supports the investigation of Hypothesis 1. The section is based on Table 8 contain-ing valuation errors and Table 9 containcontain-ing significance tests.
Hypothesis 1:
H0: The selection of peer groups based on fundamental value drivers does not affect the valuation error compared to traditional industry peer groups.
H1: The selection of peer groups based on fundamental value drivers reduces the valuation error compared to traditional industry peer groups.
In terms of EV/Sales, it appears that ‘Profitability’ is more accurate than ‘Industry’.
‘Profitability’ has a median of 0.235 compared to ‘Industry’ with 0.376, a difference of 0.141. On the other hand, ‘Industry’ is more accurate than ‘Growth’, ‘Risk’, and
‘Fundamentals’. Furthermore, it appears that ‘Growth’ and ‘Risk’ are significantly less accurate than ‘Fundamentals’. With the U-test ‘Profitability’ is significantly more accurate than all the other methods on a 1% test level. ‘Industry’ is once again significantly more accurate on a 1% test level than ‘Growth’, ‘Risk’, and
‘Fundamentals’.
For EV/EBITDA it appears that ‘Industry’ is significantly more accurate than each of the fundamentals-based selection methods. The median for ‘Industry’ is 0.232, which is 0.01 better than ‘Fundamentals’ of 0.242. Looking separately at the val-uation errors, it could be expected that there would be no significance since the difference is limited. However, due to the low IQ range for both ‘Industry’ and
‘Fundamentals’, the result is significant at a 1% level.
Table 8: Valuation Errors for Industry compared to Fundamentals
Industry Profitability Growth Risk Fundamentals EV/Sales
Median 0.376 (2) 0.235 (1) 0.539 (5) 0.529 (4) 0.528 (3) Mean 0.496 (2) 0.317 (1) 0.735 (5) 0.725 (4) 0.713 (3) IQ Range 0.420 (2) 0.292 (1) 0.469 (4) 0.469 (5) 0.453 (3)
Observations 8,320 9,829 9,950 9,640 9,359
EV/EBITDA
Median 0.232 (1) 0.285 (5) 0.263 (3) 0.275 (4) 0.242 (2) Mean 0.309 (1) 0.369 (5) 0.354 (3) 0.359 (4) 0.320 (2) IQ Range 0.294 (1) 0.327 (5) 0.323 (3) 0.327 (4) 0.297 (2)
Observations 8,136 9,717 9,826 9,585 9,306
EV/EBIT
Median 0.236 (4) 0.230 (3) 0.251 (5) 0.227 (2) 0.201 (1) Mean 0.325 (4) 0.303 (3) 0.349 (5) 0.300 (2) 0.270 (1) IQ Range 0.318 (4) 0.289 (3) 0.325 (5) 0.288 (2) 0.266 (1)
Observations 7,990 9,584 9,690 9,644 9,362
P/E
Median 0.255 (1) 0.306 (4) 0.293 (3) 0.307 (5) 0.264 (2) Mean 0.361 (2) 0.405 (3) 0.409 (4) 0.414 (5) 0.359 (1) IQ Range 0.344 (2) 0.358 (3) 0.365 (4) 0.371 (5) 0.336 (1)
Observations 9,777 11,464 11,807 11,563 11,282
P/B
Median 0.393 (4) 0.319 (2) 0.469 (5) 0.325 (3) 0.274 (1) Mean 0.500 (4) 0.422 (3) 0.592 (5) 0.410 (2) 0.366 (1) IQ Range 0.443 (4) 0.368 (3) 0.489 (5) 0.363 (2) 0.340 (1)
Observations 10,362 12,147 12,273 11,959 11,661
Note: Table 8 reports the median, mean, and IQ range for the each of the multiples EV/Sales, EV/EBITDA, EV/EBIT, P/E, and P/B based on the peer group selection methods Industry, Profitability, Growth, Risk and all Fundamentals combined. All values are horizontally ranked from from 1 to 5 with the lowest value having the lowest rank.
