Empirical results Directory UMM :Data Elmu:jurnal:M:Multinational Financial Management:Vol10.Issue2.2000:

relationship is tested by using the coefficient of variation of: i sales to market value SALESMV; ii the firm’s book value to its market value BOOKMV; and iii dividend yield. Here, a negative relationship between the degree of utilisation of derivatives and our growth option measures is expected. Following DeMarzo and Duffie 1995, high quality managers are more likely to hedge. But the choice of hedging technique and the type of exposure that is hedged can reflect managers’ perceptions of the economic effects of hedging. Titman 1992 also shows that a firm that has an optimistic outlook can use interest rate swaps to benefit from borrowing and the expected cost of financial distress will not increase. Thus a positive relationship is predicted between the use of cross-currency interest rate swaps and variability of the leverage measures. The tax treatment of both the exposure and the hedging technique can have important implications for the firm’s hedging strategy see Kramer et al., 1993. Under UK tax laws, the use of derivatives to hedge translation exposure may give rise to cash flow gains which are taxable and losses which are not tax allowable see Buckley, 1992. To avoid the adverse impacts of asymmetry in taxation, firms are likely to place more emphasis on internal techniques when hedging translation exposure. The use of forward contracts to hedge the transaction exposure emanating from re6enue transactions, results in taxabletax allowable gains and losses in the UK. All those impacts will in turn affect the level of profitability. Assuming that the tax credits can be utilised, a lower degree of variability is expected on the tax measures for firms that use forward contracts to hedge transaction exposure. The tax and profitability measures are the coefficient of variation of: i tax charge on profitloss to pre-tax profitloss TAXRATIO; ii tax charge on profit loss to market value TAXMV; iii operating profit to sales OPM; and iv trading profit to sales TPM. The terms of managers’ and employees’ compensation plans can also impact on the choice of hedging technique see Smith, 1993. Managers will use those derivatives, e.g. FX options, which increase the volatility of the firm’s stock price if a large part of their compensation is in the form of stock options. Following Smith and Stulz 1985 a positive relationship is expected between both managers’ and employees’ wealth and the extent to which the firms use derivatives, particularly when hedging economic and translation exposures. The measures of wealth are the coefficient of variation of: i directors’ remuneration to market value DIRECMV; ii employees’ remuneration to market value EMPMV; and iii BOOKMV. The predictions are further summarised in Appendix A.

3. Empirical results

3 . 1 . Some general results Since the firms are large, one would expect them to make much greater use of internal hedging techniques than external techniques. Further, much greater use of internal techniques would be expected because of the transaction cost, biased pricing, default risk, etc. see Riehl and Rodriguez, 1977 that are associated with N .L . Joseph J . of Multi . Fin . Manag . 10 2000 161 – 184 167 Table 1 Summary statistics for the extent to which hedging techniques are used by the firms a Hedging techniques Transaction exposure Economic exposure Translation exposure Percentage rating Percentage rating Percentage rating Mean 1 3 N Mean 1 3 N N Mean 1 3 Panel A. The degree of utilisation of internal hedging techniques by type of exposure 6.9 63 1.13 87.3 0.0 67 1.06 94.0 0.0 1.49 a Leads and lags 58.3 72 1.43 71.4 14.3 b Matching inflows and outflows with respect to timing of settlement 67 72 1.25 82.1 7.5 2.10 22.2 31.9 63 1.41 68.3 9.5 67 1.31 79.1 63 10.4 23.6 41.7 c Inter-company netting of foreign receipts and payments 72 2.18 62 73 1.40 69.4 9.7 66 1.14 89.4 3.0 2.11 24.7 35.6 d Domestic currency invoicing 1.31 79.0 9.7 66 1.03 98.5 1.5 e Adjustment clause in sales contract 73 1.67 46.6 13.7 62 1.47 63.3 10.0 64 1.67 53.1 60 20.3 11.3 f Assetliability management 71 1.54 57.7 g Transfer pricing agreements 1.13 70 88.3 1.7 65 1.03 96.9 0.0 1.41 61.4 2.9 60 Panel B. The degree of utilisation of external hedging techniques by type of exposure 33.3 68 1.79 50.0 29.4 71 2.38 21.1 59.2 72 a Foreign currency borrowinglending 1.94 38.9 1.75 47.8 22.4 68 1.53 61.8 67 