Table 8
Multinomial Logit Model Estimates for Liquidation and Within-Family Merger
A
OThWhnte hed
c n 咐 m
uem-otbh,uee個tcrs
ofdamoEIMfne olmybd -suel(lerulvv)naeotll simqoWunas ie dl aalarOte le Bso 2Itth6ime，n3c。3a ld7
umedl dew,d (aM2s )ua d sewSd IzzttOeh，EIEnaF阱vfea， sIItanIgnly
ad te mumtuhqte uuadel znshetisnns cd tm lomn tehmre ge aonnaTg lhydeSISvmab叫 nreanb戶 t
lees m,Itnafcno ldu rmd川e s
d nTMdhlcead hte nnuSd Imgfnmbmer1sml,y nct eatan t祖dth l sse
eptooins 1eoe，w恤自s珊，das
,nbd ay npd tehtdreen oe mpdemrocleccne ets,at r{e0tsh)pekecehetwzpIne山 l ly y
e ,Variables Modell Modelll
~血110n Within-Mer島er LiquidatÎon Wìthin-Merger
Estimates Chi-square Odds Estimates Chi-square Odds Estimates Chi-square Odds ratio Estimates Chi-squar Odds ratio
raUo raUo e
Intercept -2.717 。 974 3.303 1.047 1.904 。863 -0.657 0.078
F陶M…… Ma 目 a咽nl耐叫 flow叫{LLI心2B
I) >叫
0.003 0.994 -0.094 l962Family 0L9elvo el-1.767' 3.146 。 171 。 556 0.481 。573 -0.964... 15.955 。 381 。 189 0.352 0.828
1.020 0.498 2.773 1.833 1.049 6.253
-0.176 0.373 。 839 。 266 1.473 。 767
5.144 1.048 0.006 8.609 1.519 2.385
FN陶P…
MMU叩叫向nL喇臨 M 阿時llη
12)l))叫
l蝴
2.219 。 972 0.005 O13l ObJecuvo e 9L9e5 vel 4eH 、,.0.642 。 200 -0.594 0.151 。 552
H
H h
-4.572 。 337 0.010 7.174 0.834 0.001
-0.044 。 437 0.957 0.057 。301 。941
Ni b
Z 喜 E
-4.606 。 695 0.010 5.575 0.433 。∞4
Fund Level
SNNA…，
EUNMaUUm咽且uhm叩呵a頁 t)臨)
mummalMh叫 E t叫 4ω 叫 I(t月
21) ) 4 253 • M ? 0.014 。 943 1.209 。 389 -2.918... 13.920 。 051 -1.412 2.072 。244。 406." 16.381 0.666 -0.547'" 13.622 。 579 -0.350". 20.310 。 704 -0.469". 17.620 。 626 -0.111'" 62.729 0.895 0.006' 3.038 。 994 -0.094"
.
81.242 。 910 0.004 1.752 。 996-0.035" 5.841 0.966 -0.006 。 576 0.994
-0.048... 6.965 。 951 -0.011 0.496 0.989
-0.131... 51.574 0.860 -0.042.. 5.441 。 959 -0.117... 77.515 0.890 -0.043." 8.862 。958
。 108 2.094 1.114 0.095 1.998 1.100
。 681 0.546 1.975 1.664" 5.629 5.280
注足~ 意
b b哥
NH
peoe巾wt
a(s1ry-sn1e)at x nce(1-1) 。 0212.363 1.022 0.012." 8.142 1.012 0.004 0.195 1.004 0.012... 17.431 1.012
nPLeoikrwfeb(lh 2m.2o)a。 x nd
cerz(ut-o 2) test staOuOsOulc-s--36.747 l ∞ l 0.000" 4.791 1.0∞ o.的 1... 52.754 1.001 0.002.. 6.291 l ∞2
主位諸音
1603.97"''' 1562.55...