Table 9: Significance Test for Industry compared to Fundamentals Industry Profitability Growth Risk Fundamentals EV/Sales
Industry - ** + ** + ** + **
Profitability + ** + ** + ** + **
Growth - ** - ** -
-Risk - ** - ** + *
-Fundamentals - ** - ** +
-EV/EBITDA
Industry + ** + ** + ** + *
Profitability - ** - * - - **
Growth - ** + ** + - **
Risk - ** + - * - **
Fundamentals - ** + ** + ** + **
EV/EBIT
Industry - ** + ** - ** - **
Profitability + ** + ** - - **
Growth - ** - ** - ** - **
Risk + ** + + ** - **
Fundamentals + ** + ** + ** + **
P/E
Industry + ** + ** + **
-Profitability - ** + + - **
Growth - ** + + - **
Risk - ** - - ** - **
Fundamentals - + ** + ** + **
P/B
Industry - ** + ** - ** - **
Profitability + ** + ** - - **
Growth - ** - ** - ** - **
Risk + ** - + ** - **
Fundamentals + ** + ** + ** + **
Note: Table 9 reports the significance between Industry, Profitability, Growth, Risk, and all Fundamentals combined. The table contains both the Mann-Whitney U test and T U test. The lowest triangle contains the Mann-Whitney U test (black figures) and upper triangle contains t-test (Grey figures). A ”+” indicates that the method in the row is more accurate than the method in the column.
* significance at 5% level. ** significance at 1% level.
In terms of EV/EBIT, it appears that ‘Fundamentals’ is significantly more accurate than all other methods considered in this section. The median for ‘Fundamentals’
is 0.201, which is 0.035 better than ‘Industry’ with 0.236. Furthermore, it is worth noticing that both ‘Profitability’ and ‘Risk’ are significantly more accurate than
‘Industry’.
Even though ‘Industry’ has a lower median than ‘Fundamentals’, there is no statisti-cal evidence that ‘Industry’ is more accurate than ‘Fundamentals’ in terms of P/E.
The explanation behind this can be found in the IQ range showing that the me-dian for ‘Industry’ is more volatile than the meme-dian for ‘Fundamentals’. It appears that ‘Industry’ is significantly more precise than all of the single factor fundamental methods.
For P/B, it is observed that ‘Fundamentals’ is significantly more accurate than
‘Industry’. ‘Fundamentals’ is best followed by ‘Profitability’. The median for ‘Fun-damentals’ is 0.274, which is 0.119 better than ‘Fun‘Fun-damentals’ with 0.393.
The fundamentals-based methods are able to identify a greater number of observa-tions compared to ‘Industry’. The greater number of observaobserva-tions for fundamentals-based methods is reasoned in the minimum of five peers for the ‘Industry’ method, whereas ‘Fundamentals’ only is limited by the negative value drivers and missing selection variables. The lowest number of observations is found for EV/EBIT for ‘In-dustry’ with 7,990 observations corresponding to approximately 726 observations a year. For ‘Fundamentals’ the lowest number of observations is found in EV/EBITDA with 9,302 observations corresponding to approximately 846 observations a year. At first, it seems strange that EV/Sales and EV/EBITDA have fewer observations than EV/EBIT applying a fundamentals-based method. However, this can be explained by the fact that EBIT acts as input for the calculation of selection variables applied for the EV-based multiples concerning the SARD approach. See Table 5 inSection 4.4 for further details.
Conclusion - Hypothesis 1
The results are generally dependent on the multiples. It can be concluded that the overall lowest valuation error can be found for EV/EBIT where ‘Fundamentals’ is significantly lower than ‘Industry’. The lowest median valuation error for ‘Industry’
is found using EV/EBITDA, which is the second-lowest valuation error overall and is significantly more accurate than ‘Fundamentals’.
For all multiples except EV/Sales, it appears that the valuation error is reduced significantly when combining profitability, growth, and risk compared to using them independently.
There is enough evidence on a 1% level of significance that H0 can be rejected for all multiples except P/E when using ‘Fundamentals’. However, only EV/EBIT and P/B supportsH1 that ‘Fundamentals’ is more accurate than ‘Industry’.
There is not enough evidence to reject H0 for P/E, as the performed test does not prove any significant difference between ‘Industry’ and ‘Fundamentals’.
For Hypothesis 1 it is concluded that there is no clear pattern supporting either
‘Industry’ or ‘Fundamentals’ in regard to the accuracy of valuations.
5.2.2 The Combination of Industry and Fundamentals
This section analyses the difference in valuation accuracy between the industry-based and fundamentals-industry-based methods individually compared to the combination of industry and fundamentals, which supports the investigation of Hypothesis 2.
The section is constructed based on Table 10, which contains the valuation errors, and Table 11, which contains significance tests.
Hypothesis 2:
H0: The selection of peers within industries based on fundamental value drivers does not affect the valuation errors compared to traditional industry peer groups and peers groups based on fundamental value drivers.
