### Table 1 I. Testing the null hypothesis of a unit root: PP-tests

"... In PAGE 8: ... Along these lines, DeJong et al (1989) and Diebold and Rudebusch (1991) provide evidence that the Dickey-Fuller tests have low power against near-unit root and fractionally integrated alternatives. In the case of our dataset, the unit root null is accepted in all cases on the basis of the Phillips-Perron tests, as the first part of Table1 suggests. To address the issues of power raised above, we adopt the approach advocated by Kwiatkowski et al (1992) (referred to as KPSS).... In PAGE 9: ... ,..., (13) The asymptotic distribution of this statistic has been derived by Kwiatkowski et al (1992) under the null and under the unit root alternative, and critical values have also been tabulated. The second part of Table1 contains the results from testing the hypothesis that the level series are stationary around a linear trend. This hypothesis is clearly rejected for all the series.... In PAGE 23: ...eries is stationary around a mean. The 5% critical values are 0.146 and 0.463 respectively (source, Kwiatowski et al (1992), Table1 , pp166). Table 2 Misspecification tests for the unrestricted VAR(2) model p-values Eq st mt mt* yt yt * lt lt* AR(5) 0.... ..."

### Table 1. Patterns in the theoretical ACF and PACF of stationary time series.

"... In PAGE 12: ... PACF can be obtained by Yule-Walker equation [23]. Table1 represents the patterns in the theoretical ACF and PACF of stationary time series, which is utilized to determine the order of model. ... In PAGE 13: ...igure 5. Analysis of ACF and PACF patterns: (a) ACF; (b) PACF. (a) (b) Taking the IMF1 in Figure 4 for example, its ACF and PACF are presented in Figure 5. According to Table1 , we choose the model AR(6) . In the same way, models are chosen for all the IMFs and the final residue series.... ..."

### Table 4 Coefficient in Adjusted Autoregressive Integrated Moving Average (ARIMA) Models United States Japan Taiwan Korea United Kingdom Singapore

"... In PAGE 28: ...Table4 shows the estimates of MSW generated in hotel restaurants. The lowest amount, 18,859 tons, was recorded in 1986, and the highest was 24,191 tons recorded in 1993.... In PAGE 28: ...5% in this period. Table4 also shows the environmental costs, at the current price, of solid waste incurred by the local hotel industry. The total environmental cost attributable to MSW rose from HK$9.... In PAGE 61: ...of Equation 4 as indicated in Table4 can be compared among different countries, because the economic indicators are of different scales. The coefficients in Equa- tion 4 indicate the extent of influence on sign and magnitude of the ARIMA resid- ual of an economic indicator with respect to the stationary series of the tourist arrival from a country.... In PAGE 61: ... Moreover, one can compare the patterns or the dependent factors among the countries. For instance, tourist arrivals from Japan, as shown in Table4 , which has the highest number of factors being related, would be highly affected by economic factors. This would give us some insight on the nature or underlying economic factors that are relevant to our decision making and thus can assist us on how to revise our tourism policy or strategy upon economic changes.... In PAGE 61: ... Based on the above discussion, it might be suggested that there is a relation- ship between economic indicators and visitor arrivals. Concerning the economic indicators of Hong Kong, the results (see Table4 ) indicate that the GDP of Hong Kong is a rather relevant indicator that affects visitor arrivals from all of our stud- ied countries except Korea. The second influential indicator is the exports of Hong Kong, which affect visitor arrivals from the United States, Japan, and Tai- wan.... In PAGE 61: ... On the side of the originating countries, GNP is again an important indicator for predicting visitor arrivals. Four out of six countries are affected by this indica- tor (see Table4 ). As GNP represents the income of a country, the higher the income, the higher the spending power of its people.... In PAGE 61: ... However, the money supply of Singapore is not considered because Singapore only reports its money supply after 1997 and does not have sufficient data for the analysis. As indicated in Table4 , the CPI, unemployment rate, and imports of the origi- nating country do not have much implication on visitor arrivals to Hong Kong. Cho / TOURISM FORECASTING... In PAGE 82: ...centage scores. Table4 shows that the mean dynamic knowledge percentage score on the pre- test was .... ..."