14.7 74 b Forward exchange contracts 82.4 4.1 2.78 69 74 1.33 73.9 7.2 69 1.22 81.2 2.9 1.84 35.1 18.9 c Foreign exchange options 1.01 98.5 0.0 68 1.03 91.1 d Foreign exchange futures 0.0 73 1.07 94.5 1.4 67 1.03 97.1 0.0 – – – 68 – 74 e Factoring bills receivable 2.7 82.4 1.20 69 74 1.42 69.6 11.6 70 1.46 68.6 14.3 1.34 75.7 9.5 f Cross-currency interest rate swaps g Foreign currency swaps 1.46 74 65.2 11.6 70 1.64 57.1 21.4 1.65 54.1 18.9 69 1.09 91.3 0.0 69 1.07 94.2 69 1.4 78.4 1.4 h European currency unit 74 1.23 69 74 1.01 98.6 0.0 69 1.00 100.0 0.0 1.01 98.6 0.0 i Special drawing rights j Other currency blocs 1.07 74 94.2 1.4 69 1.04 95.7 0.0 1.20 85.1 5.4 69 1.17 88.4 5.8 69 1.10 92.8 2.9 69 k Government exchange risk guarantee, e.g. ECGD 74 1.36 68.9 5.4 a The summary statistics relate to the scores obtained on a 3-point scale where 1 denotes not used; 2 denotes occasionally used; and 3 denotes frequently used. external techniques. Table 1 shows that the degree of utilisation of internal and external hedging techniques varies with the type of exposure 6 . However, there is preliminary evidence to suggest that the firms place more emphasis on certain external techniques — a finding which has been noted elsewhere see McRae and Walker, 1980, p. 101. In particular, Panel A shows that inter-company netting and domestic currency invoicing in that order are the most commonly used techniques when hedging transaction exposure. Matching inflowsoutflows and assetliability manage- ment are respectively the most commonly used techniques when hedging economic and translation exposures. Evidence from Khoury and Chan 1988 shows that matching is the most popular internal technique used by US firms. Here, US firms consider matching to be the most flexible and ‘self-reliant’ way to hedge. Panel B also shows that the firms use a limited set of external techniques to hedge the exposures. The FX forward contract is the most commonly used hedging technique. This finding is similar for US firms see Phillips, 1995. Forward contracts are mainly used to hedge transaction exposure. While foreign currency borrowing lending is the most commonly used technique when hedging both economic and translation exposures, it is the second most commonly used technique when hedging transaction exposure. The use of foreign currency borrowinglending may reflect the desire of the firms to reduce the amount of investment that is abroad see Belk and Glaum, 1990, but the degree of usage is stronger for translation exposure than for economic exposure. Cross-currency interest rate swaps and foreign currency swaps are not commonly use by the firms. This is consistent with the findings of Glaum and Belk 1992. The utilisation rates of both FX options and futures are low for all types of exposures but FX options tend to be more widely used than FX futures see also Glaum and Belk, 1992; Phillips, 1995. The low utilisation of FX futures may be due to the effects of daily resettlement which can adversely affect the liquidity of firms. In general, external techniques appear to play a much more important role in hedging decisions then internal techniques. As the firms are large, scale economies in the use of external techniques and the availability of skilled treasury personnel may contribute to their greater use see Geczy et al., 1997. However, the firms do not appear to be very selective in their use of the techniques when hedging different types of exposures. 3 . 1 . 1 . Hedging exposures with similar internal techniques To test for a link between the utilisation rates of internal techniques, a x 2 test was applied 7 . The test was applied to determine whether or not: i the firms are selective 6 The 3-point scale identified the degree of usage as: 1 = not used; 2 = occasionally used; and, 3 = frequently used. The occasional use of hedging techniques may be associated with partial hedging andor hedging strategies which reflect expected changes in the behaviour of the financial markets. Since the aim is to capture the degree of utilisation, it would appear that the use of a larger point scale would not have altered the results see Lehmann and Hulbert, 1972. 