H1: The selection of peers within industries based on fundamental value drivers reduces the valuation errors compared to traditional industry peer groups and peers groups based on fundamental value drivers.
When using EV/Sales for the valuations, it appears that combining industry and fundamentals (‘Industry & Fundamentals’) yields significantly more accurate valu-ations than using the industry-based or fundamentals-based approach individually.
The median valuation error improved from 0.376 to 0.322, a difference of 0.054, compared to industry, which is the most precise alternative under consideration.
However, ‘Industry & Fundamentals’ is still less effective than ‘Profitability’ on its own as found inSection 5.2.1.
In terms of EV/EBITDA, the ‘Industry & Fundamentals’ peer selection method also decreases the valuation errors significantly compared to both ‘Industry’ and ‘Fun-damentals’ according to both the U-test and the t-test. The median for ‘Industry’ is 0.218 compared to 0.232 and 0.242 for ‘Industry’ and ‘Fundamentals’, respectively.
Table 10: Valuation Error for Industry and Fundamentals compared to Industry &
Fundamentals
Industry &
Industry Fundamentals Fundamentals
EV/Sales
Median 0.376 (2) 0.528 (3) 0.322 (1)
Mean 0.496 (2) 0.713 (3) 0.436 (1)
IQ Range 0.420 (2) 0.453 (3) 0.397 (1)
Observations 8,320 9,359 7,484
EV/EBITDA
Median 0.232 (2) 0.242 (3) 0.218 (1)
Mean 0.309 (2) 0.320 (3) 0.292 (1)
IQ Range 0.294 (2) 0.297 (3) 0.284 (1)
Observations 8,136 9,306 7,429
EV/EBIT
Median 0.236 (3) 0.201 (1) 0.204 (2)
Mean 0.325 (3) 0.270 (1) 0.284 (2)
IQ Range 0.318 (3) 0.266 (1) 0.291 (2)
Observations 7,990 9,362 7,491
P/E
Median 0.255 (2) 0.264 (3) 0.239 (1)
Mean 0.361 (3) 0.359 (2) 0.329 (1)
IQ Range 0.344 (3) 0.336 (2) 0.315 (1)
Observations 9,777 11,282 9,097
P/B
Median 0.393 (3) 0.274 (1) 0.303 (2)
Mean 0.500 (3) 0.366 (1) 0.411 (2)
IQ Range 0.443 (3) 0.340 (1) 0.396 (2)
Observations 10,362 11,661 9,544
Note:Table 10 reports the median, mean, and IQ range for the each of the multiples EV/Sales, EV/EBITDA, EV/EBIT, P/E, and P/B based on the peer group selection methods Industry, Fundamentals, and Industry & Fundamentals combined. All values are horizontally ranked from from 1 to 5 with the lowest value having the lowest rank.
Table 11: Significance Test for Industry and Fundamentals compared to Industry &
Fundamentals
Industry &
Industry Fundamentals Fundamentals EV/Sales
Industry + ** - **
Fundamentals - ** - **
Industry & Fundamentals + ** + **
EV/EBITDA
Industry + * - **
Fundamentals - ** - **
Industry & Fundamentals + ** + **
EV/EBIT
Industry - ** - **
Fundamentals + ** + **
Industry & Fundamentals + **
-P/E
Industry - - **
Fundamentals - - **
Industry & Fundamentals + ** + **
P/B
Industry - ** - **
Fundamentals + ** + **
Industry & Fundamentals + ** - **
Note: Table 11 reports the significance between Industry, Fundamentals, and Industry &
Fundamentals combined. The table contains both the Mann-Whitney U test and t-test. The lowest triangle contains the Mann-Whitney U test (black figures) and upper triangle contains t-test (Grey figures). A ”+” indicates that the method in the row is more accurate than the method in the column.
* significance at 5% level. ** significance at 1% level.
P/E also illustrates a significant increase in accuracy (i.e. a decrease in valuation errors) when combining the approaches. Both the U-test and the t-test support this.
The median valuation error decreases from 0.255 to 0.239 compared to ‘Industry’, which is the most accurate of the two approaches independently. As found in Sec-tion 5.2.1, there is no significant difference between ‘Industry’ and ‘Fundamentals’.
The results for EV/EBIT are different. The median valuation error of ‘Industry
& Fundamentals’ is slightly higher than for ‘Fundamentals’ (0.204 compared to 0.201). The U-test shows no significant difference between the two. However, one should note that the t-test shows significance at the 1% level. Both methods show significantly better results than the ‘Industry’ method.
