Negative Externality of Traffic Congestion and Housing Values : Evidence from the US Great Lakes Megaregion

Transcription

1 Mid-Continent Transportation Research Symposium Negative Externality of Traffic Congestion and Housing Values : Evidence from the US Great Lakes Megaregion October 24, 2016 Jangik Jin TOPS Lab, UW-Madison

2 Outline Research Background 1. Research Question 2. Research Objectives Methodology Results 1. Empirical Model 2. Study Area / Data 3. Congestion Measure 1. Descriptive Statistics 2. Panel Data Analysis Conclusions/Discussion

3 What are the problems of Traffic Congestion? Noise Air Pollution Stress Energy Consumption Wasted Commuting Time

4 How do households respond to Traffic Congestion? Residential Location Choice

5 So What is the matter? Residential Property Values Residential Location Choice

6 Traffic congestion decreases property values?

7 We don t know There is little evidence..

8 Fortunately, there are some related studies. - Decrease Housing Prices Traffic noise Air pollution Crime Environmental disamenities + Increase Housing Prices Good accessibility Good transportation services Good transportation networks Good neighborhood environments

9 Unfortunately, - There are few studies examining the effects of directly measured traffic congestion on housing prices. Why??? - There have not been available data to measure traffic congestion at the local level. Available Congestion Data Commuting time data (US Census) - cannot directly measure traffic congestion Traffic Congestion Data (Texas Transportation Institute: TTI) - available only at the metropolitan level

10 NPMRDS (National Performance Management Research Data Set) Released October 2013, enables researchers to examine the conditions of transportation performance management at the micro level as well as across the extensive regions.

11 Research Objectives (1) Measure traffic congestion at the local level (2) Explore how traffic congestion affects residential property values

12 Analytical Framework Housing Price = f(housing demand, housing supply) HP t = f(q t D, Q t S Traffic Congestion Q t D = g(t t, M t, D t, L t, S t Q t S = h(i t, M t, C t, U t, V t, Z t HP t = f(t t, M t, D t, L t, S t, I t, C t, U t, V t, Z t j HP it = α + β 1 T it + β 2 T i,t 1 + β 3 HP i,t 1 + β 4 HP i,t 1 + β k k=5 X k,t 1 + δ i + ϵ it

13 Study Area US Great Lakes Megaregion 10 States Wisconsin Minnesota Michigan Iowa Illinois Ohio Kansas Indiana Kentucky Missouri 10 Metros Chicago Detroit Minneapolis Saint Louis Cleveland Pittsburgh Cincinnati Kansas City Milwaukee Indianapolis

14 How to Measure Traffic Congestion? Planning Time Index 95% Speed it PTI it = Free Flow Speed it

15 Planning Time Index

16 Planning Time Index Spatial and Temporal Distribution of Growth in Traffic Congestion

17 Median Housing Prices from Zillow Inc.

18 Median Housing Prices

19 Median Housing Prices Spatial and Temporal Distribution of Growth in Housing Prices

20 Descriptive Statistics Variables Mean S.D. Data Source Median Housing Price ($) 164, ,061 Zillow (2013June~2016May) Growth in Median Housing Price (%) Planning Time Index(PTI) HERE (2013June~2016May) Growth in Planning Time Index (%) Interest rate (%) Federal Reserve Bank Mortgage rate (%) Freddie Mac Construction price index Census Unemployment rate (%) BLS Number of household (N) HUD (USPS vacancy data) Number of firm (N) HUD (USPS vacancy data) Residential vacant rate (%) HUD (USPS vacancy data) Business vacant rate (%) HUD (USPS vacancy data) Urban Census Tiger/shp file Distance to CBD (mile) Census Tiger/shp file Park (%) USGS Lake (%) USGS Spring (March-May) Summer (June-August) Fall (September- November) Winter (December-February)

21 Dynamic Housing Panel Model Variables Model (1) Model (2) Model (3) Model (4) Model (5) Model (6) PTI Growth *** *** *** *** *** *** (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) HP Growth (t-1) *** *** *** *** *** *** (0.0026) (0.0025) (0.0025) (0.0026) (0.0026) (0.0025) ln(hp) (t-1) *** *** *** *** *** (0.0038) (0.0050) (0.0051) (0.0050) (0.0640) (0.0605) PTI (t-1) *** *** *** *** ** (0.0014) (0.0017) (0.0017) (0.0017) (0.0037) (0.0035) Interest Rate (t-1) *** *** *** *** ** (0.0450) (0.0261) (0.0261) (0.0266) (0.0271) (0.6136) Mortgage Rate (t-1) *** *** *** *** *** (0.0162) (0.0162) (0.0166) (0.0165) (0.1991) Construction Cost Index (t-1) *** *** *** *** *** (0.0014) (0.0014) (0.0014) (0.0014) (0.0087) Unemployment Rate (t-1) *** *** *** *** *** (0.0015) (0.0015) (0.0015) (0.0028) (0.0035) Housing Vacant Rate (t-1) *** *** *** *** (0.0008) (0.0008) (0.0008) (0.0043) (0.0042) ln(household) (t-1) *** (0.0025) (0.0027) (0.0027) (0.1263) (0.1192) Firm Vacant Rate (t-1) *** *** *** *** * (0.0005) (0.0005) (0.0005) (0.0021) (0.0019) Park * ** (0.0006) (0.0005) Lake (0.0006) (0.0006) Distance to CBD (0.0001) (0.0001) Urban (0.0056) (0.0054) Metro (0.0147) (0.0143) Summer *** *** (0.0065) (0.0064) Fall *** *** (0.0062) (0.0062) Winter *** *** (0.0059) (0.0058) Constant *** *** *** *** *** (0.2187) (0.2197) (0.2248) (1.3161) (2.0940) Fixed-effects (Zip code) No No No No Yes Yes Fixed-effects (Month) No No No No No Yes Number of Observation 74,552 74,552 74,552 74,552 74,552 74,552 R

