url
29
Apr
2014

ON THE DETERMINANTS OF DERIVATIVE HEDGING BY INSURANCE COMPANIES: Results of Participation Extent Model

All variables employed in the participation decision (probit) model are also included in the participation extent of use model. The results of the participation extent analysis for the life and non-life insurance sectors are given in Tables VI and VII. In comparison with the probit model, there are fewer significant variables for both insurance sectors. This corresponds to the empirical studies carried out by Hoyt and Colquitt in the analysis of hedging behavior among life insurers and by Hardwick and Adams in the analysis of derivative use among life insurers in the United Kingdom. To analyze the determinant variables in the extent of use model, we focus mainly on the pooled estimation results. Foreign exposure is found to be an influential factor on the level of derivative use for both life and non-life insurers, indicating that insurance companies are inclined to engage in currency risk management with derivative instruments.
The result that the current ratio is positively related to the level of derivative use among life insurance firms is unexpected. It contradicts the argument that firms with higher financial limitations will hedge through derivatives (Geczy, Minton, and Schrand, 1997).
The coefficient for the domestic dummy variable is positive and significant for the life insurers. This finding reflects the current circumstances of the finance industry. Foreign branches in Taiwan tend not to hedge using derivatives for two reasons. First, due to the fact that the financial market used to be oligopolic and legally protected, market channels and related expertise have been established by native enterprises. Subsidiaries of most foreign firms are limited to a small number of businesses with relatively low currency exposure. Consequently, the need for foreign currency trading in these branches is lower than in domestic companies. Second, the small market size in Taiwan discourages foreign companies from expanding the scope of their business so the exchange risk is even lower and there is little need for derivative hedging.
The anticipated substitute relationship between reinsurance and derivative hedging is confirmed for the non-life insurance sector. This reflects the fact that catastrophic futures and options are being used increasingly as a substitute for conventional risk management methods in that sector.
Although the coefficient of the interest rate risk variable (represented by net interest margin) is negative and significant for the non-life insurance sector, the unexpected sign might imply that in practice, non-life insurers focus more on volatility of interest cost rather than interest income. This finding also suggests that non-life insurers are trying to prevent interest costs from rising by means of derivatives. Haushalter indicates that the factors affecting the likelihood of employing derivative instruments are not necessarily the same as those affecting the extent of use. The discrepancy between the two estimations would be hidden if the tobit approach is used to measure the hedging activities (Colquitt and Hoyt, 1997; Cummins, Phillips and Smith, 2001). As to the effect of size in the life and non-life insurance sectors, there are inconsistent outcomes reported in the participation and extent models. this