Using P/B as the valuation multiple, the median valuation error for ‘Industry &
Fundamentals’ is 0.029 higher than for ‘Fundamentals’ (0.303 vs. 0.274). There is statistical evidence at the 1% significance level that this difference is significant according to both the U-test and the t-test. On the other hand, there is also statistical evidence at the 1% significance level that ‘Industry & Fundamentals’
is more accurate than ‘Industry’.
It should be noted that the number of observations is lowest for the ‘Industry & Fun-damentals’ approach. This is natural since the method both excludes observations with less than five industry peers and observations with a missing fundamental se-lection variable, whereas each of the other methods only excludes observations based on one of the requirements.
Conclusion - Hypothesis 2
For all multiples except EV/EBIT and P/B, it appears that the valuation accu-racy is improved significantly when combining ‘Industry & Fundamentals’ rather than using the industry-based or fundamentals-based method independently. For EV/EBIT, there is no significant difference between ‘Industry & Fundamentals’ and
‘Fundamentals’. However, ‘Industry & Fundamentals’ are significantly more precise
than ‘Industry’. In terms of P/B, ‘Industry & Fundamentals’ is significantly more precise than ‘Industry’ but less precise than ‘Fundamentals’.
Based on the U-test, there is enough evidence on a 1% level of significance thatH0 can be rejected for EV/Sales, EV/EBITDA, P/E, and P/B. However,H1 can only be accepted for EV/Sales, EV/EBITDA, and P/E because P/B’s median valuation error for ‘Industry & Fundamentals’ is significantly lower than for ‘Industry’.
Based on the U-test, there is not enough evidence on a 5% level of significance that H0 can be rejected for EV/EBIT due to insignificance between ‘Fundamentals’ and
‘Industry & Fundamentals’ for this multiple.
For Hypothesis 2 it is concluded that there is a trend towards supporting ‘Industry
& Fundamentals’ over ‘Industry’ and ‘Fundamentals’ in regard to the accuracy of valuations.
5.2.3 Industry and Fundamentals in an Intra-Regional Setting
This section analyses the valuation accuracy of the industry-based method in a global relative to a regional setting, which supports the investigation of Hypothe-sis 3. Furthermore, it analyses the valuation accuracy of the fundamentals-based method in a global relative to a regional setting, which supports the investigation of Hypothesis 4. The section is constructed based on Table 12, which contains valuation errors, and Table 13, which contains significance tests.
Hypothesis 3:
H0: The selection of peers within regions based on the industry approach does not affect the median valuation error compared to a cross-regional setting.
H1: The selection of peers within regions based on the industry approach reduces the median valuation error compared to a cross-regional setting.
Hypothesis 4:
H0: The selection of peers within regions based on fundamental value drivers (i.e.
profitability, growth, and risk in combination) does not affect the median valuation error compared to a cross-regional setting.
H1: The selection of peers within regions based on fundamental value drivers (i.e.
profitability, growth, and risk in combination) reduces the median valuation error compared to a cross-regional setting.
Table 12: Valuation Error for Industry and Fundamentals compared to a Regional Setting
Region & Region &
Industry Industry Fundamentals Fundamentals EV/Sales
Median 0.376 (2) 0.324 (1) 0.528 (2) 0.270 (1)
Mean 0.496 (2) 0.477 (1) 0.713 (2) 0.367 (1)
IQ Range 0.420 (2) 0.416 (1) 0.453 (2) 0.339 (1)
Observations 8,320 4,033 9,359 9,359
EV/EBITDA
Median 0.232 (2) 0.221 (1) 0.242 (2) 0.235 (1)
Mean 0.309 (2) 0.295 (1) 0.320 (2) 0.311 (1)
IQ Range 0.294 (2) 0.283 (1) 0.297 (2) 0.294 (1)
Observations 8,136 3,896 9,306 9,306
EV/EBIT
Median 0.236 (2) 0.209 (1) 0.201 (1) 0.202 (2)
Mean 0.325 (2) 0.301 (1) 0.301 (2) 0.272 (1)
IQ Range 0.318 (2) 0.298 (1) 0.266 (2) 0.258 (1)
Observations 7,990 3,838 9,362 9,362
P/E
Median 0.255 (2) 0.236 (1) 0.264 (2) 0.254 (1)
Mean 0.361 (2) 0.349 (1) 0.359 (2) 0.349 (1)
IQ Range 0.344 (2) 0.323 (1) 0.336 (2) 0.330 (1)
Observations 9,777 5,211 11,282 11,282
P/B
Median 0.393 (2) 0.350 (1) 0.274 (2) 0.271 (1)
Mean 0.500 (2) 0.453 (1) 0.366 (2) 0.361 (1)
IQ Range 0.443 (2) 0.417 (1) 0.340 (2) 0.336 (1)
Observations 10,362 5,495 11,661 11,661
Note:Table 12 reports the median, mean, and IQ range for the each of the multiples EV/Sales, EV/EBITDA, EV/EBIT, P/E, and P/B based on the peer group selection methods Industry, Region & Industry, Fundamentals, and Region & Fundamentals. All values are horizontally ranked from from 1 to 5 with the lowest value having the lowest rank.