7 The results where the statistical test yields a value whose associated probability under the null hypothesis is 5 or less P-value 5 0.05 are reported. This cut-off point is applied throughout this study unless explicitly stated otherwise. in their use of the hedging techniques; and ii certain techniques are perceived to have special attributes such that they would only be used to hedge specific exposures. The null hypothesis that there is no difference in the utilisation rate of matching when the firms hedge transaction and economic exposures is rejected x 4 2 = 13.996; P-value = 0.007. The contingency table suggests that the firms make greater use of matching when hedging transaction exposure compared with eco- nomic exposure. For example, 32 of the 45 firms 71.11 that do not use matching to hedge economic exposure, also use matching occasionally and frequently to hedge transaction exposure. Using the Cramer test statistic, C see Siegel and Castellan, 1988 the association appears to be moderate C = 0.333. The null hypothesis of no difference in the degree of utilisation of matching when hedging economic and translation exposures is also rejected x 4 2 = 33.668; C = 0.521; P- value = 0.000. Here, 50 of the firms do not use matching to hedge translation exposure and 90.00 of those firms do not hedge economic exposure with this technique either. However, the firms have a much stronger preference for inter- company netting when hedging transaction exposure compared with economic exposure overallx 4 2 = 13.910; C = 0.332; P-value = 0.008. While 43 of the firms do not use inter-company netting to hedge economic exposure, 72.10 of those firms hedge transaction exposure with inter-company netting. Similar inference can be made for the use of assetliability management and leads and lags across exposures, but in general, the firms appear to prefer to use those techniques to hedge transaction exposure. 3 . 1 . 2 . Hedging exposures with similar external techniques External techniques such as currency swaps and foreign currency borrowinglend- ing, allow firms to borrow more cheaply than would otherwise have been possible. Those techniques also enable firms to reduce or eliminate the amount of their foreign investments see Glaum and Belk, 1992. Cross-currency interest rate swaps also share those attributes. In general, if those techniques enable firms to reduce the amount of their foreign investment, one would expect their use to be more strongly associated with economic and translation exposures. The x 2 test provided some support for this prediction. The null hypothesis of no difference in the degree of utilisation of foreign currency borrowinglending when hedging economic and translation exposures is rejected overallx 4 2 = 16.904; C = 0.355; P-value = 0.002. Of the 39 firms that hedge translation exposure with foreign currency borrowing lending, 35.89 of them frequently hedge economic exposure with the same technique. However, more than half of those firms 21 out of 39 do not use foreign currency borrowinglending to hedge economic exposure. Thus it seems that the firms prefer to hedge translation exposure with foreign currency borrowinglending. In contrast, the firms make much greater use of currency swaps when hedging economic exposure compared with transaction exposure overall x 4 2 = 20.680; C = 0.387; P-value = 0.001 but most of the firms do not use cross-currency interest rate to hedge their exposures. If managers believe that FX options provide a genuine hedge but see, Giddy and Dufey, 1995, they are more likely to use them to hedge economic and translation exposures. The null hypothesis of no difference in the extent to which the firms use FX options to hedge economic and translation exposures cannot be rejected. However, the null hypothesis can be rejected for transaction and translation exposures x 4 2 = 10.224; C = 0.272; P-value = 0.037. Here, FX options are primar- ily used to hedge transaction exposure. Up to 37 of the 56 firms that do not use FX options to hedge translation exposure use the derivative to hedge transaction exposure. 3 . 1 . 3 . Hedging transaction exposure with different techniques Since it is more difficult to match the maturity of derivatives with those of the underlying economic and translation exposures, firms would be expected to make much greater use of internal techniques. Although, the preliminary evidence sug- gests that the firms place a stronger emphasis on external techniques, the hypothe- sised relationships are directly tested here. The null hypothesis of no difference in the extent to which the firms use foreign currency borrowinglending and assetli- ability management when hedging transaction exposure is easily rejected x 4 2 = 13.109; C of 0.308; P-value = 0.011. Thirty-nine firms do not use assetliability management to hedge transaction exposure and more than half of those 56.41 do not use foreign currency borrowinglending either. Indeed, the contingency table suggests that there is a stronger preference for foreign currency borrowinglending. The null hypothesis is also rejected for the extent to which forward contracts are used compared with matching and domestic currency invoicing 8 . The results are similar for the extent to which the firms use FX options compared with other internal techniques 9 . 