Visitor Attraction Model of Trade and Service Facilities at the Business Unit Level in the Planned Government Centre Area of Serang Regency

Estimating the traffic impact of the planned Serang Regency Government Centre requires attraction models for every constituent land use, including the trade and service facilities that dominate the Serang-Jakarta arterial corridor. This study develops a visitor attraction model from 138 business units across nine activity types in Ciruas and Kragilan Districts, using daily visitor counts as the dependent variable and three independent variables: parking area, building area, and number of employees. The analysis works at the business unit level, unlike the common practice of aggregating to village level, which reduces the sample to 27 observations. All three independent variables correlate positively and significantly with the dependent variable, at 0.875 for parking area, 0.730 for building area, and 0.552 for employees. Multiple regression yields a coefficient of determination of 0.836, yet the employee coefficient turns negative and significant (B = -28.182; p < 0.001). Adding business type controls raises the coefficient of determination to 0.862 without restoring the sign, and logarithmic transformation of all three predictors merely relocates the reversal. The linear specification also fails the Breusch-Pagan test (LM = 847.2; p < 0.001) and is dominated by 14 market units whose removal lowers the slope by 78 percent. The elasticity model ln Y = 2.068 + 0.501 ln X, with X the parking area, is the only specification satisfying every criterion, with a coefficient of determination of 0.372 and Breusch-Pagan p = 0.320. The elasticity of 0.501 implies diminishing returns: doubling parking area raises visitor numbers by about 41 percent