Table 13: Significance Test for Industry and Fundamentals compared to a Regional Setting
Region & Region &
Industry Industry Fundamentals Fundamentals EV/Sales
Industry - + ** - **
Region & Industry + ** + ** - **
Fundamentals - ** - ** - **
Region & Fundamentals + ** + ** + **
EV/EBITDA
Industry - * + * +
Region & Industry + ** + ** + *
Fundamentals - ** - **
-Region & Fundamentals - - ** + **
EV/EBIT
Industry - ** - ** - **
Region & Industry + ** - ** - **
Fundamentals + ** + ** +
Region & Fundamentals + ** + ** +
P/E
Industry - * -
-Region & Industry + ** + * +
Fundamentals - - **
-Region & Fundamentals + - ** + *
P/B
Industry - ** - ** - **
Region & Industry + ** - ** - **
Fundamentals + ** + **
-Region & Fundamentals + ** + ** +
Note: Table 13 reports the significance between Industry, Region & Industry, Fundamentals, and Region & Fundamentals. The table contains both the Mann-Whitney U test and t-test. The lowest triangle contains the Mann-Whitney U test (black figures) and upper triangle contains t-test (Grey figures). A ”+” indicates that the method in the row is more accurate than the method in the column.
* significance at 5% level. ** significance at 1% level.
Based on the performed U-tests as illustrated in Table 13 there are on a 1% test level significantly more accurate valuations for ‘Region & Industry’ in comparison to ‘Industry’, which refer to industry in a cross-regional setting. This supports Hypothesis 3.
For EV/Sales the valuation error is improved with 0.052 percentage points to a median of 0.324. In relation to EV/EBITDA, it is improved with 0.011 percentage points to a median of 0.221. Furthermore, EV/EBIT is improved with 0.027 per-centage points to a median of 0.209. P/E is improved with 0.019 perper-centage points to a median of 0.236. Lastly, P/B is improved with 0.043 percentage points to a median of 0.350.
The t-test supports the results from the U-test for all multiples except EV/Sales as of the t-test does not show a significant difference. It should be noted that both EV/EBITDA, P/E, and P/B have only proved significant on a 5% test level in the t-test.
Adding the regional restriction for peer group selection to the fundamentals-based method gives a significantly more accurate result for EV/Sales (at the 1% level).
It improves the valuation error from 0.528 to 0.270 and further reduces the IQ range from 0.453 to 0.339, which indicates less volatility in the result for ‘Region &
Fundamentals’.
When evaluating EV/EBITDA, it appears that ‘Region & Fundamentals’ is sig-nificantly more accurate than ‘Fundamentals’ at the 1% level. This once again supports Hypothesis 4. The median for ‘Region & Fundamentals’ is 0.235, which is 0.007 percentage points better than the median for ‘Fundamentals’ of 0.242. This is, however, not supported by the t-test: it does not show any significant difference between ‘Region & Fundamentals’ and ‘Fundamentals’. This supports Hypothesis 4. It is worth noticing that when applying a regional restriction, the significant result supporting ‘Industry’ over ‘Fundamentals’ from Section 5.2.1 now become insignificant to ‘Region & Fundamentals’ – not even at a 5% level of significance.
Looking at EV/EBIT, it appears that ‘Region & Fundamentals’ reduces accuracy when compared to ‘Fundamentals’. The median for ‘Region & Fundamentals’ is 0.202, which is 0.001 higher than the median for ‘Fundamentals’ of 0.201. However, the difference in medians is not significant, and therefore it does not support Hy-pothesis 4. It is noticeable that the t-test concludes that ‘Region & Fundamentals’
is more accurate than ‘Fundamentals’. This is the opposite of the U-test, but this result is still not statistically significant.
