The Effects of SARS on the Hong Kong Tourism Industry: An Empirical Evaluation

Transcription

1 Asia Pacific Journal of Tourism Research, Vol. 10, No. 1, March 2005 The Effects of SARS on the Hong Kong Tourism Industry: An Empirical Evaluation Alan K. M. Au 1, Bala Ramasamy 2 and Matthew C. H. Yeung 1 1 School of Business and Administration, The Open University of Hong Kong, Hong Kong 2 Centre for Europe-Asia Business Research, University of Nottingham, UK The objective of this study is to assess the temporal impact of SARS on the tourists arrival in Hong Kong. An econometric strategy was carefully selected to determine the existence of unit roots in data series containing the number of tourist arrivals from 36 source countries between 1978 and The existence of unit roots can detect the stationary properties of the series. The analysis finds that data series of 24 countries contain unit roots and hence any form of exogenous shocks, like the SARS epidemic, can have permanent impact on the number of tourist arrivals. Included in this category are Japan, Taiwan, the US and the UK, which are the main source of tourists for Hong Kong. The paper recommends that authorities take source-country-specific measures to manage the negative effect of SARS. Key words: SARS, Hong Kong Tourism Industry, Unit Root Introduction In a span of 6 years, Hong Kong, together with other East Asian economies, struggled with 3 major crises. The 1997 financial crisis caused both the foreign exchange and stock markets to nose dive. In 2001, the terrorist attacks on the United States and its aftermath sent shock waves through out the region. As if these were not enough, early 2003 saw the economic growth in the region screeching to halt as the SARS epidemic swept nations bordering the Pacific Ocean. So debilitating was the effect of SARS that Mark Clifford of the BusinessWeek (2003, April 11) compared it to the 1967 Cultural Revolution in China. Among all affected cities, Hong Kong was the worst hit with 1,755 confirmed cases with 299 fatalities (WHO, 2003). The number of cases reported in Hong Kong represents 22 percent of all cases and 63 percent of cases outside mainland China. The myeung@ouhk.edu.hk ISSN print/issn online/05/ # 2005 Asia Pacific Tourism Association DOI: /

2 86 A. K. M. Au et al. city-state that was known as the Pearl of Asia due to its legendary beauty at night soon was labelled the Masked City as during the height of the outbreak people who dared to walk the streets of Hong Kong were all masked. Of course, the media (for instance, Bray, CNN.com 2003, October 2) all over the world printed these photos as their headlines. The entire resident population of Amoy garden, for instance, where a large cluster of SARS cases broke out, were sent to quarantine camps in late March 2003, was yet another media headlines. These types of comprehensive coverage of the outbreak provided by media all over the world had alerted most prospective travellers. Tourism is one of the most important economic pillars of the Hong Kong economy as it is the largest foreign exchange earner (Heung & Qu, 1998). It contributed around 6% to Hong Kong s GDP over the last decade (Zhang et al., 2001). The tourism industry in Hong Kong faced its severest challenge from this outbreak. The drop in number of tourist arrival was phenomenal. Figure 1 shows that the drop in the number of tourists arrival was by a factor of 4. The average drop in passengers flying Hong Kong-based Cathay Pacific and Dragonair was particularly evident. Cathay axed 45 percent of all flights resulting in a loss of HKD 3 million per day (see Cathay s 2003 Interim Report), while Dragonair experienced a drastic drop of 76 percent in its passengers compared to the same season last year. Overall, a 50% drop in scheduled flights was equivalent to almost half a million-abandoned seats (see Dragonair s 2003 Operational Statistics). It was not surprising then that the Asian Development Bank estimated that Hong Kong would suffer the greatest drop in annual GDP due to the SARS outbreak (see CNN.com, 2003, May 2). It could also force a slide in the expected 3.5 percent economic growth for 2003 (Kwok, K. C., a chief economist with Standard Chartered Bank, CNN.com, 2003, March 26). The effect of the epidemic on the tourism sector in Hong Kong was so severe that Rick Miller, vice-president of research and economics at the World Travel and Tourism Council (WTTC) claimed that The impact of SARS on these countries (China, Hong HK Tourists Arrival No. of tourists 1,800,000 1,600,000 1,400,000 1,200,000 1,000, , , , ,000 0 Nov-02 Dec-02 Jan-03 Feb-03 Mar-03 Apr-03 May-03 Period Jun-03 Jul-03 Aug-03 Sep-03 Oct-03 Figure 1 Hong Kong Tourists Arrival November 2002 October 2003.

