### Table 4 N onlinear Estimation: Heterogeneous Error Term

### Table 6 N onlinear Estimation: Heterogeneous Error Term

### Table 3. Econometric estimates from time-varying advertising parameter models. Variable Parameter Fluid Milk Cheese

"... In PAGE 13: ...9 and BCGW and GCGW are the brand and generic cheese advertising goodwill variables, respectively. Estimation and Testing Results Estimation results are displayed in Table3 . Before discussing those results, we need to evaluate the heteroskedastic nature of the residuals.... In PAGE 14: ... Estimation results reveal both models demonstrate reasonable explanatory power with adjusted R-square values at or above 0.94 ( Table3 ). Wald tests were constructed to test the structural heterogeneity of the advertising parameters.... In PAGE 15: ... The shorter lag-distribution for cheese relative to fluid milk is consistent with the empirical results in Kaiser that applied five-quarter lags to generic fluid milk advertising and three-quarter lags to generic cheese advertising using a polynomial distributed lag structure. Demand Elasticities Given the nonlinear specification of the time-varying parameter models, the regression results of Table3 are most usefully evaluated in terms of calculated elasticities. Table 4 provides selected elasticities for the time-varying models evaluated at the sample means.... ..."

### Table 2: The Collier-Hoeffler Model, no unobserved heterogeneity$

2001

"... In PAGE 10: ... We therefore test both variants. The results are reported in Table2 . The model performs remarkably badly in either variant.... ..."

Cited by 7

### Table 4. Empirical Bias and Standard Error Estimates of Variance Components and Heterogeneity Parameter from Simulations with Variance Heterogeneity Parameter True Value m Empirical Bias Empirical SE Naive Estimated SE Estimated SE

"... In PAGE 15: ... Note that our xed e ects estimators and their standard errors under the heterogeneous variance model coincide with those under the = 0 model. Table4 is here. Table 4 gives the results for the estimates of the variance components and the hetero- geneity parameter and their standard errors.... In PAGE 15: ... Table 4 is here. Table4 gives the results for the estimates of the variance components and the hetero- geneity parameter and their standard errors. \Naive Estimated SE quot; denotes the square root of the mean of 5000 (or 1000) naive estimates of variances, which are given by the diagonal elements of ?1.... ..."

### Table 3. Empirical Bias and Standard Error Estimates of Fixed E ects from Simulations with Variance Heterogeneity

"... In PAGE 14: ... We generated 5000 simulated data sets with m = 79, m = 157, and m = 314, and 1000 with m = 628. Table3 is here.... In PAGE 15: ...Table3 gives the results for the estimates of the xed e ects and their standard errors. \Empirical bias quot; denotes the mean of 5000 (or 1000) estimates minus the true value, the \Empirical SE quot; denotes the standard error of the estimates, and \Estimated SE quot; denotes the square root of the mean of 5000 (or 1000) estimated variances.... ..."

### Table 4: Estimated posterior mean , std. error, 2.5% quantile, median and 97.5% quantile for the parameters of the Poisson-GLM with heterogeneous error term variance (3.8) for the absolute price changes of the Call option on the XETRA DAX index with strike price 2600 and expiration month March 2003 (c1 = 80, c2 = 1) based on the last 400 recorded iterations in each chain .

2006

"... In PAGE 15: ... We had to choose slightly more informative priors for this simulation due to problems of exponential overflow in WinBUGS with the less informative priors used for the simulations of the models discussed in the previous sections. Table4 shows the results of the final MCMC simulation. 4 Assessment of model adequacy While the estimation results of all of the three previously discussed models lead to the same general conclusions about the influence of the explanatory variables on the absolute option price changes (in fact, the fitted regression surfaces as shown in Figure 4 for the simple Poisson-GLM look very similar in shape for each of the three models), the question arises which of the discussed models is the most adequate one.... ..."

### Table 4: Estimated posterior mean , std. error, 2.5% quantile, median and 97.5% quantile for the parameters of the Poisson-GLM with heterogeneous error term variance (3.8) for the absolute price changes of the Call option on the XETRA DAX index with strike price 2600 and expiration month March 2003 (c1 = 80, c2 = 1) based on the last 400 recorded iterations in each chain .

2006

"... In PAGE 15: ... We had to choose slightly more informative priors for this simulation due to problems of exponential overflow in WinBUGS with the less informative priors used for the simulations of the models discussed in the previous sections. Table4 shows the results of the final MCMC simulation. 4 Assessment of model adequacy While the estimation results of all of the three previously discussed models lead to the same general conclusions about the influence of the explanatory variables on the absolute option price changes (in fact, the fitted regression surfaces as shown in Figure 4 for the simple Poisson-GLM look very similar in shape for each of the three models), the question arises which of the discussed models is the most adequate one.... ..."

### Table 3: Estimates for the preferred model in Table 2. Without With Heterogeneity Heterogeneity

1995

"... In PAGE 15: ...2 Cohort was coded in 5-year age groups, 15{19, 20{24, : : :, 45{49. Initially, period was coded as in Table3 , with dummy variables for periods chosen so as to make the standard errors similar. Later, for exploratory purposes, period was coded with a dummy variable for each year, and nally, models were t in which the period e ect was modeled parametrically.... In PAGE 19: ... Model 8 indicates that death of the previous child increased fertility. Table3 shows the estimates for the preferred model, with and without population het- erogeneity. The estimates were similar, and we focus on the results from the model that includes population heterogeneity.... In PAGE 20: ... The preferred model is shown in bold. # Other variables 2 p BIC 1 None 13359 10 ?13247 2 Y3 13505 18 ?13302 3 C 13394 16 ?13214 4 Y3 C 13525 24 ?13255 5 Y3 S 13621 19 ?13407 6 Y3 Ch 13566 19 ?13352 7 Y3 S Ch 13632 20 ?13407 8 Y3 S M 14531 20 ?14306 9 Y3 S M C 14560 26 ?14268 10 Y3 S M Ch 14537 21 ?14300 11 Y1 S M 14592 37 ?14176 NOTE: The independent variables are as follows: Y3 = period (coded as in Table3 below); C = cohort (7 levels); S = size of the place (5 categories) in which the woman resides; Ch = place of childhood residence (city/village); M = child mortality (1 if the previous child was alive; 0 if not); Y1 = period (coded as one dummy variable for 1900{1952, and one for each year 1953{1977). The quantities 2, p and BIC are de ned by equation (4).... ..."

Cited by 6

### Table 1: End-client and Network Heterogeneity

1998

"... In PAGE 1: ...1 Dealing with heterogeneity in the multicast environment Though multicast applications reap enormous perfor- mance bene ts from the underlying multicast service, they are fundamentally challenged by the heterogene- ity that is inherent in the disparate technologies that comprise the Internet, both within the end systems and across the network infrastructure. Table1 shows the high variance in client and network capabilities to- day. End devices range from simple palm-top personal digital assistants (PDAs) to powerful high-end desk- top PCs, while network link characteristics can vary by many orders of magnitude in terms of delay, capac- ity, and error rate.... ..."

Cited by 48