摘要:The research results show that sign-orientedmodels have better performance in the prediction of abnormal earnings compared totraditional model. Using forecasted abnormal earnings in all of the valuation functionsindicate that none of the studied models can appropriately forecast firm’s value.
eal values are
equal.
It is obvious that if the model could estimate the prices correctly, the results of two above tests
should indicate the equality of average and mean of market prices (P) and estimated values (V). In both
tests above, the null hypothesis is confirmed on the equity of means and/or medians. And, hence, it is
necessary to confirm the null hypothesis for both tests to support the second assumption. Table 7,
display the results obtained from T test and Wilcoxon signed-ranks test in four models.
Table 7: The Results of means and medians Equality Test
T test(H0: V P μ = μ ) Wilcoxon signed rank test(H0:medv=medp) Models
T statistic Sig Z statistic sig
1 7.414 0.000 -15.372 0.000
2 7.409 0.000 -15.618 0.000
3 0.748 0.455 -9.057 0.000
4 6.949 0.000 -14.686 0.000
The contents of this table, in most models fail the null hypothesis in the 1% significant level.
So, it can be concluded that there is significant discrepancy between means and medians of market
prices and estimated values in all models (except the third model for which only the means equality
assumption-and not medians equality assumption-has been confirmed). Therefore, the EVR hypothesis
in all four studied models is failed. But, it is required to compare the valuation power of four models
studied in this research to find the superior model. For the same purpose, the comparison of the
average absolute value of the valuation errors of four models is used. Results show that traditional
models cause lower errors in the firm’s valuation and, thus, have a better function compared to signoriented
models. Second column in Table 8 illustrates this comparison totally for all estimation
periods. This table indicates the occurrence of more valuation errors when using sign-oriented models.
71 International Research Journal of Finance and Economics - Issue 36 (2010)
Moreover, the results of this table show that the third model cause less valuation errors compared to the
first model(for the aim of comparing traditional models).
In most researches involving in the review of the application of Ohlson (1995) and Feltham-
Ohlson (1995) linear models, the estimated values are usually lower than real values in most cases. In
this research, as shown in the last three columns of table 8, first model in 93%, second model in 95%,
third model in 78%, and fourth model in 96% of observations, respectively, have underestimated the
firm’s value.
Table 8: Comparative survey on the Valuation Power of Four Models
Models Average of Absolute
valuation errors
Total number of
observations Observations with V>P Observations with V
1 696207 348 24 324
2 731786 348 17 331
3 335740 348 75 273
4 778770 348 15 333
In view of all cases presented above, though the EVR hypothesis has not been supported in all
models, the statistical results regarding this assumption, unlike the EFR hypothesis, shows that
traditional models have a better function compared to sign-oriented models. But, with considering the
non-ability of any of the four models in firm’s valuation, and the better performance of sign-oriented
models in forecasting abnormal earnings one cannot certainly introduce a superior model in
本论文由英语论文网提供整理,提供论文代写,英语论文代写,代写论文,代写英语论文,代写留学生论文,代写英文论文,留学生论文代写相关核心关键词搜索。