3 The effects of SARS on the Hong Kong Tourism 87 Kong, Vietnam and Singapore) has been four or five times the impact of September 11 in the States (Clarks, BBC News 2003, May 15). Miller also predicted that the impact of SARS on the tourism industry in Hong Kong would be temporary, with a return to normality by To what extent is this accurate? Given that SARS has had an unprecedented effect on tourism, it is unclear if the impact is temporary or has ruined the reputation of Hong Kong as a tourists Mecca indefinitely. Previous studies have attempted to forecast the impact of an exogenous shock on the tourism sector. Enders et al. (1992), for example used the ARIMA model to predict the effect of terrorism on the tourist industry in Western Europe. To our knowledge, there has been no empirical study that has considered the impact of shocks on the Hong Kong tourism industry. Specifically, the objective of this research note is to determine if the impact of SARS on tourist arrivals in Hong Kong is temporary or permanent in nature. It is expected that tourists from different countries would act differently to any form of external shocks, and so the impact would vary from one source country to another. By choosing an appropriate econometric strategy we evaluate the impact of SARS on tourist arrivals in Hong Kong at the source country level. While one could argue that the SARS is a transient issue, the recent SARS cases reported in Taiwan and China (see Koo & Fu, 2003 and Clark, BBC News 2003, May 15) may be an indication that it would be beneficial if Hong Kong prepares itself for any further outbreak of the disease or any similar shocks to its tourism industry. The study also offers methodological contributions to tourism research and future research directions by suggesting a way to quantify arrival patterns with respect to the concept of stationarity and demonstrates the use of a research method that has been rarely used in the discipline. Adopting the concepts in the present study and replicating the methodological approach could empirically evaluate the effects of other tourism crises. The rest of the note is organized as follows: The following section explains the methodological strategy we have used. This is followed by a discussion of our findings and finally we end with some concluding remarks. Data and Methodology In simple terms, a series is said to be stationary if its statistical property remains constant along a time path. Under such circumstances, any form of external shock could have a transient and diminishing effect on the series, implying that the series will naturally return to its original property over time. Conversely, a series is said to be non-stationary if it has no long-run statistical property over a time path. Econometricians refer to this type of series as a random walk. Since this type of a series does not possess its own property, any exogenous shock will persist and hence the effect on the series will be long-term in nature. In order to determine whether a particular series is stationary or non-stationary, one has to detect whether the data series contains a unit root. There exist a number of procedures for testing the presence of unit root in a data series. One could agree that Dickey and Fuller s (1979, 1981) ADF approach is perhaps the most recognized and frequently used method. Dickey and Fuller (1979; 1981) suggest a number of statistics, including t t, t at, t bt, F 3, F 2, t m, t am, F 1 and t, to test the presence of unit root in a data series. In order to use these statistics effectively, a strategy needs to be employed. The present study

4 88 A. K. M. Au et al. exploits the strategy based on Enders (1995) and the modification introduced by Hoffmann et al. (2003). Table 1 describes the methodology employed in this study. Data on the number of tourists entering Hong Kong from 36 countries between 1978 and 2001 were collected for the present study. These 36 countries make up nearly 100 percent of the tourists who visit Hong Kong. Data were sourced from various issues of Hong Kong Tourism Statistics published by the Hong Kong Tourism Board. Table 1 Methodology Step 1 Estimate equation (1) with OLS Dy t ¼ a 0 þ gy t 1 þ a 2 t þ S k 1 b idy t 1 þ 1 t (1) Step 2 where y is the data series of interest; t denotes time; D is the first difference operator; a 0, g, a 2 and b i are the parameters to be estimated; 1 denotes the error term. The value of k is first determined. This step is crucial because lagged differences could whiten the error term such that the limiting distributions and critical values obtained by Dickey and Fuller can be assumed to hold. Ng and Perron (1995) show that information criteria such as the AIC chooses very short lag length which leads to a high size distortion, while on the other hand, the general-to-specific approach has a tendency to choose a higher lag length which results in a loss of predictive power. Thus, both approaches are not considered here. Instead, a bottom-to-top approach is used to determine the value of k. Using an LM test, autocorrelations up to the second order are tested. If both LM tests do not reject the null hypotheses of no first order and second order autocorrelation, it is an indication that k ¼ 0 is acceptable. However, if any of the null hypotheses are rejected, the value of k will be increased until the null hypotheses are accepted. Such a bottom-to-top lag length selection strategy is especially useful when data length is modest, as it could minimize the loss in the degrees of freedom (Hoffmann et al, 2003). Equation (1) with the selected k is used to test for H 0 : g ¼ 0 using the t t statistic. If the null hypothesis is accepted, we progress to the next step. Otherwise, one would conclude that the series is trend-stationary. Establish a null hypothesis of a 2 ¼ 0 given g ¼ 0 and test it using the t bt test. If this null is accepted, the joint test, F 3 (Dickey and Fuller, 1981), is used to test H 0 : a 2 ¼ g ¼ 0 in order to reconfirm the results. F 2 could also be used to gain additional insight. If the null hypothesis of a 2 ¼ 0 given g ¼ 0is rejected, the standard normal distribution is used to re-test the H 0 : g ¼ 0. Rejection of H 0 : g ¼ 0 leads to the conclusion that the series is trendstationary. If the H 0 : g ¼ 0 is accepted, we proceed to the next step. (continued)

