"... In PAGE 26: ... 5.4 Regression results Table7 contains the results from regressing (-1, +1) and (-3, +3) related firm CARs (stated in percent terms) on various combinations of the explanatory variables. We include both sets of CARs to extract patterns that persist outside of the three days surrounding the event.... In PAGE 26: ... The p-values under the coefficient estimates (in parentheses) are based on the same bootstrapping procedure used in Table 4 that preserves the cross-sectional error structure of firms associated with the same event. The first thing to note from the results in Table7 is that most of the coefficient estimates are statistically insignificant and that the variables together explain very little of the variation in the related firm CARs. The highest adjusted R-squared using the (-1, +1) CAR is 1.... ..."

"... In PAGE 11: ...ural-to-urban poverty rates has leaped up to 1.4 as shown in Figure 3. The 1990s have seen rising rural incidence relative to urban. Table2 summarizes the trend rates of poverty reduction by state underlying Figure 1. The trend rates of poverty reduction have varied greatly across states, with Kerala the highest rate of poverty reduction (both as a proportion and in percentage points per year) and Assam the lowest, with Bihar close behind (and Jammu and Kashmir, in the linear case).... ..."

"... In PAGE 30: ...) This model was estimated by GLS, using the sum of the variances of the estimated returns to education in 1970 and 1980 as the measure of the variance of the dependent variable.28 Table10 presents the results. All regressors except those shown drop out in the first-differenced model.... ..."

"... In PAGE 23: ... When Na is the number of enterprises in the sample, ra measures labor redundancy in the entire economy. The estimated values of ra across all sectors of activity and in the entire economy are reported in Table7 . The last four columns in this table are based on the same ri indicators used to draw Figure 2.... ..."

"... In PAGE 13: ... Projected population estimates for these time periods were obtained from the US Census Bureau. Table2 contains a summary of the number of urban cells per person for each county and the number of future urban cells required given the projected population growth for each county. Note that Grand Traverse County has a population density that is twice that of the other counties.... In PAGE 16: ...nd 1990 required 4.52 cells (or 0.452 ha) of urban land. Over 92,000 people will be added to the six county area between 1980 and 2020, ( Table2 ) representing an increase of 73% in total population. Percentage decadal growth for each county is expected to average around 13%, with many counties potentially increasing total population by nearly 20% in some 10-year periods (Table 2).... In PAGE 18: ....3. Watershed-scale land use projections Four projections for the six county area were produced for the years 1990, 2000, 2010, and 2020 (Fig. 7) using the PID values derived from US Census Projections ( Table2 ).... ..."

"... In PAGE 3: ... The FIPA-based template of the event-driven DSC-based lightweight agent. In order to exemplify the DSC-based modeling of agent behavior, the specification of the ADSC of a mobile event-driven state-based lightweight agent is shown in Figure 3; Table1 reports state variables, methods and events of the example agent specification. The agent overall goal is that of moving across a set of agent servers according to a predefined itinerary for monitoring a set of remote processes.... ..."

"... In PAGE 5: ... 5 Multimodal rendering based on contact states and events The generated events can be used with several rendering techniques to enhance the perception of contact. Among the numerous possibilities of association between an event and a rendering technique, we chose to implement an intu- itive rendering of contact for all the modalities ( Table1 ) to improve the realism of the interaction in a complex virtual... ..."

"... In PAGE 5: ... Two different areas have been simulated, a small area with 7 base stations and a large area with 67 base stations. Table1 presents the most important characteristics of the respective models. In our current implementation the state size does not change between the small and large areas.... In PAGE 6: ...he method of Cleary et. al. [4] (EQ 5) is presented in Table 3. When compared to the figures in Table1 we see that this method predicts that both TISS and UISS should outperform CSS for all benchmarks with one exception when TISS is used for the large FCA model. The experi- mental results of Section 4.... In PAGE 6: ...5) compared to CSS. When contrasted to the frac- tion of the total execution time spent on state saving using CSS, Table1 , we see that this is a near optimal result. The a.... In PAGE 7: ... Speed-up for the large FCA model on 4 processors Figure 5. Checkpoint interval for the large FCA model on 4 processors The DCA models features radically different character- istics compared to the FCA models, Table1 . In particular the state size is much larger and the fraction of the state modified in an event execution is much smaller.... In PAGE 8: ...than what could be attributed only to the reduction of the state saving overhead. As indicated by Table1 , the fraction of the execution time spent on state saving when CSS is used is 84% for this particular model. A complete reduc- tion of this overhead cannot account for more than a 6.... In PAGE 8: ... For the large DCA model, Figures 8 and 9, the perform- ance difference between the SSS mechanisms and the ISS mechanisms is not as pronounced as for the small model, Figure 6. This can be explained by the larger event execu- tion times for this model, Table1 . Furthermore, the maxi- mum speed-up is close to the 5.... In PAGE 9: ...3 Mb and 27.8 Mb respec- tively, Table1 . Thus only memory consumption over these figures are due to state (or event) saving.... ..."