Baseline model specification Clause Samples

Baseline model specification. To investigate the relation between geographic location and cash holdings, we supplement the following cash model as developed by ▇▇▇▇▇ et al. (1999) with geographic variables and industry and year fixed effects: Cash = β0 + β1 Geographic variables + β2 Control variables + Industry Dummies + Year Dummies+ ε, (1) where Cash is cash holdings computed as the natural logarithm of the ratio of cash and marketable securities to net assets defined as total assets minus cash and marketable securities. We use the 2 ▇▇▇.▇▇▇▇▇▇▇▇▇▇▇.▇▇▇. logarithmic transformation to reduce problems associated with skewness and to mitigate the influence of outlier observations. Appendix 1 provides a description of our variables. All financial variables are winsorized at the 1% and 99% percentile levels to reduce effects of outliers and our regression estimates are based on standard errors adjusted for clustering at the firm level (▇▇▇▇▇▇▇▇, 2009). To determine the geographical location, we first obtain the four-digit ZIP code of the county in which the firm is headquartered the year it entered Worldscope database.3 We next estimate its proximity relative to Paris, the French financial center (Guillain and ▇▇ ▇▇▇▇▇, 2010) using three proxies.4 First, we define Distance as the natural logarithm of one plus distance in kilometers between firm’s i headquarter and Paris, P, using the arc-length formula, as in ▇▇▇▇▇ and ▇▇▇▇▇▇▇▇▇ (1999): DistanceiP = ar cos (deg)∗ 2 Л r/360, (2) deg = cos(lati)∗ cos(loni)∗ cos(latP)∗ cos(lonP) + cos(lati)∗ sin(loni)∗ cos(latP)∗ sin(lonP) + sin(lati) ∗ sin(latP), (3) where r is the radius of the earth (6378 km); lat is latitude; and lon is longitude. Second, we define Distance Road as the natural logarithm of one plus road distance in kilometers between the location of a firm’s i headquarters and Paris region (ZIP Code 75000) as automobile is still the primarily transportation mode in France (▇▇ ▇▇▇▇▇, 2009).5 Finally, we construct a dummy variable Outside_Paris that takes the value 1 if a firm is headquartered outside the Paris region, and 0 if the firms is in Paris (county code 75) or inner-ring suburbs (92, 93, 94) and outer-ring suburbs (77, 78, 91, 95). We use a number of control variables. Following Opler et al. (1999), cash holdings can be explained by a number of firm characteristics. Firm size can be negative due to the presence of 3 Some cases of relocation such as Bioalliance Pharma, Imecom SA and Bac Majestic, are made within the same region. Ca...