5 The effects of SARS on the Hong Kong Tourism 89 Table 1 continued Step 3 Estimate equation (2) with OLS Dy t ¼ a 0 þ gy t 1 þ S k 1 b idy t 1 þ 1 t (2) Step 4 Step 5 The value of k in equation (2) is selected based on the processes described in Step 1. H 0 : g ¼ 0 is tested using the t m statistics. The process is stopped if the null is rejected. Otherwise, the testing continues to step 4. Test the null hypothesis of a 0 ¼ 0 given g ¼ 0 using the t am statistics. Proceed to step 5, if this null hypothesis is accepted. Additionally, the F 1 test with H 0 : a 0 ¼ g ¼ 0 based on equation (2) is then used to reconfirm the acceptance of this null. If the null hypothesis of a 0 ¼ 0 given g ¼ 0 is rejected, the standard normal distribution is used to retest H 0 : g ¼ 0. Rejection of H 0 : g ¼ 0 leads to the conclusion that the series is stationary. Estimate equation (3) with OLS Dy t ¼ gy t 1 þ S k 1 b idy t 1 þ 1 t (3) Determine k using the processes recommended in Step 1. H 0 : g ¼ 0 is tested using the t statistic. If the null is rejected, one could conclude the series is (zero-mean) stationary. Results and Discussion The test statistics based on the strategy employed are provided in Appendix 1. The results of our findings are reported based on five categories as follows: (1) Stationary the impact of external shocks would gradually diminish over time. This would imply that in the long run, the number of tourists from countries in this category would return to the original constant level. (2) Trend-stationary the impact of external shocks would gradually diminish over time. In this case the number of tourists from these countries would return to its long-term trend. Thus, in these cases the impact created by shocks is temporary. (3) Random walk the impact of external shocks on non-stationary series would not diminish over time. This would imply that there would be a permanent effect on the number of tourists from countries that come under this category. (4) Random walk with a drift the impact of external shocks on a non-stationary series with a drift behaves exactly the same as (3) except that the series is either drifting upward or downward. In the context of this study, the interpretation of the drift is redundant but in short, the impact of shocks on such a series is permanent. (5) Random walk with a drift and a trend the impact of shocks on a non-stationary series with a drift and a trend would not diminish over time. However, since the