In terms of P/E, it appears that ‘Region & Fundamentals’ is significantly more accu-rate than ‘Fundamentals’ at the 5% level. The median for ‘Region & Fundamentals’
is 0.254, which is 0.01 lower than the median for ‘Fundamentals’ of 0.264. It should be noted that the t-test does not support the significant difference between ‘Region
& Fundamentals’ and ‘Fundamentals’. It is worth noticing that when applying a regional restriction, the insignificant difference between ‘Industry’ and ‘Fundamen-tals’ from Section 5.2.1becomes significant in favour of ‘Region & Fundamentals’
relative to ‘Region & Industry’.
Moving on to the last multiple, P/B, there is no significant difference between ‘Re-gion & Fundamentals’ and ‘Fundamentals’. The median for ‘Re‘Re-gion & Fundamen-tals’ is 0.271, which is 0.003 lower than the median for ‘FundamenFundamen-tals’ of 0.274. The t-test supports the U-test: both reveal that the median valuation error for ‘Region
& Fundamentals’ lower than for ‘Fundamentals’ but without any significance.
Concerning observations, there are no fluctuations between ‘Region & Fundamen-tals’ and ‘FundamenFundamen-tals’. This is due to the fact that the algorithm requires at least 10 other companies within a region which is met for all regions. When look-ing at observations for the industry-based method in a regional settlook-ing, it is clear that the number of observations drops significantly. The reason behind this can be found in the natural limitation of the sample in which 1,200 companies are di-vided into five different regions each year; this decreases the likelihood of fulfilling the requirement of five peers. The lowest number of observations can be found for
EV/EBIT with 3,825, which corresponds to approximately 348 observations a year.
The implications of this are discussed in Section 6.2.
Conclusion - Hypothesis 3
There are significant improvements when adding the regional factor to the industry-based method when compared to the cross-regional setting for all multiples. The t-test supports the U-test for all multiples except EV/Sales where the t-test does not show any significance.
There is enough evidence on a 1% test level of significance that H0 can be rejected for all multiples. Thus, H1 is likely to be true on a 1% level.
For Hypothesis 3 it is concluded that there is substantial evidence supporting ‘Region
& Industry’ over ‘Industry’ concerning the accuracy of valuations.
Conclusion - Hypothesis 4
It can be concluded that adding a regional factor to the fundamentals-based method improves the accuracy significantly except for EV/EBIT and P/B. Furthermore, there is a pattern of ‘Region & Fundamentals’ being the most accurate when analysing the ranks in Table 12. This is supported by Table 13, which illustrates significantly more accuracy for EV/Sales, EV/EBITDA, and P/E.
However, performing the t-test for EV/EBITDA, EV/EBIT, P/E, and P/B shows no significant difference between ‘Region & Fundamentals’ and ‘Fundamentals’. This paints the opposite picture than the results from the U-test. For EV/Sales both the U-test and the t-test show a significant deviation from ‘Fundamentals’.
There is enough evidence on a 1% level of significance that H0 can be rejected for EV/Sales and EV/EBITDA, and P/E at a 5% level. In all three cases H1 can be accepted.
There is not enough evidence to rejectH0 for EV/EBIT and P/B as the performed
test does not prove any significant difference between ‘Fundamentals’ and ‘Region
& Fundamentals’.
For Hypothesis 4 it is concluded that there extensive evidence supporting ‘Region
& Fundamentals’ over ‘Fundamentals’ in regard to the accuracy of valuations.
5.2.4 Fundamentals in an Intra-Regional and Industry-specific Setting
This section analyses the difference in valuation accuracy between the fundamentals-based method when industry and regional restrictions are imposed simultaneously and pairwise combinations of industry-based, region-based, and fundamentals-based methods. This analysis supports the investigation of Hypothesis 5. The section is constructed based on Table 14, which contains valuation errors, and Table 15, which contains significance tests.
Hypothesis 5:
H0: Combining industry, region, and fundamentals (i.e. profitability, growth, and risk in combination) does not affect the median valuation error.
H1: Combining industry, region, and fundamentals (i.e. profitability, growth, and risk in combination) reduces the median valuation error.
In terms of EV/Sales, ‘Region, Industry & Fundamentals’ increases the median val-uation error compared to ‘Region & Fundamentals’ from 0.270 to a median of 0.294, an increase of 0.024. Based on both the U-test and the t-test, there is statistical evidence at the 1% level that this difference is significant. On the other hand, ‘Re-gion, Industry & Fundamentals’ are significantly more accurate than both ‘Industry
& Fundamentals’ and ‘Region & Industry’.