8 The test statistics for the extent to which FX contracts and the relevant internal hedging techniques are used when hedging transaction exposure are as follows: Use of FX forward contracts P-value C x 4 2 Use of: 12.361 Matching 0.293 0.015 0.035 0.266 10.317 Domestic currency invoicing 9 The test statistics for the extent to which FX options and the relevant internal hedging techniques are used when hedging transaction exposure are as follows: Use of FX options P-value C Use of: x 4 2 0.006 14.488 0.317 Leads and lags 13.858 Domestic currency invoicing 0.308 0.008 Transfer pricing arrangements 0.002 0.350 17.159 3 . 1 . 4 . Hedging economic exposure with different techniques Some significant results were also found for the degree of utilisation of certain techniques when hedging economic exposure. For example, the null hypothesis for the degree of utilisation of foreign currency borrowinglending and assetliability management is rejected x 4 2 = 24.199; C = 0.453; P-value = 0.000. Most of the 29 firms 86.21 that do not use foreign currency borrowinglending use assetliability management. The results are also significant for the relationship between the degree of utilisation of: i FX options and certain internal techniques; and ii foreign currency swaps and certain internal techniques 10 . The relationships are weak to moderate and reflect the stronger emphasis on external techniques. 3 . 1 . 5 . Hedging translation exposure with different techniques Evidence from Collier et al. 1990 indicates that some treasury managers of both UK and US firms are concerned about the adverse impacts of translation risk on leverage, distributable reserves and the overall balance sheet value. One implication of this finding is that managerial attitudes towards translation exposure would vary, particularly when firms are faced with hedging techniques which increase the variability of those measures. The results indicate a moderate association between the degree of utilisation of foreign currency borrowinglending and assetliability management x 4 2 = 15.525; C = 0.348, P-value = 0.004. The firms generally make much greater use of foreign currency borrowinglending to hedge translation exposure. For example, of the 34 non-users of assetliability management, up to 61.77 of them are occasional 17.65 and frequent 44.12 users of foreign currency borrowinglending. Most firms that use forward contracts also use match- 10 The test statistics for the extent to which FX options and the relevant internal hedging techniques are used when hedging economic exposure are as follows: Use of FX options x 4 2 Use of: C P-value 10.027 0.004 Matching inflowsoutflows 0.282 21.674 Inter-company netting 0.415 0.000 0.046 Invoicing in domestic currency 0.279 9.667 11.695 0.312 Transfer pricing arrangements 0.020 For economic exposure, the test statistics for the extent to which foreign currency swaps and the relevant internal hedging techniques are used are as follows: Use of foreign currency swaps x 4 2 C Use of: P-value 25.457 Inter-company netting 0.449 0.000 Domestic currency invoicing 0.025 11.186 0.300 14.817 0.351 0.005 Assetliability management ing, inter-company netting and domestic currency invoicing 11 although the empha- sis on forward contracts is not strong. Similarly, the null hypothesis that there is no difference in the degree of utilisation of foreign currency swaps and assetliability management is rejected x 4 2 = 12.275; C = 0.312; P-value = 0.015. In general, the firms place a much weaker emphasis on translation exposure. 3 . 