Amazon S3



VI. Supplementary Appendix (for Online Publication Only)Appendix A: Locations Within 50 Kilometers of a Referring Court, 1964-2013Appendix B: The Correlates of Region-Level Yearly References with Year-Fixed Effects, 1997-20121: Italy2: Italy (w/o Lazio)3: North Italy4: South ItalyDV: # Yearly PRsDV: # Yearly PRsDV: # Yearly PRsDV: # Yearly PRsExport Share-0.089****-0.07***-0.0560.095(-5.35)(-3.15)(-1.09)(0.25)ln(Population)1.48****1.44****0.507**2.23***(8.24)(7.68)(2.10)(2.69)Domestic Litigation0.041*-0.0040.517**-0.085(1.74)(-0.11)(2.52)(-1.53)Civic Participation0.0080.02050.063**0.129(0.27)(0.63)(2.04)(0.76)ln(GDP/capita)3.55****2.71****-2.723.67(7.26)(3.47)(-1.03)(0.92)Year FE?YesYesYesYes(Intercept)-45.37****-36.83****22.9-53.96**(-8.93)(-4.77)(0.84)(-1.27)N272258108112t statistics in parentheses* p < 0.10, ** p < 0.05, *** p < 0.01, **** p < 0.001Notes: The dependent variable is the number of yearly references at the regional level. The analysis leveraged heteroskedasticity-robust standard errors for all models. Measures for covariates were obtained from ISTAT and are detailed in footnote 11.Appendix C: The Correlates of Region-Level Yearly Common Market References, 1997-20121: Italy2: Italy (w/o Lazio)3: North Italy4: South ItalyDV: # Yearly Economic PRsDV: # Yearly Economic PRsDV: # Yearly Economic PRsDV: # Yearly Economic PRsExport Share-0.052**-0.014-0.0220.473(-2.32)(-0.51)(-0.29)(0.91)ln(Population)1.187****1.248****0.3321.828**(5.05)(5.32)(1.31)(2.41)Domestic Litigation0.068**-0.070.465-0.284****(2.24)(-1.21)(1.39)(-3.37)Civic Participation0.00820.03810.0353-0.0997(0.20)(0.83)(0.85)(-0.42)Ln(GDP/capita)2.202****0.255-3.421-2.176(3.50)(0.25)(-1.41)(-0.85)(Intercept)-31.73****-12.8031.217.134(-4.65)(-1.29)(1.22)(0.26)N2182078788t statistics in parentheses* p < 0.10, ** p < 0.05, *** p < 0.01, **** p < 0.001Note: Dependent variable is the number of yearly references at the regional level related to EU competition, taxation, free movement of goods and services, and establishment provisions.Appendix D: Predicted Number of Preliminary References by Domestic Litigation Levels across Northern Italy (Left) and Southern Italy (Right)Notes: Appendix D plots the predicted yearly number of regional references (with 95% confidence intervals) across various levels of domestic litigation (holding all other covariates at their means) across northern and southern regions separately. The different values for the X-axes in Figure 3 are a function of the different average yearly number of cases before regional administrative courts in the north and south: For the north, the mean is approximately 2,000 (standard deviation of 1,700); for the south the mean is approximately 4,000 (standard deviation of 4,500).Appendix E: Hotspot Analysis of all Preliminary References Including Rome (left) and Excluding Rome (right), 1964-2013 Notes: Basemap source is Eurostat (2013); Grid is comprised of 25 vertical lines and 25 horizontal lines; Projected coordinate system is ETRS 1989 UTM Zone 33N.Appendix F: Hotspot Analysis of Free Movement of Goods and Services References (Left), Taxation References (Center), and Social Provisions References (Right) with Alternative 50x50 Grid, 1964-2013 Notes: Basemap source is Eurostat (2013); Grid is comprised of 50 vertical lines and 50 horizontal lines; Projected coordinate system is ETRS 1989 UTM Zone 33N.Appendix G: Hotspot Analysis of Free Movement of Goods and Services References (Left), Taxation References (Center), and Social Provisions References (Right) Excluding Rome, 1964-2013 Notes: Basemap source is Eurostat (2013); Grid is comprised of 25 vertical lines and 25 horizontal lines; Projected coordinate system is ETRS 1989 UTM Zone 33N.Appendix H – Information on Getis-Ord Gi* Hotspot analysisTo assess subnational clustering in EU preliminary reference activity, we aggregated city-level point data into a standardized 25-by-25 grid of polygons, such that each polygon aggregates the total number of references originating from each city within its contours. This aggregation is desirable because running the Getis-Ord Gi* on the city-level point data would produce misleading results, since the analysis would omit all cities that never referred cases to the ECJ. Consequently, the Gi* statistic would only capture hotspots among the subset of cities where at least one reference originated. Aggregating point data within a polygon grid remedies this problem: Polygons lacking any referring city register a value of 0 for EU preliminary reference activity and are reintroduced into the hotspot analysis. Because the Gi* statistic is a tool for spatial analysis, one needs to provide a conceptualization for the spatial relationships underlying the data. The standard conceptualization is the fixed distance band, which in this case was computed by running a global spatial autocorrelation analysis, leveraging the Global Moran's I statistic, separately for each issue-specific litigation map, and uncovering the average distance from each polygon at which spatial autocorrelation is maximized. This distance is then leveraged to compute the Getis-Ord Gi* statistic: Neighboring polygons within the specified fixed distance band around any given polygon are included in the hotspot calculations, whereas polygons falling outside the band are excluded from the analysis.Robustness checks including Rome are meant to showcase which hotspots are least sensitive to global increases in the mean polygon-level number of references. Since Rome is the origin of a substantial amount of all types of preliminary references, its inclusion increases the global mean number of references per polygon; Consequently, it becomes harder for any given polygon to register as a hotspot. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download