6 90 A. K. M. Au et al. data series exhibits a trend, both the trend and the shock contribute to the changes in the number of tourists. Although the impact of these shocks on such a series is permanent, the growth in the number of tourists is governed by a trend. Thus, the impact is not as harmful as in (3) and (4). Let us demonstrate how the United States fit into the random walk with a drift category. In the first step, we find that the t t statistics is not significant which leads us to step 2. The null hypothesis of a 2 ¼ 0 given g ¼ 0 is tested using the t bt statistics. The results, however, are not significant. An insignificant F 3 and F 2 would confirm that the trend should not have been included in equation 1. Thus, we proceed to step 3. Given an insignificant t m, we proceed further to step 4. In step 4, the null hypothesis of a 0 ¼ 0 given g ¼ 0 using the t am statistics is tested. An insignificant t am suggests that the drift should not have been included in equation 2. This is confirmed by an insignificant F 1. Consequently, we proceed to step 5. Here, a positive t suggests an explosive process in the series. Enders (1995) and Elder and Kennedy (2001) suggest that the best option at such an uncomfortable corner is to rule out the possibility of an explosive series, since it does not really make economic sense. We may conclude that it occurs by chance or due to misspecification in the model. Therefore, one would be wise to work backwards and conclude that the drift term was not equal to zero in equation 2. As the t m statistics based on equation 2 is insignificant, we finally classify the US series as a random walk series with a drift. Using similar strategy we identify countries that fit into the five categories as shown in Table 2. Our results from Table 2 highlight several important points. First, among the 36 source countries, 24 depict random walk properties. This would imply that tourists from these countries are vulnerable to any form of external shocks and make adjustments to their tourism destination whenever news of these shocks is received. As mentioned earlier, the extensive media coverage of SARS would have a debilitating effect on the number of tourist arrivals from these countries. It must be noted that among these 24 source countries are Japan, Taiwan, the US and the UK, which together make up nearly 60 percent of tourists arrivals into Hong Kong. Second, among the 12 countries that portray stationary properties are Australia, Singapore, the Philippines, Indonesia, South Korea and Canada. Among these countries, the Philippines and Indonesia are important sources of imported labour for Hong Kong. Singapore, Australia and Canada, on the other hand, are prominent migration destination among Hong Kong s professionals. Thus, the relationship with these countries may go beyond typical tourists relationship per se into family and work related ties. This may provide plausible explanation as to the temporary nature of shocks on these source countries. Third, and perhaps more importantly, it is imperative that the tourism authorities handle the SARS issue differently according to the categories described above. For instance, in markets like Japan, Taiwan, the US and the UK where shocks cause a permanent impact on tourist arrivals, the authorities will need to pacify the touring community by perhaps emphasizing the efforts of the government in isolating the SARS carriers etc. A greater allocation of funds to these markets is required so that an introduction of a positive shock will counter the negative effect of SARS. It must be noted that a positive shock can have a permanent impact on tourist arrivals from these countries as well. Image building

7 Table 2 Countries by Stationarity Categories Category A Pure Random Walk A Random Walk with a Drift A Random Walk with a Drift and a Trend Trend-Stationary Stationary Country Austria USA Netherlands Canada Argentina Switzerland Mexico Israel Denmark Australia Portugal Brazil India Norway Bahrain Venezuela Taiwan South Africa Japan UK Kuwait Sweden Saudi Arabia France South Korea Belgium Indonesia Italy Philippines Spain Singapore UAE NZ Malaysia Thailand Implication Moving up and down randomly, reacting to news as it is received; a shock, or an innovation, has a sustained effect in these series The series is not centered on zero; a shock, or an innovation, has a sustained effect in these series Over the long run, the number of tourists from these countries are increasing; there appears to be a random walk around a trend; a shock, or an innovation, has a sustained effect in these series Over the long run, number of tourists from these countries are increasing (decreasing for Kuwait and Saudi Arabia); a shock, or an innovation, has a diminishing effect over time Mean reversing; time invariant; have a broadly constant amplitude; a shock, or an innovation, has a diminishing effect over time The effects of SARS on the Hong Kong Tourism 91