的PNSNVE〉hEbhg豆、HNmE毛主凶~害怕"川口~~
Table 9
Multinomial Logit Model Estimates for Liquidation and Within-Family Merger: ExcIuding Financial Crisis
A three-outcome multinomial logit model was used 10 investigate the distinction among different 叫itfomlS. The fund family can dispose funds by three choices: (0) keep the fund in the family; (1) liquidate出efund: (2) a within-family mutual fund merge. The variables included fund numbers, n叫刮目tfl側5，and performance 剖thefamily,
。阿ective，and fund level. We also included fund size. age，甜d肌肉ueness田 theanalysis. .**, *., and" indicate significant at the one, five, and ten per臼01，respectively The numbers of observations excluding financial crisis are斜，234 ，respectively
Variables M血lell Modelll
88 The Exit Determinants and Wea/th Effects in the Taiwan Mutua/ Fund Jndustry
Table 10
Performance Changes Following Mutual Fund Mergers
This table pr官sentsthe median values of single司 factoralpha, multifactor alpha and objective-adjusted retums for target and acquiring funds. Objective-adjusted retums are computed as the difference between the funds' annual retum and the average retum on all funds in the same investment objective. We estimated performance for the four years following the fund merger. Year -1 is one-year prior to the m咕咕ermonth, and year -2 is two-year prior to the merger, etc. The changes of year relative to merger represent the difference of performance between po叫-mergerand pre-merger periods for all acquiring funds. The first two p-values represent the significance of the di 仔erence acro俗 thetarget and acquiring funds in the years preceding the merger, and the last two p-values represent the significance of the characteristic variations for acquiring funds following the merger.
Annualized Performance (%)
Y'ear Relative to Merger Chan!!es ofYear
-2 -1 +1 +2 -lto+1 p-values -lto+2 p-values cris的，and confirms the results in Table 10. The target funds performed worse than the acquiring funds during the pre-merge period. The investors of the target funds benefited from the merge activity since their retums increased from year -1 to year + 1 and to year +2, regardless of the retum measure; however, the investors of acqumng funds suffered from deterioration in performance subsequent to the merger.
Chiao Da Management Reνlew均 1.31No. 2, 2011 89
Table 11
Performance Changes Following Mutual Fund Mergers: Excluding the Period of Global Financial Crisis
This table presents the median values of single-factor alpha, multifactor alpha and objective-adjusted retums for target and acquiring funds. Objective-a句 ustedretums are computed as the difference between the funds' annual retum and the average return on all funds in the same investment objective. We estimated performance for the four 抖的following the 臼ndmerger. Year -1 is one-year prior to the merger month, and year -2 is two-year prior to the merger, etc. The changes ofyear relative to merger represent the difference ofperformance between post-merger and pre-merger periods for all acquiring funds. The first two p-values represent the significance of the
di缸訪問cein median across the target and acquiring funds in the years preceding the merger, and the last two p-values represent the significance of the characteristic variations for acquiring funds following the merger.
Annualized Performance (叫
Year Relative to Merger Changes ofYear
-2 +1 +2 -1 to+1 p-values -1 to +2 p-values Panel A: Single-factor Alpha
Thanrgdes t -0.072 -0.185 。 167 0.029 。 138 0.033
Acquiring -0.025 -0.006 -0.009 -0.042 -0.049 。 074 -0.040 0.045 funds
p-values 0.051 0.019
Panel B: Multi-factor Alpha Target
-0.264 -0.358 0.201 0.035 。 298 0.027
funds Acquiring
-0.197 -0.315 -0.147 -0.034 -0.096 0.017 -0.009 0.038 funds
p-values 0.048 0.062
Panel C: Objective-adjusted retum Target
-0.063 -0.032 0.006 0.191 0.033 0.087
funds Acquiring
-0.061 0.166 -0.031 0.002 -0.293 0.008 -0.012 0.041 funds
主values 。 204 0.009
5.2.
