Spatial aggregation error and agglomeration benefits from transport improvements

Aggregating data to other spatial resolutions can result in spatial aggregation errors, which can affect the data and subsequent analyses. This paper investigates the nature of spatial aggregation errors in the calculation of agglomeration benefits. The finding shows that inappropriate aggregation can result in significant errors in the monetised benefits used in transport appraisals. The error is analysed using both a stylised model and four case studies. Using arithmetic employment-weighted averages for transport costs when calculating the access to economic mass (ATEM) of the aggregated zones resulted in an error of +/-70%. An arithmetic trip-weighted average reduces errors. The error is further reduced using a harmonic employment-weighted average in the ATEM calculation, but it is not eliminated. The remaining error stems from the altered structure of the data stemming from the aggregation process, which is called the modifiable area unit problem (MAUP). Our findings, which concur with the broader MAUP literature, indicate that this error is unavoidable. Our policy recommendation is to calculate the agglomeration benefits as disaggregated as possible, but if aggregation is necessary, use the harmonic employment-weighted average for transport costs.