2 . Bi6ariate test of hedging techniques and firms ’ characteristics In this sub-section, the extent to which cross-sectional variation in the character- istics of the firms can explain the degree of utilisation of the hedging techniques is assessed. The data for the characteristics of the firms is not normally distributed. Therefore, the distribution-free Kruskal – Wallis statistic has been used. To illustrate the testing procedure, the test statistics associated with the use of FX options and the characteristics of the 64 non-anonymous firms are shown for the case of transaction exposure see Table 2. If the degree of utilisation of FX options is associated with the financial measures, one would expect to observe differences in the variability of the financial measures. The table shows, for example, that frequent users of FX options exhibit less variability on dividend yield compared to both occasional and non-users; the associated Kruskal – Wallis test statistic is significant P-value = 0.015. Thus the null hypothesis that the k samples are from identical populations with the similar medians can be rejected and it can be inferred that the degree of usage is associated with differences in the variability of the measure. 3 . 2 . 1 . Internal hedging techniques and firms ’ characteristics In most cases, the degree of utilisation of internal techniques is positively related with the measures of internationalisation. In the case of transaction exposure, the Kruskal – Wallis statistic indicated that both occasional and frequent users of matching, domestic currency invoicing and transfer pricing tend to be larger in terms of both NCOUNT and NSUBS. As expected, a higher degree of internation- alisation appears to be associated with an increase in the use of internal techniques. Similarly, firms that use inter-company netting are larger in terms of both measures, but in addition, the magnitude of PERSALE is also larger. As expected, firms that use assetliability management to hedge transaction exposure exhibit less variability on QAR. However, PERHEDGE and the leverage measures were not found to be 11 The test statistics for the extent to FX forward contracts and the relevant internal hedging techniques are used when hedging translation exposure are as Use of FX forward contracts P-value C x 4 2 Use of: 0.297 11.267 0.024 Matching inflowsoutflows 14.980 Inter-company netting 0.342 0.005 Domestic currency invoicing 0.025 0.297 11.106 N .L . Joseph J . of Multi . Fin . Manag . 10 2000 161 – 184 173 Table 2 The characteristics of the firms conditioned on the degree of utilisation of foreign exchange FX options when hedging transaction exposure a Test statistic Combined sample FX options not used FX options occasionally used FX options frequently used S.D. N Mean S.D. K–W P-value N Mean Mean S.D. N Mean S.D. N 1 . Coefficient of 6ariation of financial measures 20 30.703 20.340 13 44.244 25.781 5.827 0.054 61 32.304 20.358 26.784 12.952 SALESMV 28 20.888 13 35.646 24.416 2.493 0.287 61 30.095 28.972 20 19.165 BOOKMV 28. 9.708 23.060 29.117 22. 13.894 20 27.199 16.017 8.416 0.015 58 29.567 19.621 32.047 25.750 16 Dividend 21 20.200 13.720 13 35.271 24.773 4.508 0.105 63 23.016 18.968 19.317 18.977 OPM 29 14.517 13 21.059 15.109 2.603 0.272 63 14.497 16.769 18.226 15.901 29 TPM 21 17.250 28.802 20 16.696 28 27.078 11.845 0.565 0.754 61 26.939 13.617 24.242 9.384 28 GCASHMV 18.280 13 39.969 21.817 1.671 0.434 61 CASHMV 45.314 20 22.682 51.367 28.036 28 43.472 10.322 13 14.865 11.833 0.991 0.609 63 16.583 16.280 21 10.428 QAR 29 10.112 16.739 13.148 21 9.447 13 11.733 9.026 1.478 0.477 63 13.046 8.834 13.720 8.154 29 WCR 17.561 13 39.226 CGEAR 41.907 21 2.760 0.252 63 25.334 25.622 24.529 19.517 29 19.906 19.565 13 47.102 39.092 0.867 0.647 63 34.537 39.158 29 26.108 IGEAR 21 40.627 24.208 33.631 13 56.070 43.403 2.268 0.322 61 LTBORMV 44.729 20 33.461 44.282 24.786 28 39.784 33.872 13 50.682 46.518 3.346 0.188 63 29.034 38.421 TLBOR 48.002 29 63.107 43.792 21 9.562 21 8.234 13 − 42.116 197.284 1.089 0.580 63 − 0.961 89.895 9.986 14.704 29 TAXRATIO 24.001 13 44.005 25.503 9.302 0.010 61 39.953 35.850 20. 22.237 TAXMV 28 11.547 24.803 30.267 21 17.575 13 41.375 28.508 4.022 0.134 62 30.836 20.894 25.072 17.869 28. DIRECMV 28 31.666 21.041 13 40.627 19.813 4.073 0.131 61 32.396 18.860 EMPMV 28.058 20 13.386 2 . Internationalisation measures 117.719 14 63.286 87.848 2.368 45.400 0.306 NSUBS 73 56.178 101.195 25 85.444 34 61.176 34 20.147 16.500 14 36.000 27.016 8.188 0.017 73 22.164 20.725 17.160 NCOUNT 19.375 25 26.982 PERSALE 14 26 46.429 33.479 5.423 0.066 73 42.671 30.551 31.731 31.239 33 49.967 26.549 13 62.692 25.545 1.440 0.487 67 55.690 60.075 29 25.768 PERHEDGE 25 63.800 25.219 a The summary statistics are conditioned on the degree of usage of foreign exchange options and are also