Journal of Student Research 2018

The Impact Of Uber’s Presence On Taxi Fare

49

To account for differences among the markets other than the presence of Uber, I used each market’s zip-code to collect information from the 2010 U.S. census, to gather a sufficient set of control variables. The control variables included population and region of the country. The summary statistics for the potential control variables are displayed in Table 3.

2010 Population

1-mile

5-mile

10-mile

Mean

$5.99

$18.07

$32.40

26,951 19,453

Standard Deviation

1.15

2.8

5.03

Count

Table 3: Sample summary statistics 265 Table 3: Sample summary statistics 265 265 265

Econometric Framework

As mentioned, the presence of Uber results in an increase in the supply of ride-sharing services in any market which the app enters. Thus, Hypothesis H1 predicts that Uber’s presence results in a lower average taxi fare in the market. The primary regressor in my econometric analysis is the binary measure of Uber presence; “1” if Uber is present, “0” if Uber is not present. A parsimonious model, i.e., a model with only the primary regressor included, was run with the average taxi fare for one-mile, five-mile, and 10 mile rides serving as three separate dependent variables. As shown in Table 4, the presence of Uber does not significantly affect average taxi fare in this model (similar to the univariate analysis presented above).

Coefficients t Stat

Intercept

32.2979

36.800

Uber

0.121345 0.129367

Adjusted R Square -0.00374 Standard Error 5.041717

Table 4: Estimation results for parsimonious model Table 4: Estimation results for parsimonious model

Taxi fare is a function of many factors other than Uber presence, such as the distance of the ride, the time spent in the taxi, and the region of the country. Therefore, the insignificant results for the parsimonious model are not unexpected. Lack of control variables in this model specification contribute to the positive coefficient associated with Uber. To correct for the omitted variable bias certainly present in the parsimonious model, a multiple linear regression model was used to identify the effect of the presence of Uber on taxi fare. A linear equation allows us to