Net Asset Flows Surrounding the MergerTable 12 presents the values of net asset flows and objective-adjusted net asset flows for target and acquiring funds before and after the merger. It is found that in the pre-merger period both acquiring and target funds experienced negative net asset flows
,
with the acquiring funds experiencing higher net asset flows than the target funds. In particular, the net asset flows of the acquiring funds in year -190 The Exit Delerminants and Wealth ξtfects in Ihe Taiwan Mutual Fund lndustry
and year -2 were -0.799 percent and -1.153 percent
,
respectively. However,
the net asset flows of the target funds in year -1 and year -2 were -2.326 percent and -2.297 percent, respectively. In addition, in tenns of objective-adjusted net asset flows, both the target funds and the acquiring funds also encountered negative objective-adjusted net asset flows. The significant p-values indicate that the target funds suffered more acute net redemptions than the acquiring funds did before the merger. This phenomenon suggests that the merger may have been motivated by the fund family's managerial strategies. On one hand, the fund family eliminated poor-performing funds by merging; on the other hand, the acquiring funds attracted additional assets by merging. Hence, the fund family gained two advantages by a single move.In the post-merger period, both the net asset flows and the objective-adjusted flows continued to post negative retums of -1.993 percent and -1.362 percent in year + 1, respectively. The negative flows arose from the fact that the investors of acquiring funds redeemed assets from combined funds due to their deterioration in performance. Moreover, though the target fund investors could enjoy benefits together with the acquiring fund investors after the merger, the possibility remained that they might change their investment tactics and chose di仔erentfunds in which to invest subsequent to the merger. This could lead to a pattem of net redemption after the merger. However, the right panel shows that the difference in the net asset flows of acquiring funds between one year before and one year (a two year period) after the merger was an insignificant -1.381 (-0.422) percent.
Similar results were found in terms of the objective-adjusted net asset flows. This indicates that, even though cash kept flowing out the fund after the merge activi旬，
it was insignificant compared to the magnitude of outflows that occurred during the pre-merge period
To avoid the confounding effects of the 2008 global financial crisis, we exc1uded the data from this sample period. As shown in Table 13, both target and acquiring funds suffered from cash outflows within the one-year and two-year period before the merge activity; acquiring funds also experienced insignificant
Chiao Da Managemenl Review 均1.31No. 2, 2011 91
fund outtlows after the merge activity. These results are similar to those including the financial crisis period.
In summary, even though the combined funds suffer a decrease in performance in the post-merger period, the subsequent net outtlow of the combined funds is insignificant. This was consistent with many studi郎，based on both US and Taiwan data, namely that document an asymrnetric relationship between fund performance and asset tlows, with a positive relationship between good performance and subsequent intlows but an insignificant relationship between poor performance and net outtlows (Ippolito, 1992; Chevalier and Ellison, 1997; Sirri and Tufano, 1998; Lin, 2004)
Table 12
Changes in Net Asset Flows around Fund Mergers
This lable presenlS Ihe median values of nel assel flows and objeclive-adjusled nel as臼1flows (as a percenlage) for largel and acquiring funds. Objeclive-adjusted net assel flows are c心mpuledas the di 仟erencebetween the funds' annual net assel flow and Ihe average nel asset flows of all funds in Ihe same investment objeclive. Year -1 is one-year prior 10 the merger month, and year -2 is 山e
second year prior to Ihe merger, elc. The changes of year relative 10 Ihe merger represent the
di仔erenceof net as認1flows between post-merger and the pre-merger periods for all acquiring funds. The first 1\νo p-values represent Ihe significance of the difference in median across Ihe target and acquiring funds in the years preceding Ihe merger, and the last Iwo p-values represenl the significance of characterislics variation for acquiring funds following the merger
Year Relative to Merger Changes ofYear
-2 -1 +1 +2 -lto+1 -110+2
Panel A: Net asset flows
Target funds -2.297 司 2 .3 26
Acquiring funds -1. 153 -0.799 -1.933 -0.972 -1.381 -0.422
p-values 0.000 0.000 0.146 0.747
Panel B: Objeclive-adjusted nel assel flows
Targel funds -2.372 -2.378
Acquiring funds -1.191 -0.862 -1.362 -0.462 -1.449 -0.421
p-values 0.000 。 000 。.156 0.418
92 The Exit Determinants and Wealth Effects in the Taiwan Mutual Fund Jndustry
Table 13
Changes in Net Åsset Flows around Mutual Fund Mergers: ExcIuding the Period of Global Financial Crisis
This t油 lepresents the median values of net asset flows and objective叫Uustednet asset flows (as a percentage) for target and acquiring fu且ds.Objective-adjusted net asset f10ws are computed as 出E
difference between the funds' annual net asset flow and the average net 的setf10ws of all funds in the same investment objective. Year -1 is one-year prior to the merger month, and year -2 is 由e
second year prior to the merger, etc. The changes of year relative to the merger represent the difference of net asset flows between post-merger and the pre-merger periods for all acquiring funds. The first two p-valu閃閃prese叫tthe significance of the difference in median across the target and a心quiringfunds in the years preceding the merger, and the last two p-values represent the significance of characteristics variation for acquiring funds following the merger
Target funds Acquiring funds
企旦控豆豆
Target funds Acquiring funds 乙旦出2
Year Relative to Merger Changes ofYear
-2 - +1 +2 -1 to+1 -1 to司令2
Panel A: Net asset f10ws -0.020 -0.025
-0.007 -0.006 -0.018 -0.017 -0.015 -0.008
0.039 0.033 0.491 0.574
Panel B: Objective-adjusted net asset flows -2.027 -2.500
-0.680 -0.686 -1.489 -1.731 -1.450 -0.918
0.000 0.000 0.102 0.199
5.3.