given for the firms combined. The variable representing firm characteristics are described in the appendix. S.D. is the standard deviation. K–W is the Kruskal–Wallis test statistic which tests the null hypothesis that the characteristics of the firms conditioned on the degree of usage have the same median. The P-values50.05 are show in bold. For both PERSALE and PERHEDGE, respondents were asked to tick one percentage value or range of values from [0], [1–10], [11–20]…[91–100]. None of the firms recorded a score of 0. The statistics reported are based on the mid-point of those ranges. This table is for illustrative purposes only. significantly associated with any hedging technique. The results for translation and economic exposure are insignificant and reflect the low priorities the firms give to those exposures. 3 . 2 . 2 . External hedging techniques and firms ’ characteristics The Kruskal – Wallis test statistic also indicated that the use of external tech- niques varies with the characteristics of the firms. One interesting result is that occasional users of currency lendingborrowing tend to hedge a much larger percentage of PERHEDGE than frequent users, when transaction exposure is hedged. If firms partially hedge on the expectation of benefiting from FX trading see Hakkarainen et al., 1998, it is possible that the percentage of exposure that is hedged as well as the degree of utilisation of certain techniques will vary. Firms that frequently use FX forwards to hedge transaction exposure tend to exhibit much lower variability on CASHMV compared to occasional users. This result is expected since the mis-match of the cash flows from the instrument and the underlying exposure would not occur, in the absence of default. The degree of utilisation of FX forwards is also positively related to NCOUNT. Furthermore, the degree of utilisation of FX options is positively related to dividend yield, TAXMV and NCOUNT as well as the length of time since the firms had established their formal corporate hedging policies. Thus it appears that greater experience in exposure management see also Dolde, 1993 increases the firms’ confidence in using more complex techniques. Firms that are occasional and frequent users of factoring tend to hedge a larger percentage of PERSALE. They are also larger in terms of NSUBS. All those considerations apply to transaction exposure. The long-term nature of economic exposure presents special problems for firms such that those hedging techniques which reduce the amount of foreign investments are likely to be preferred. Furthermore, the economic exposure arising from the long leads of growth options is likely to re-enforce the incentive to partially hedge. In general, the results indicate that both occasional and frequent users of foreign currency borrowinglending exhibit lower variability on certain growth option measures, i.e. BOOKMV and SALESMV. The degree of variability on those measures is lowest for frequent users. Further, the degree of utilisation is positively related to PERSALE. It should be noted that firms that hedge more than 81 of their global exposure exhibit greater variability on SALESMV, BOOKMV and dividend yield while firms that hedge less than 40 of global exposure exhibit the least variability on those measures. But the statistical results are not significant P-values ] 0.060. However, there is the potential for the degree of utilisation to impact on the percentage of exposure that is hedged. Frequent users of FX options tend to exhibit less variability on the growth option measures as well as GCASHMV, DIRECMV, EMPMV, while occasional users exhibit the highest level of variability on those measures. Extending Smith’s 1993 argument, it had been suggested that the terms of managerial compensation would provide an incentive for using FX options, forwards and futures particularly when hedging economic and translation exposures. The results suggest that the incentive to increase the firms’ volatility is greater only for occasional users of FX options. Finally, the results indicate that firms that use foreign currency borrowinglend- ing to hedge translation exposure exhibit less variability on OPM, IGEAR, and LTBORMV. The lower variability of the leverage measures is unexpected. How- ever, firms that hedge less than 40 of their global exposure exhibit less variability on both CGEAR and IGEAR while those that hedge between 41 and 80 of their global exposure exhibit more variability on those measures P-value 5 0.025. Thus it appears that the extent to which exposure is hedged impacts on the hypothesised relationship. The use of foreign currency borrowinglending when hedging transla- tion exposure is also positively related with NSUBS and PERSALE. While occa- sional users of foreign currency swaps exhibit greater variability on both OPM and TAXMV, frequent users exhibit less variability on OPM. These contrasting results may be due to the inflexibility inherent in the use of foreign currency swap agreements. 