### Table 7: Bayesian Ex Post Probabilities of a Stationary Series Conditional on Significant Evidence of Stationarity

"... In PAGE 10: ...September 9, 1992 3:36 pm Page 10 of 20 post probabilities of having a true unit-root DGP conditional on the tests indicating the presence of a unit root (or in the case of the PP test, simply failing to reject the null). Table7 shows the corresponding probabilities of having a truly stationary DGP conditional on the tests indicating stationarity. Table 6 shows that in large samples (T = 200) there is little difference between the PP and PP-KPS tests in our confidence when detecting a unit-root.... In PAGE 10: ...5, while the joint test allows us to be more confident of our results, particularly for large k. Table7 shows significant differences between the single and joint testing procedure in all sample lengths when we conclude the series is stationary. Based on the single test, we would believe this conclusion to be true between 60 and 75 percent of the time for most choices of k and T, while with the joint test these probabilities rise to the 70 to 95 percent range.... In PAGE 10: ... However, we should expect the joint test to produce frequently inconclusive results. As is shown in Table 8, using the same priors used to construct Table 6 and Table7 , we should expect such results roughly 25 to 50 percent of the time, but particularly in short samples with large k and in large samples with small k. Some may suspect that the superior reliability of the joint test is simply the result of comparing it to the PP test, which we know suffers from severe size distortion in the presence of negative MA parameters.... ..."

### Table 4. Sampling Experiments for the Stochastic Volatility Model

2001

"... In PAGE 14: ... Due to the computational demands, we limit the posterior draws to 20,000. Each experiment was repeated 200 times, and the resulting parameter estimates are summarized in Table4 . Also, , which posterior distri- bution requires an extensive number of draws to estimate accurately, is xed to 1 2 in the simulation study.... In PAGE 14: ... For the parameter measuring the reversion in the log-volatility process, z, the estimates are also slightly downward biased, while the estimate of the di usion coe cient for the log-volatility process, z, is virtually unbiased using a sample size of 2000. When interpreting the results in Table4 , it should be kept in mind that the reversion coe cients can be inter- preted as approximately quot;autocorrelation - 1 quot;. This, cou- pled with the well-known fact that sample autocorrelation coe cients are downward biased for processes with near unit roots may explain, at least partially, the direction of the bias in the sampling experiments.... In PAGE 14: ... It does, however, in uence the dispersion of the posterior means, as measured by the RMSE. Hence, the results in Table4 are likely to underestimate the e ciency of the MCMC estimator. Overall, the performance of the MCMC method for es- timating the parameters in the stochastic volatility model is di cult to assess.... ..."

Cited by 30

### Table 2: Tests of the time-series properties of the dataa

"... In PAGE 7: ...ppendix 1. The data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Appendix 2. Table2 : Tests of the time-series properties of the data .... In PAGE 16: ... Both the ADF test and the modified Phillips-Perron test allow us to test formally the null hypothesis that a series is I(1) against the alternative that it is I(0). The results from the tests of the time-series properties of the data can be found in Table2 in Appendix 2. ADF critical values are generated to account for the finite-sample distribution of the series by performing Monte Carlo simulations with 5,000 replications for the level of inventories, the level of new orders, capacity utilization, the price of raw materials and the yield spread.... In PAGE 16: ... Evidence was found that capacity utilization contains a moving-average component, while the yield spread appears to follow an autoregressive moving- average process.9 Table2 (Appendix 2) indicates that both the ADF and the tests suggest that inventories, new orders, and raw material prices are non-stationary or I(1) processes in levels. The ADF test rejects the null hypothesis of a unit root in the level of the yield spread at conventional levels of significance and also provides evidence that capacity utilization is characterized as a stationary or I(0) process.... ..."

### Table 2 Computation time versus computer characteristics for the studied configuration Computer Type specifications Simulation time for 1 shot*

2001

"... In PAGE 12: ... For instance, as a compromise between complexity and speed of the delivery, in this study we had to step back from a preferred 12 tank-in-series plug-flow system to a 3 tank-in-series complete-mix configuration. Table2 summarises the calculation time required to run a 1-year shot on three different PCs... ..."

Cited by 2

### TABLE I: Details on time series from di erent subjects with normal and irregular phonation. The sampling rate was 3704 frames per second for all time series. The di erence in length is due to the di erent length of high-speed sequences obtained from the clinical investigation and due to the selection of stationary segments. With respect to the typical oscillation frequencies the number of measured cycles are su cient for a statistical analysis such as our eof estimation calculation.

### Table 4 Fisher Test of Ho: There is a common unit root vs Ha: At least one series is stationary No Intercept, No Trend in ADF Specifications All variables are real (deflated by GDP deflator) and measured per capita in logarithms

2001

"... In PAGE 12: ... In this formulation the coefficient on housing market wealth is significant in all specifications, while the coefficient of financial wealth is essentially zero. Table4 presents tests for the presence of unit roots in the time series data we analyze. For most, but not all, of the state series we can reject the hypothesis of unit roots in the data.... In PAGE 13: ... The housing market appears to be more important than the stock market in influencing consumption in developed countries. 8 The specific test we report in Table4 uses a model with no intercept and no trend in conducting the augmented Dickey-Fuller (ADF) tests. The table also relies upon a four-quarter lag for the state panel, and a one-year lag for the country panel.... ..."

Cited by 2