Other Characteristics Effects Following a MergerTable 14 presents the values of asset size, expense ratio, fund tumover, and units of benefit for target and acquiring funds surrounding the merger. First of al1, in the years preceding the merger, the size of the acquiring funds was larger than that of the target funds. Specifically, the size of the acquiring funds for one year and two years prior to the merger was 478.9 million NT dollars and 709.0 million NT dollars, respectively. However, the corresponding size of target funds was only 313.5 miIlion NT dollars (514.6 million NT dollars) in year -1 (-2).
Furthermore, in comparison of size, the target and acquiring funds in each of the pre-merger years were statistically and significantly different (p-value = 0.000)
In the post-merger period, the size of acquiring funds increased slightly 企om
478.9 to 493.2 (524.3) million NT dollars in year +1 (+2). The difference in the size of acquired funds increased significantly. This increase in size can partly be attributed to the merger ofthe assets oftarget funds. But the redemptions after the merge activity made the post-merger assets less than the combined assets of target
Chiao Da Management Review 均1.31No. 2, 2011 93
and acquiring funds (493.2 < 313.5+478.9). Based on the above results, we infer that an important motivation for a fund merger is to achieve economies of scale.7 Hence, in the material below, we detect the variation in expense ratio in the pre-and post-merger period to determine whether a merger can reduce the expense ratio by achieving economies of scale
Table 14
Chan~es in Other Characteristics Concernin~ Mutual Fund Mer~ers This table presents the median values of assets (NT dollar盲，millio明)，缸mdturnover (percent), and beneficiary(numbers) for target and acquiring funds. We estimate characteristics for the four ye訂S
following the fundmerger. Year -1 is one-year prior to the merge month, and year -2 is tbe second year prior to the merger, etc.The changes of year relative to the merger represent the differences of fund characteristics between thepost-merger and pre-merger periods for all acquiring funds. The first two p-values represent the significance ofthe difference in median across the target and acquiring funds in the years preceding the merger, and the lasttwo p-values represent the Significance of the variable characteristics for a叫 uiringfunds following the merger
Target funds median expense ratios of the acquired funds were 0.149 percent and 0.148 percent, respectively, which were significantly higher than the 0.138 percent and 0.133
7 Due to the lack of confidence in the merger activi吟" some investors withdrew their money from the combined funds. This consequently led to the increase in fund size being moderate after the merger
94 The Exil Delerminanls and Weallh 芝加clSin the Taiwan Mutual Fund Industry
percent expense ratios of the acquiring funds. To investigate if the merger could reduce the expense ratio, we calculated the difference in expense ratios between the pre-merger and post-merger periods. The results show that even though there was no signi ficant change in the expense ratio of the acquiring fund one year after the merger (year + 1) (p-value = 0.308), the expense ratio decreased significantly in the second year after the merger (year +2) (p-value = 0.038). The reduction in the expense ratio of acquiring funds in the second post-merger year suggests that the combined funds gained efficiency and improved significantly after the merger.
The evidence of dec1ine in expense ratios indicates that fund investors would benefit from reducing expenses if a fund family were to achieve the economies of scale via mergers
With respect to fund turnover, target and acquiring funds did not have any significant difference in fund turnover in the one year and two year pre-merger periods. The acquiring funds also did not show any significant changes in turnover ratio between pre-and post-merger periods (p咱value= 0.731 and 0.926).
This suggests that the mutual managers' trading strategies did not change following the merger activity.
Finally, we compared fund beneficiaries between target and acquiring funds before and after the merger. In Table 14, the numbers of beneficiaries of target funds are smaller than those of acquiring funds in the years prior to merging, and the difference across the target and acquiring funds in the pre-merger period is significant, especially in the year before merging (year -1) (p-value < 0.01). The phenomenon of the target funds having fewer beneficiaries could reflect the poor performance of target funds. Fund beneficiaries, in order to protect their own property, will choose to redeem assets when they are conscious ofthe funds' poor performance or become aware of poor fund management.