3 . 3 . Multi6ariate test of hedging techniques and firms ’ characteristics Bivariate tests tend to be weak since they do not allow for interactions among the explanatory variables. To further assess the choice of hedging techniques, a logistic regression was applied. The dependent variable of the logistic regression is deter- mined by using the ratings that represent the degree of usage. Here, the score of 1 not used is coded as 0 and, the scores of 2 and 3 occasionally and frequently used are coded as 1. The existence of missing explanatory variables and the anonymous responses result in an overall sample size of 54 firms. To minimise the potential problems of small sample size, we present the results for the models where: i at least 20 of the firms can be allocated to either group 0 or 1, a priori; ii each empirical model outperforms a naive proportional chance model see Joy and Tollefson, 1975; and iii each model’s x n 2 statistic is significant at the 5 0.05 level. No evidence was found that suspiciously large regression residuals had an adverse effect on the estimated coefficients. 3 . 3 . 1 . Internal hedging techniques Panel A of Table 3 shows that the explanatory variables exhibit some discrimina- tory power for the degree of utilisation of internal hedging techniques. Only the coefficients associated with transaction exposure are significant. As expected, the use of internal techniques is positively related with measures of internationalisation. The use of leads and lags is positively related with the coefficients of both SALESMV and NCOUNT but those coefficients are marginally significant. The coefficient value of 0.138 for SALESMV means that as its variability increases, all else held constant, the likelihood that the firm will use leads and lags to hedging transaction exposure increases. In this case, each unit increase in the variability of SALESMV increases the log odds by a factor of 1.148; that is, e 0.138 . EMPMV makes the greatest contribution to the explanatory power of the model and the coefficient of PERHEDGE is always positive. Both EMPMV and DIRECMV have negative coefficients for leads and lags, and transfer pricing, respectively P- value 5 0.05. N .L . Joseph J . of Multi . Fin . Manag . 10 2000 161 – 184 176 Table 3 Logistic regression for the use of internal and external hedging techniques and the characteristics of the firms a Transaction exposure Domestic cur- Coeff Coeff R Leads and lags R Transfer pricing Coeff R rency invoicing Panel A : Internal hedging techniques CASHMV − 0.182 SALESMV − 0.065 − 0.186 − 0.148 − 0.251 − 0.038 EMPMV 0.020 0.070 0.027 LTBORMV 0.042 SALESMV 0.377 TAXMV 0.106 0.152 0.298 0.138 0.018 0.014 0.040 − 0.055 − 0.249 PERHEDGE 0.024 DIRECMV 0.120 0.122 0.027 NCOUNT 0.025 0.014 0.015 NCOUNT 0.071 Constant 0.400 Constant − 1.254 − 0.402 1.182 0.023 0.754 0.382 PERHEDGE 0.050 0.019 − 4.179 Constant 1.516 Diagnostics Statistics Diagnostics Statistics Diagnostics Statistics Model’s x 3 2 12.909 Model’s x 5 2 24.552 10.348 Model’s x 3 2 classified 88.980 classified 70.370 classified 75.930 S. Residuals 88.890 − 0.022 S. Residuals S. Residuals − 0.051 N 1 , N 2 − 0.015 N 1 , N 2 30, 24 N 1 , N 2 32, 22 12.42 N .L . Joseph J . of Multi . Fin . Manag . 10 2000 161 – 184 177 Table 3 Continued Transaction exposure Coeff Foreign Currency Cross-currency Coeff R Factoring Coeff R R Coeff R exchange borrowing interest rate swaps options lending Panel B : External hedging techniques − 0.187 − 0.244 DIRECMV − 0.120 GCASHMV − 0.308 − 0.321 TLBOR TPM − 0.039 − 0.480 − 0.223 0.076 0.017 0.053 0.163 TAXMV 0.181 OPM 0.307 EMPMV 0.127 0.210 0.388 IGEAR 0.049 0.268 0.200 0.024 0.048 0.051 0.062 CASHMV − 0.200 NSUBS 0.013 − 0.256 − 0.096 0.413 CASHMV NCOUNT − 0.035 − 0.047 − 0.137 0.020 0.022 0.016 0.005 BOOKMV 0.095 0.118 PERSALE − 0.036 − 0.234 CASH 0.084 0.176 0.042 0.055 0.016 SALES 0.187 0.266 0.071 Constant − 3.389 PERHEDGE 0.227 0.068 0.029 1.038 Constant Constant − 1.242 0.934 1.246 Constant − 0.163 1.245 1.106 Diagnostic Statistics Diagnostic Diagnostic Statistics Statistics Statistics Diagnostic Model’s x 3 2 19.075 Model’s x 6 2 33.527 Model’s x 4 2 12.344 Model’sx 4 2 25.314 classified 87.040 classified 79.630 79.63 classified 87.040 classified S. Residuals 0.007 S. Residuals − 0.099 S. Residuals − 0.121 S. Residuals 0.003 N 1 , N 2 43, 11 N 1 , N 2 41, 13 N 1 , N 2 18, 36 18, 36 N 1 , N 2 N .L . Joseph J . of Multi . Fin . Manag . 