to match observed sorting patterns and segment-specic prices and then we use it to conduct counterfactual experiments that change the compo- sition of housing stock. Our ndings show that misallocation is substantial; to equalize price per m2 segments would require 20−37 percentage point reallocation with houses with < 150m2,nogarden being oversupplied and > 150m2 being under supplied. The counterfactual scenarios indicate that housing type target. Firstly, with a 10 percentage points (pp) increase in houses with < 150m2,nogarden would lead increase in welfare by 2.5% for childless households but leads to a loss in welfare for families due to their increase in the price index. Secondly, for the increase in aging population drives seniors into <150m2,nogardenin Rotterdam by 5 percentage points. A cross-sectional analysis also reveals that the growth in the demographic is primarily driven by the change in the the growth of their own preferred housing type. Lastly, when we increased the increase in large houses (>150m2) by 10 pp, then the price index decreases by approximately by 15%, while at the same time the price indices for the other segments would also change.
Essays in Spatial and Urban Economics
MUKHERJEE, SANMOY
2026-07-09
Abstract
This thesis advances spatial and urban economics by quantifying how local policies, heterogeneity in preferences and supply composition are interconnected in housing markets. These chapters demonstrate that treating housing markets as a uniform and spatially isolated market overlooks critical spillovers, segmentation and misallocation. Chapter 1 exploits the data for 228 English districts for 1996Q1-2023Q4 to test for spatial autocorrelation in house prices and to estimate spatial spillovers for planning refusal density on prices. We constructed a three year rolling mean of the number of refusals per km2 of a district for major housing applications and estimates baseline IV FE model and Spatial Durbin Model. Our results depict that there is a signicant amount of spatial autocorrelation (around ρ = 0.8) which proves that district-wise the housing markets are well interconnected with a well division between the North and the South of England. For every 1% increase in planning refusal density will lead to a 0.4−0.8 increase in standard deviation of house prices. There is also presence spatial spillovers as seen from the SDM with the indirect eects lining up to be 3 times stronger as compared to its direct counterpart. This implies that a planning is not local with consequences that lead to supply issues to the neighboring districts using lagged planning delays as an IV. Next, in Chapter 2, we test for the hypothesis to what extent are the housing markets segmented by its type and household groups and does the relative segment-specic sup- ply matter for prices. For this, we used the data from the Woon suvrey and the Dutch NVM housing transaction data. It documents dierences in oorspace consumption by demographic groups and denes housing types on the basis of oorspace and demo- graphic groups. Here we considered three demographic proles; households with kids, households without kids and retirees along with 3 housing types (based on oorspace); < 150m2,nogarden, < 150m2,garden and > 150m2. Our results documented for the three demographic groups, they have dierent preferences for oorspace and gardens. There is clear evidence of segmentation; households with kids occupy housing with dwellings with garden while while households with no kids choose oorspace less than 100m2. We then also employed an structural expenditure regression in which helped us identify the substitutability parameter (ρ). This helped us identify ρ= 1.11 which makes substitutability dicult for demographic groups in housing types with own-segment sup- ply elasticities of −0.14. ix Finally in Chapter 3 we progress towards how does the composition of dierent hous- ing types across locations eect equilibrium sorting, prices, wages and welfare in a spatial setting. The model is calibrated to match observed sorting patterns and segment-specic prices and then we use it to conduct counterfactual experiments that change the compo- sition of housing stock. Our ndings show that misallocation is substantial; to equalize price per m2 segments would require 20−37 percentage point reallocation with houses with < 150m2,nogarden being oversupplied and > 150m2 being under supplied. The counterfactual scenarios indicate that housing type target. Firstly, with a 10 percentage points (pp) increase in houses with < 150m2,nogarden would lead increase in welfare by 2.5% for childless households but leads to a loss in welfare for families due to their increase in the price index. Secondly, for the increase in aging population drives seniors into <150m2,nogardenin Rotterdam by 5 percentage points. A cross-sectional analysis also reveals that the growth in the demographic is primarily driven by the change in the the growth of their own preferred housing type. Lastly, when we increased the increase in large houses (>150m2) by 10 pp, then the price index decreases by approximately by 15%, while at the same time the price indices for the other segments would also change.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