8 92 A. K. M. Au et al. advertising and providing incentives to travel agents in these source countries are some examples of these positive shocks. Similarly, in source markets that are price sensitive, price discounts in air tickets and accommodation could be emphasized. The sliding of the US dollar, which has a direct effect on the pegged Hong Kong dollar, could help in this regard. On the other hand, countries showing stationary properties may not require similar attention, as there is a natural tendency for tourist numbers to return to normality in the short-run. In terms of recovery marketing strategies, Sonmez, Apostolopoulos and Tarlow (1999) suggest that teams or management units for marketing and promotional activities at the destination level must be formed to manage the recovery marketing process including any required re-imaging or branding activities. Ritchie (2003) suggests that these teams should comprise of representatives from local government, travel and tourism industry professionals and community leaders. Given the various categories from the analysis, the recovery teams could be formed according to the grouping of countries we have devised. Conclusion The impact of SARS on Hong Kong s tourism industry is said to be more damaging than the 9-11 episode or the 1997 Asian Financial crisis. This paper dwells into the temporal effects of shocks, such as SARS, on tourist arrivals in Hong Kong. Using data that spans 23 years ( ) and 36 source countries, our analysis finds that 12 countries-argentina, Australia Canada, Denmark, Norway, South Africa, Kuwait, Saudi Arabia, South Korea, Indonesia, Philippines and Singapore do not contain unit roots and so possess stationary properties. This implies that any shock that would likely affect the tourism industry in Hong Kong would have only a temporary impact. On the other hand, data relating to the other 24 countries do contain unit roots and so shocks can cause permanent diversions from the trend; thereby causing a sustained effect on the number of tourists from these countries. Among these latter countries are Japan, Taiwan, the US and the UK, which make up 60 percent of tourist arrivals. We advice policy makers to undertake damage control activities at a country specific level i.e. based on the temporal nature of these tourists. It must be noted that our analysis uses total tourist arrivals. Future research could consider the effect of shocks like SARS on particular classification of tourists, for instance business travelers, backpackers, group travelers etc. References Bray, M. (2003). Shortcomings in HK SARS outbreak, CNN.com. Accessed online on October at 10/02/hk.sars.report/index.html Cathay Pacific Airways Limited (2003). Interim Report Accessed online at com/cx/intracx/content/documents/71496/ 71496_71496_ir2003e.pdf Clark, E. (2003). Sars strikes down Asia tourism, BBC News. Accessed online on May at news.bbc.co.uk/2/low/business/ stm Clifford, M. L. (2003). How SARS Is Strangling Hong Kong, BusinessWeek. Accessed online on April at dnflash/apr2003/nf _8175_db065.htm CNN.com (2003). SARS infecting economies, CNN.com. Accessed online on May at com/2003/world/asiapcf/east/05/10/sars.finance/ Dickey, D. A. & Fuller, W. A. (1979). Distribution of the Estimators for Autoregressive Time Series with a Unit Root, Journal of the American Statistical Association, 74,

9 The effects of SARS on the Hong Kong Tourism 93 Dickey, D. A. & Fuller, W. A. (1981). Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root, Econometrica, 49, Dragonair (2003). Operational Statistics 2003, at template?series ¼ 1&lang ¼ eng&article ¼ 1082 Elder, J. & Kennedy, P. E. (2001). Testing for Unit Roots: What Should Students Be Taught? Journal of Economic Education, 31, Enders, W. (1995). Applied Econometrics Time Series, John Wiley, New York. Enders, W., Sandler, T. & Parise, G. F. (1992). An Econometric Analysis of the Impact of Terrorism on Tourism. Kyklos, 45(4), Heung, Vincent C. S. & Qu, Hailin (1998). Tourism Shopping and its Contribution to Hong Kong, Tourism Management, 19(4), Hoffmann, R., Lee, C. G. & Ramasamy, B (2003). Shocks to Malaysia s exports: Temporary or Permanent? Journal of Asia-Pacific Business, 5(1), Koo, J, & Dong, F. (2003). The Effects of SARS on East Asian Economies, Federal Reserve Bank of Dallas Expand Your Insight, Accessed online on July at Kwok, K. C. (2003). WHO wants more China help on SARS, CNN.com. Accessed online on March at Ng, S. & Perron, P. (1995). Unit Root Tests in ARMA Models with Data-Dependent Methods for the Selection of the Truncation Lag, Journal of American Statistical Association, 90, Ritchie, B. W. (2003). Chaos, crises and disasters: a strategic approach to crisis management in the tourism industry, Tourism Management, 25(6), Sonmez, S. F., Apostolopoulos, Y. & Tarlow, P. (1999). Tourism in crisis: Managing the effects of terrorism, Journal of Travel Research, 38(1), World Health Organisation (2003). Communicable Disease Surveillance & Response (CSR), at Zhang, Q. H., Wong, K. F. & Or, Y. S. (2001). An analysis of historical tourism development and its implications to the tourism industry in Hong Kong, Pacific Tourism Review, 5(1 2),