10 2000 161 – 184 178 Table 3 Continued Economic exposure Translation exposure Coeff R Currency bor- Coeff Coeff R Currency bor- Foreign cur- Foreign ex- R Coeff R rowinglending rowinglending change options rency swaps 0.254 0.188 OPM − 0.092 0.118 GCASHMV − 0.363 OPM TPM 0.338 0.046 0.339 0.047 0.125 0.023 0.035 SALESMV 0.049 Dividend yield 0.251 CGEAR − 0.041 − 0.093 − 0.382 OPM − 0.340 − 0.061 − 0.374 0.018 0.117 0.021 0.038 IGEAR − 0.249 TAXMV 0.059 − 0.120 − 0.065 0.282 EMPMV TAXRATIO 0.179 − 0.053 0.305 0.038 0.023 0.027 0.072 Constant − 1.601 Constant PERSALE 0.026 0.149 NSUBS 0.151 1.049 0.219 0.015 0.007 0.714 0.862 PERHEDGE 0.038 0.225 Constant − 0.447 0.019 0.619 Constant − 0.470 1.357 Diagnostics Statistics Diagnostics Diagnostics Statistics Statistics Diagnostics Statistics Model’s x 5 2 20.915 Model’s x 4 2 13.597 28.689 Model’s x 3 2 6.108 b Model’s x 3 2 classified 85.190 classified 72.220 a classified 72.220 classified 75.930 S. Residuals − 0.016 S. Residuals − 0.055 0.047 S. Residuals − 0.017 S. Residuals N 1 , N 2 13, 41 N 1 , N 2 28, 26 N 1 , N 2 23, 31 N 1 , N 2 37, 17 a The explanatory variables are entered intoremoved from the logistic regression using a stepwise procedure. The likelihood-ratio test is used for both entering cut-off P-value50.05 and removing cut-off P-value]0.10 the explanatory variables intofrom the model. For this reason some coefficients are significant at the 10 level, which is considered to be marginal. The Wald statistic is used to test the null hypothesis that each coefficient of the model is zero. The standard errors of the coefficients are in parentheses. R is the partial correlation between the dependent and independent variables. S. Residuals is the average of the standardised residuals of the logistic regression model. For the ith case its residual is divided by p i 1−p i where p i is the predicted value. The chi-square x n 2 statistic tests the null hypothesis that all coefficients in the model except the constant are simultaneously zero against the alternative that at least one coefficient is non-zero. The percentage correctly classified is a measure of the classificatory efficiency of the model. The level of significance indicates that the classificatory efficiency of the empirical model is superior to that of a naive proportional chance model see, Joy and Tollefson, 1975. N 1 indicates the number of firms in the sample that do not use the hedging techniques while N 2 indicates the total number of occasional and frequent users. The results are presented where: i it was possible to allocate at least 20 of the firms to group 0 or 1 a priori; ii the percentage correctly classified by the empirical model outperforms a proportional chance naive model P-valueB0.05; one tailed; and iii each model’s x n 2 statistic is significant at the 5 level or less. The test statistic is significant at ]5 but 510 level. The test statistic is significant at ]1 but 55. The test statistic is significant at 51 level. 3 . 3 . 2 . External hedging techniques The results for the degree of utilisation of external hedging techniques are shown in Panel B of Table 3. Here, the degree of utilisation of foreign currency borrowing lending is associated with each type of exposure. The cash flow and profitability measures appear to explain the degree of utilisation, and in most cases, their coefficients carry the expected sign. Notice that the coefficient for TAXMV is positive implying that the use of FX options for hedging transaction exposure will increase as its variability increases. For transaction exposure, PERSALE carries an unexpected negative sign for cross-currency interest rate swap, and is marginally significant but positive in the case of translation exposure. In general, the characteristics of the firms can explain the choice of hedging technique but the explanatory power of the logistic regression is much stronger for the use of external techniques.

4. Summary and conclusions