### Table 3: Gravity Model of Trade

2004

"... In PAGE 16: ...5. Discussion of the Results Results from the baseline gravity equation, for 1995 to 1999, are reported in Columns (1) through (5) of Table3 . Columns (6) through (10) include the Internet variable, HOST.... In PAGE 16: ... This suggests that there is significant multicollinearity between the host variable and these variables, most likely because the number of Internet sites in a country is positively correlated with per-capita income, and GDP and population together capture per-capita income in the baseline gravity equation (log(GDP/POP)=logGDP-logPOP). This implies that our estimate of the coefficient on the Internet variable in Table3 overestimates its true value, and should be interpreted as an upper bound of the effect of the Internet on trade. To account for this possible upward bias, we include initial trade patterns from 1995 in the regression equation.... In PAGE 17: ... In contrast to the coefficients on population and GDP, the effect of distance on trade changes only slightly when HOST is included in the regression equation. In the baseline gravity equation, as shown in the first five columns of Table3 , the coefficient on distance falls slightly over time. When the host variable is included in the regression equation, as shown in the next five columns, the coefficient on distance becomes more stable, providing some evidence that the Internet has reduced the way in which distance impacts trade.... In PAGE 20: ... For both the exporter and importer the impact of the Internet is increasing over time. As in the general total trade results from Table3 , the coefficient estimates fall slightly in 1999 but are still highly statistically significant. The coefficients from 1999 can be interpreted as implying that a relative increase of 10 percent in domestic hosts would lead to about a 2.... ..."

Cited by 1

### Table 3: Estimated travel speed, given road grade and weather

1992

"... In PAGE 6: ... The duration of a trip is the sum of the driving times for each segment on the itinerary, plus the time needed for loading and unloading shipments during the trip. The driving time for a segment is the length of the road used for that segment divided by the estimated speed for that road (as determined by the weather and road grade, see Table3 ). Union drivers can be on a trip for at most 11 hours, while non-union drivers can be on a trip for at most 12.... ..."

Cited by 1

### Table 7: Transport cost inclusive gravity model, 1994

"... In PAGE 16: ... The drawback with this procedure is that we implicitly assume that all the independent variables affect each of the 2- digit SITC product groups in an identical way. 19 Table7 presents our estimation results. The tariff plus transport cost variable performs according to expectations: it has a negative sign and is statistically significant at the 1 percent level.... ..."

### Table 5: Gravity Model of Trade, Lagged Host

2004

"... In PAGE 18: ...trade a lot have greater incentive to launch web initiatives to facilitate that trade. Although we cannot completely control for this possibility, in Table5 we use two- and four- year lagged values of HOST instead of the contemporaneous variable.10 The results show that the lagged HOST variable is not statistically significant in 1996 in the regression, indicating that the Internet had no or very little systematic effect on trade flows during this time period.... ..."

Cited by 1

### Table 2 Summary of the 21 real road networks used in the evaluation No. state number

"... In PAGE 4: ... One important characteristic of a real road network is the degree of connectivity measured by the arc-to-node ratios. It can be seen in Table2 that the arc-to-node ratios range from 2.... ..."

### Table 4 FDI and Ethnic Chinese Networks: Gravity Equation

"... In PAGE 14: ... It often works reasonably well in explaining multinational production (Ekholm (1998), Brainard (1997)). Results based on the gravity equation are collected in Table4 . The dependent variable is the log of cumulative inward FDI over the period of 1983-1997.... ..."

### Table 3: FTP and TVTP Logistic Models for GDP

"... In PAGE 11: ... In light of this outcome, no similar two indicator TVTP analysis was undertaken. The sign and significance of the mean growth rates for the Hamilton FTP model in Table3 indicate classical business cycle behaviour, with the data being classified into positive and negative growth rate regimes, with a mean quarterly growth rate of .76% in regime 1 (expansion) and - .... In PAGE 11: ... Note that the model selection criteria AIC and SIC cannot be compared with the linear AR specification as an indication of regime switching non-linearity because of the non-standard conditions that are involved (see Hamilton and Perez- Quiros, 1996). The TVTP logistic results are also presented in Table3 . The estimates of the regime dependent means ( m0 + m1 and m0 ) associated with these models are statistically significant and again indicate classical business cycle behaviour in GDP.... In PAGE 30: ...ogarithms. The first difference also performed better at forecasting. 7 As the logistic model for the Treasury Bill yield time varying recession slope coefficient was insignificant a variation of this model was tried by fixing this probability to be zero. The resultant SIC for this model was lower than the reported model in Table3 , but the forecast MSFE was no better. 8 In these circumstances the statistic delivers 0/0.... ..."

### Table 2. Comparison between Harvest TOPEX altimeter bias estimates using 1-day JGM-2 and orbits computed using the TOPEX tuned gravity eld model. These orbits were computed using SLR and Doris Doppler tracking data

"... In PAGE 9: ...5 cm at Lampedusa (JGM-2 bias will be more negative). Table2 shows the comparison between bias estimates at Harvest using JGM-2 orbits and using orbits computed with the TOPEX tuned gravity eld model for 22 over ights. These biases were estimated using a 5th-order polynomial as interpolator and using both 10/sec and 1/sec data.... In PAGE 9: ... Tide gauge values were obtained by taking the average of all available tide gauge systems at a certain over ight. Table2 clearly shows the existence of geographically correlated radial orbit di erences, presumably caused by errors in the JGM-2 orbits. The mean of the di erences between the individual bias estimates is 1.... In PAGE 9: ...5 cm). Table2 shows that the mean of the bias estimates using the tuned gravity eld model is equal to -17.... In PAGE 9: ...o -17.7 cm with an rms about mean of 3.3 cm (fast-rate data) at Harvest. Table2 also shows that the bias estimates based on the fast-rate data are in close agreement with the bias estimates... In PAGE 12: ... Again, Table 4 shows that a positive o set exists between the estimates using the two di erent types of orbits. The o set value compares very well with the orbit di erences between the JGM-2 orbits and the orbits computed using the TOPEX tuned gravity eld model ( Table2 ). It is also interesting to note that the JPL determined TOPEX bias estimate (using 6 over ights) has a value of -17.... In PAGE 12: ... It is also interesting to note that the JPL determined TOPEX bias estimate (using 6 over ights) has a value of -17.7 cm, which agrees very well with the mean bias estimate displayed in Table2 , which is also -17.7 cm using the TOPEX tuned gravity eld model (22 over ights).... ..."

### Table 1. Properties of the used road networks.

2006

"... In PAGE 12: ... All graphs5 have been taken from the DIMACS Challenge website [17]. Table1 summarises the properties of the used networks. 4 14 countries: Austria, Belgium, Denmark, France, Germany, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and the UK 5 Note that the experiments on the TIGER graphs had been performed before the final versions, which use a finer edge costs resolution, were available.... ..."

Cited by 13

### Table 4. Gains (over two years) induced by a 1% increase in road length per capita

"... In PAGE 15: ... Growth spillovers and the provision of infrastructure This section examines the possibility of a better targeted infrastructure policy in Africa. The estimated coefficients are used to calculate gains and costs associated to an increase in road length per 1000 inhabitants by 1% in one given country i at the beginning of period t ( Table4 ). Due to growth spillovers among neighbouring countries, the representative neighbour j of country i will also indirectly benefit from the infrastructure investment made by i .... ..."