4.1h <p>Analysis of WDC trends in the US</p>
Analysis of WDC trends in the US
P.I. Name & Address
“Why do warehouses decentralize more in certain metropolitan areas?”
Keywords: land use and transportation, warehousing location change, urban freight movement
During the past decade the demand for customized goods has increased exponentially. To satisfy it, goods producers have drastically altered how/where goods are produced and distributed. As a result, goods production has been dispersed all over the world, and the logistics industry has reprioritized from storage to throughput to move a large volume of resources and products quickly and reliably. Consequently, warehousing and distribution centers (WDCs) have been relocated to urban peripheries, where land is cheaper and more available. This relocation, or decentralization, results from inventory and transportation cost tradeoffs in which the gains of lower rents outweigh the increase in transportation costs as WDCs move further away from the market.
Since efficient freight movement is essential for the smooth functioning of metropolitan areas, WDC decentralization should occur everywhere. However, this is not necessarily true. The extent of decentralization varies across metropolitan areas, depending on their population size, industry mix, foreign trade involvement, and geography. According to my previous study in California, not all major metro areas have experienced decentralization. San Francisco, notwithstanding its role in high-tech manufacturing and trade, has a geographically limited land supply, which has hindered decentralization. Whereas, Sacramento, even with its plentiful land availability, has not been significantly involved in foreign trade. Likewise, San Diego, despite being a border city, has physical constraints – the border, coast, and hilly terrain – that have inhibited significant decentralization. In Los Angeles, in contrast, WDC decentralization was inevitable because land for large-scale operations was available on the outskirts of the urban area. This decentralization mostly occurred because this unique environment is the largest trade node and the second largest metropolitan area in the U.S.
In this study, I test the metro-level factors that might explain the disparity in the patterns of WDC decentralization across major metropolitan areas in the U.S. Metropolitan areas have unique characteristics that could be either favorable or hostile to WDC operations. This study contributes to the theoretical understanding and empirical testing of the WDC decentralization.
To explain the disparity, I use four factors. The first is metropolitan size, which is correlated with density – a proxy for demand (land prices). The largest metro areas have the highest peak density and average density. Therefore, land intensive activities – manufacturing, trade, and warehousing – require cheap land, which is available in the urban peripheries. The second is the employment density gradient, which is approximated by a negative exponential curve of employment density ( ), where D(r) is the employment density at distance r from the urban center; r is distance to the urban center; D0 is the employment density at the center; β is the density gradient; u is the error term. As high land rent pushes land-intensive businesses away from the center, the distance at which it decreases to a favorable level for WDCs depends on the rent gradient. The third is the extent to which a metro area is involved in foreign trade. WDCs that accommodate foreign trade are more likely to decentralize to maintain sufficient capacity for global commerce, as opposed to those oriented to local markets. The fourth factor is industry mix. Freight flow and WDC demand are a function of industry mix; thus, wholesaling or warehousing-oriented metro areas would generate more freight demand than those that are service-oriented. As with the third factor, more freight demand implies more decentralization.
As decentralization is an action of location change over time, I formulate multiple OLS regression models in which the change in the average distance from WDCs to the urban center between 2003 and 2013 is a function of baseline circumstances of the four factors in 2003. It is the longest duration for which WDC location information is available in ZIP Code Business Patterns datasets. The general OLS regression model is,
Change in WDC location 2003-2013 = f(metro size (or employment density gradient), foreign trade, industry mix, all in 2003)
This model, controlling for all other factors, estimates the true effect size of a factor. As the collinearity between the first and second factors is expected, both factors will not be included simultaneously. The unit of analysis used in the model is each of the 67 metropolitan areas in the U.S. I document a non-linear relationship between WDC decentralization and metropolitan size and explain how their statistical significance and explanatory power vary across estimated models.
- Bowen Jr., J., 2008. Moving places: the geography of warehousing in the US. Journal of Transport Geography, 16, 379–387.
- Cidell, J., 2011. Distribution centers among the rooftops: the global logistics network meets the suburban spatial imaginary, International Journal of Urban and Regional Research, 35:4, 832-851.
- Dablanc, L. and Ross, C., 2012. Atlanta: A Mega Logistics Center in the Piedmont Atlantic Megaregion (PAM), Journal of Transport Geography, 24, 432-442.
- Hesse, M. and Rodrigue, J-P., 2004. The transport geography of logistics and freight distribution. Journal of Transport Geography, 12:3, 171–184.