10 94 A. K. M. Au et al. Appendix 1 Tests Statistics Country t t t bt t at F 3 Argentina (0.008) (0.273) (0.256) (0.027) Australia (0.015) (0.053) (0.004) (0.034) Austria (0.012) (0.027) (0.243) (0.026) Bahrain (0.038) (0.071) (0.007) (0.031) Belgium (0.014) (0.022) (0.736) (0.036) Brazil (0.054) (0.074) (0.581) (0.150) Canada (0.009) (0.009) (0.964) (0.029) Denmark (0.001) (0.001) (0.239) (0.002) France (0.010) (0.015) (0.833) (0.030) India (0.374) (0.218) (0.833) (0.232) Indonesia (0.001) (0.001) (0.035) (0.003) Israel (0.940) (0.244) (0.470) (0.074) Italy (0.463) (0.786) (0.056) (0.382) Japan (0.007) (0.030) (0.390) (0.025) Kuwait (0.003) (0.025) (0.003) (0.010) Malaysia (0.046) (0.050) (0.449) (0.129) Mexico (0.007) (0.061) (0.031) (0.021) Netherlands (0.021) (0.014) (0.518) (0.037) Norway (0.002) (0.003) (0.947) (0.008) NZ (0.007) (0.012) (0.012) (0.023) Philippines (0.004) (0.007) (0.739) (0.013) Portugal (0.108) (0.423) (0.737) (0.217) Saudi Arabia (0.003) (0.169) (0.009) (0.008) Singapore (0.002) (0.002) (0.542) (0.007) South Africa (0.002) (0.007) (0.022) (0.008) South Korea (0.005) (0.006) (0.108) (0.015) Spain (0.026) (0.022) (0.145) (0.067) Sweden (0.076) (0.154) (0.241) (0.116) Switzerland (0.238) (0.667) (0.145) (0.288) Taiwan (0.056) (0.027) (0.197) (0.073) Thailand (0.037) (0.160) (0.056) (0.092) Turkey (0.390) (0.301) (0.266) (0.009) UAE (0.047) (0.120) (0.180) (0.134) UK (0.198) (0.283) (0.107) (0.291) USA (0.015) (0.022) (0.022) (0.047) Venezuela (0.123) (0.135) (0.394) (0.261) Values in parentheses are p-values for t-tests not t tests.

11 The effects of SARS on the Hong Kong Tourism 95 F 2 t m t am F 1 t (0.059) (0.013) (0.023) (0.042) (0.567) (0.019) (0.082) (0.031) (0.044) (0.224) (0.055) (0.176) (0.147) (0.331) (0.796) (0.057) (0.055) (0.044) (0.123) (0.834) (0.045) (0.405) (0.062) (0.044) (0.501) (0.134) (0.460) (0.195) (0.316) (0.437) (0.007) (0.907) (0.058) (0.001) (0.002) (0.004) (0.193) (0.118) (0.285) (0.914) (0.021) (0.330) (0.095) (0.163) (0.159) (0.019) (0.236) (0.801) (0.014) (0.003) (0.003) (0.337) (0.150) (0.242) (0.518) (0.009) (0.049) (0.838) (0.005) (0.001) (0.200) (0.170) (0.053) (0.098) (0.359) (0.051) (0.103) (0.098) (0.230) (0.705) (0.023) (0.057) (0.065) (0.153) (0.921) (0.046) (0.831) (0.082) (0.010) (0.012) (0.037) (0.109) (0.078) (0.198) (0.790) (0.000) (0.517) (0.084) (0.001) (0.001) (0.004) (0.468) (0.112) (0.140) (0.241) (0.017) (0.285) (0.108) (0.153) (0.283) (0.016) (0.421) (0.102) (0.136) (0.513) (0.366) (0.118) (0.178) (0.279) (0.397) (0.020) (0.005) (0.010) (0.017) (0.231) (0.002) (0.370) (0.095) (0.005) (0.009) (0.013) (0.133) (0.084) (0.208) (0.442) (0.022) (0.060) (0.345) (0.014) (0.005) (0.024) (0.874) (0.320) (0.133) (0.079) (0.166) (0.131) (0.085) (0.220) (0.984) (0.405) (0.126) (0.100) (0.249) (0.882) (0.012) (0.622) (0.175) (0.049) (0.040) (0.126) (0.093) (0.056) (0.153) (0.841) (0.003) (0.004) (0.636) (0.001) (0.657) (0.108) (0.207) (0.079) (0.162) (0.471) (0.047) (0.295) (0.065) (0.085) (0.220) (0.032) (0.584) (0.156) (0.058) (0.054) (0.180) (0.520) (0.222) (0.270) (0.291)