22 Dynamic Housing Panel Model (Urban vs. Rural // Metro vs. Non-Metro) Urban Rural Metro Non-Metro (1) (2) (3) (4) PTI Growth ** *** (0.0002) (0.0002) (0.0001) (0.0014) HP Growth (t-1) *** *** *** *** (0.0033) (0.0040) (0.0026) (0.0187) ln(hp) (t-1) *** *** *** *** (0.0765) (0.0984) (0.0605) (0.5408) PTI (t-1) (0.0041) (0.0064) (0.0035) (0.0417) Interest Rate (t-1) ** ** (0.7552) (1.0263) (0.6109) (6.6993) Mortgage Rate (t-1) *** *** *** (0.2455) (0.3318) (0.1983) (2.1223) Construction Cost Index (t-1) *** *** *** (0.0107) (0.0144) (0.0086) (0.0927) Unemployment Rate (t-1) *** *** (0.0043) (0.0058) (0.0034) (0.0480) Housing Vacant Rate (t-1) *** *** ** (0.0046) (0.0104) (0.0041) (0.0728) ln(household) (t-1) ** ** (0.1965) (0.1565) (0.1192) (1.1217) Firm Vacant Rate (t-1) *** ** * (0.0024) (0.0033) (0.0019) (0.0178) Constant *** *** *** *** (2.8522) (3.2900) (2.0900) ( ) Fixed-effects (Zip code) Yes Yes Yes Yes Fixed-effects (Month) Yes Yes Yes Yes Number of Observation 43, ,994 1,558 R

23 Dynamic Housing Panel Model (Distance from the CBD) All < 5 miles 5~10 miles 10~15 miles 15~20 miles (1) (2) (3) (4) (5) PTI Growth *** ** ** ( ) (0.0004) (0.0003) (0.0002) (0.0004) PTI Growth CBD *** ( ) Other variables Yes Yes Yes Yes Yes Fixed-effects (Zip code) Yes Yes Yes Yes Yes Fixed-effects (Month) Yes Yes Yes Yes Yes Number of Observation 74,552 8,875 14,252 10,327 7,720 R

24 Summarize the Results 1. Traffic congestion growth is negatively associated with growth in median housing prices. Such results are slightly changed when including control variables and fixed-effects terms, but almost consistent and robust 2. While the effects are not statistically significant in rural and non-metro areas, the effects are significant in urban and metro areas. 3. There is no significant effect of congestion growth on the housing price growth in areas within 5 mile from the CBD. One possible explanation is that urban amenity-premium such as a better access to jobs, shopping, restaurants, and recreational facilities are greater than disamenities of traffic congestion in areas near CBD. 4. Congestion growth has a negative effect on housing price growth in areas 5~15 miles away from the CBD. But, there is no significant relationship between congestion growth and housing price growth in areas over 15 miles away from the CBD.

25 Conclusion 1. The negative externalities of traffic congestion not only include noise, air pollution, stress, energy consumption, and wasting travel time as previous studies demonstrated, but also is related to residential property values. 2. Although direct mechanism of the relationship between traffic congestion and housing prices is not explained in this study, we verified that the traffic congestion functions as a disamenity in a local neighborhood. 3. Transportation policies that aim to relieve traffic congestion should consider that severe traffic congestion may have an adverse effect on local property values. 4. The negative effect of traffic congestion does not cause a decrease in housing values in areas close to the CBD. This is probably because the benefits of urban amenities exceed the disamenities of traffic congestion. However, getting away from the CBD, the benefits of urban amenities become smaller, consequently, the negative effect of congestion on property values is clear. 5. Although more investigation with consideration of location-specific characteristics and sensitivity analysis should be necessary for providing further policy implications, congestion reduction efforts can bring people to move into these areas.

26 Negative Externality of Traffic Congestion and Housing Values : Evidence from the US Great Lakes Megaregion Questions and Comments?? Jangik Jin jjin8@wisc.edu TOPS Lab, UW-Madison