In the post-merger period, the numbers of beneficiaries of the combined funds increased from 1657 to 2632. The right panel shows that the increase in fund beneficiaries in the years following the merger is significant (p-value 0.000 and 0.063 at years + 1 and +2, respective\y). The increase resulted from the fact that the beneficiaries, originally be\onging to the acquired funds, were shifted to the acquiring funds, and they shared benefits with the beneficiaries of the
Chiao Da Management Review Vol. 31 No. 2. 2011 95 combined funds afterwards because these funds did not perform well subsequent to the merger.
Table 15
Changes in Other Characteristics Concerning Mutual Fund Mergers:
Excludin~ the Period of Global Financial Crisis
This table presents the median values of assets (NT dollars, million), fund tumov釘 (percent)，and beneficiary (numbers) for target and 的quiringfunds. We estimate characteristics for the four years following the fund merger. Year -1 is one-year prior to the merge month, and year -2 is the second year prior to the merg鯽，etc. The changes of year relative to the merger represent the di作erencesof fund characteristics between the post-merger and pre-merger periods for all acquiring funds. The first two p-values represent the significance of the difference in median across the target and
acqu lfI且 g funds in the years preceding the merger, and the last two p-values represent the Significance of the variable characteristics for acquiring funds following the merger.
Target funds Acquiring funds
企旦世E
TAacrqgue1t rhing ridfs unds p-values
Year Relative to Merger Changes ofYear -2 -1 + 1 +2 -1 to + 1 -1 to +2
96 The Exit Delerminanls and Weallh EjJects in the Taiwan Mutual Fund Jnduslry
during the pre-merge period. Through merging other funds' assets, acqumng funds had more fund assets under management and the number of beneficiaries increased from year -1 to year + 1 and to year +2. Acquiring funds also had a decline in expense ratios in year +2, which indicates that the target and acquiring fund investors would benefit from reducing expenses if a fund family were to achieve the economies of scale via mergers.
6. Conclusion
When mutual-fund investors are con企onted with poor performance, they often head for the exits. Increasingly, fund companies are doing the same. They exit by merging weak funds into better performers, or by liquidating a fund's holdings and returning the proceeds to investors. This paper examined the determinants of the mutual fund exit forms, liquidation and a wit血hin-晶mily merger, as well as the subsequent impacts on fund investors after a merger. lt was found that the likelihood of a fund exit was inversely related to fund size, fund performance and fund tlows, regardless of liquidations and mergers. Specifically, we found that a fund family may liquidate a poor-performing and long-term outf1ow fund, but it will merge a poor-performing fund with another one within a family if the 臼nd has experienced only short-term fund outf1ows. This phenomenon may be attributed to the fact that, following a merger, a fund family can retain valuable client sources and distribution channels within the farnily, if a fund with poor performance has short-term asset outf1ows. However, fund outf1ows at the family level only a釘ectthe liquidation decisions
Acquiring fund investors in our study experienced a significant deterioration in performance subsequent to the merger activity. In contrast, the target fund investors appeared to b巴nefit 企om these combinations， 的 their fund's performance improved in the year after the merger. The difference in performance between acquired and acquiring funds suggests some wealth transfer effects from investors of acquiring funds to the target funds. In addition, the net asset f10ws continued to remain negative for the combined fund in the year following the
Chiao Da Management Reνlew 均 1.31 No. 2, 2011 97
merger. This indicates that a merger may not be a remedy to improve the net redemption of acquiring funds unless the performance of combined funds improves after the merger. However, the greater assets under management after a merger prove that the action of merging real\y adds to the economies of scale
After a fund family achieves economies of scale in operations, the expense ratio of combined funds following the merger decreases
7. References
Agrawal, A., Jaff，巴， J. F., and Mandelker, G. N. (1992), “The Post-Merger Performance of Acquiring Firms: A Re-Examination of an Anomaly,"
Agrawal, A., Jaff，巴， J. F., and Mandelker, G. N. (1992), “The Post-Merger Performance of Acquiring Firms: A Re-Examination of an Anomaly,"