Table VI Participation Extent Model for Life Insurance Sector

Variable ExpectedSign 2001-2003OLS 2001OLS 2002OLS 2003OLS
Coefficient r-statistic Coefficient r-statis-tic Coefficient г-statistic Coefficient r-statistic
Consent 0.1153 09120 -0.0058 -0.0739 -0.1886 -1.2601 0.2946 -016
SIZE -5- -0.00» -12910 -0.0016 -03176 0.0037 0.3562 -0.0007 -0.0259
HN -3- 0.0000 0.4070 0.0000 0.0279 0.0000 2.0S22 0.0000 0.0920
MISAST -0.0014 -0.4370 0.0065 0.3773 0.00 IS 0.5318 -0.0061 -0.6331
MISUA -0.0009 -12800 -0.0001 -0.1621 -0.0008 -0.986 -0.0027 -0.4746
GROWTH + -0.0005 -0.8510 0.0000 0.0860 0.0007 0.6388 -0.0015 -05 8
FEINS 0.0010 05250 0.0008 1341В 0.0016 1.1327 -0.0056 -1.1811
TLCF -0.0010 -0.7350 -0.0006 -1.4798 -0.0023 -0.9757 -0.0332 -05499
CH 0.0431 1.3110 0.0232 12175 -0.0356 -1.1101 0.1563 0.8219
MH -3- 2.5405 0.8490 1.0248 0.6857 -12459 -0.4180 -1.8688 -0.1451
FX -3- 0.6601 5.0380 02685 2.0705 0.6431 2.7943
0.5551 12058
NTM 0.0319 03900 0.0213 0:9197 -0.0289 -0.0537 -2.2562 -05226
ENV +/— 0.0012 15800 0.0426 03032 0.0013 1.3686 0.0014 05770
CR 0.0028 3 3660 0.0013 3.6132 0.0009 0.7226 0.0038 15328
Variable ExpectedSign 2001-2003OLS 2001OLS 2002OLS 2003OLS
CodBciaA r-statistic Coefficient r-statistic G&sfficfent r-statistic Coefficient r-statistic
DOMESTIC -s 0.0715 2.7090″ 0.0269 1.7691 0.0102 03630 02732 2.3561
HERFC -0.1323 -1.5930 -0.0629 -09967 0.2514 23226″ -0.7922 -2.2881
IR + 0.0038 1.0620 0.0020 1.1083 -0.0048 -0.6418 0.0327 1.113S
Number of observations 79 27 28 24
adjusted R 0.4244 05031 0.4442 0.3548
Ftest 4.3600[0.0000] 25 188” [0.0938] 224S5 [0.1096] 1.7561[0.2511]
LMtest 1.4800[0.2237]
Hrisnun test 4.2500[0.9984]
White test 1.1743 [0.3103]

Table VII Participation Extent Model for Non-Life Insurance Sector

Variable ExpectedSign 2002-2003OLS 2002OLS 2003OLS
Coefficient r-sta tistic Coefficient r-statistic Coefficient r-statistic
Const*nt 0.0441 0.6330 -0.1068 -1.0357 -0.8340 -1.7776
SIZE -0.0005 -0.1140 0.0042 0.7375 0.0456 1.7161
FTN 0.0000 0.6510 0.0000 1.2374 0.0000 1.1046
MISAST 0.0000 02250 0.0007 1.4935 -0.0009 -1.169P
MISL1A 4).00£S -0.8810 0.0004 0.9275 0.0012 0.7419
GROWTH -0.0007 -0.7980 0.0003 0.1956 0.0037 1.4964
REINS -0.0006 -25780
0.0002 0.5071 0.0015 159 85
TLCF 0.0013 0.7950 0.0013 0.3500 0.0414 2.4778
CH 0.0046 0.3120 0.0283 1.3605 -0.06 -1.1455
MH 0.0042 0.1560 0.0282 0.8030 0.1512 1.4728
FX 0.0070 69680 -0.0011 -0.5731 0.0064 23268
KIM -0.8307 -4S400
0.1125 0.1810 5.2769 22236
CR 0.0005 03660 0.0016 0.6096 0.0037 0.6168
DOMESIIC 1 -o.oos -05090 0.0011 0.0658 -0.0430 -0.6533
HEEPC ~ ^0.0003 -0.1640 -0.0060 -1.5209 0.0081 0.1038
Number cf observations 71 24 24
adjusted R: 0.4460 0.0571 0.6156
Ftest 4.7400[0.0000] 1.0909[0.4779] 3.4022[0.0548]
LM test 25800[0.1082]
Hauanantea 0.0100[1.0000]
White test 2.4235[0.0060]

All variables employed in the participation decision (probit) model are also included in the participation extent of use model. The results of the participation extent analysis for the life and non-life insurance sectors are given in Tables VI and VII. In comparison with the probit model, there are fewer significant variables for both insurance sectors. This corresponds to the empirical studies carried out by Hoyt and Colquitt in the analysis of hedging behavior among life insurers and by Hardwick and Adams in the analysis of derivative use among life insurers in the United Kingdom. To analyze the determinant variables in the extent of use model, we focus mainly on the pooled estimation results. Foreign exposure is found to be an

About The Author

Kevin J. Brandon

Home | Site Map | Contacts

Copyright © 2013 - 2019 Investment And Finance Online. All rights reserved