—— :24: :9' T":- .‘ .- '{E' U355?“ “ngqu! ,1; 'IfiIU #0333 UU'U I"? U§U~U g3 3‘: I ‘ I), .. , UUU3UUU'U I'II.UI-.I{' ' 'IUIIUIUIUIU 333 UIUU’UU3UI3UIU UUUUIIUUI ’3fo UHUU'U'UUUUUUIfiU'U U3UUUU UUIUU'U [U '33 +fi;l[3:l UU ,I'.UUUI "3":ij h-(I UUU'TJUU ifs-1333:.) -UI.' UUV '3 U'UIUII‘5UUIU'I'UI UI'II‘I'I‘ I33! I U,‘ "DII 'I'III' .35” II IIUI .A' I IIIIIIIIIIIII :IJ' 3U}. U 'U .UUU3'3r'I- "‘II ' 33,3 'II'U . .w U’ U .3313. .. "‘3' UIUUUU “IE? "\UU‘U'I' UU 'UU'UU' UUU' 313i 3“ “2'32 _Z‘Ufifl""qh ' UUI' 3", I‘I 'U 3;“ L. .3)..." '35-‘39: ,JLW'H ‘ 4-.“ MS‘II ...‘LU LrIJ 1II'IV . '3 .'U I I II] I'. “Iv: L, IS? Ibgahfif ‘U. 2;; ‘25)» ’I II n-I‘II, . I; I“ “U {I I fl»? . ..ISIIEIII.II.II'III;I.,III.II- '.I';'.:. I U43 31-33:: UU'UU'UUIU Pk); “If" Ialg I . . . . .- _ I. I l~I-I-.I III. I“ ,IIqII .. I.) I. 5:115!” :U‘UUUd U‘J' UU U 'U " ” 'UU U UUUUUU ,‘cl'UUJ UU IUU: Up. ‘ I 'II ”I; "'3'" 1:3IUUUIUII \ WE? ‘_ U III? In I} I 3. - .U UI .3 \ "L‘fiUIIS-w '.-.Ir:II‘- MI .1ZI '. '. '4'!» U . 5 U ‘ ' 4| Itfi' M V r I 233,3 WI 7 “‘Sr '1 3‘... ‘ U I IIIIIIII - UM . xx‘ls‘ :75. -_.._.. I. If; . ‘il @UUJI'I'IULU . "1:3: " 1:17". ' D‘I U III p .93 EIIII.” i1 I34-7I’U53I3AHU‘: . \gUC-‘UEILUHU‘ I ._ p: #55wa T y-“ z —’ -I - 0;wa ‘ , r, %2;~212.h "’- '~ ‘ xg’rI 2.2 II III cox '33,); 3. ' g...) - I‘VE-i H‘Uu‘f‘¥h 3'3U‘UU'UU' I 1‘33: _ U "IU'L'Is ‘ R‘v'glgé' .3313. II V I ‘3th o “‘3 3 I #313.” U:."-1’U:II II S 7 ‘ U-U .III'( I I"II.-I'U Iffi‘rUk J 3:133 U"I,'.,U “Ur" ”33:! U" ‘31'333‘3' IUr"I I4" '33 U” 1UU3PV 33E" Wu IEIIIIUIJUAU .I'IUEUUIsUP: ‘UU '3" ,I U‘UUU"UU U‘UU'UIUI’U 3U"IU 135 ‘Uz".'.rIU;é'UUU}' U‘Ugjbr 3 {3:303} UH? III 3"? NIIIUIII 1U '3'3'133333'». 3'" 'UU‘U‘UUUU ”3' UUUU U'3U/U3 UU-II‘i-‘III'I-i‘IiiIiU 3'" "U' I ‘I.l ",I ;:I...u. U U'U‘UIUI‘. "U U'UUUIUU ’UU'UUUUUU I.U(UJI| "-'.' " I"I IUl-IIII .III ' II III I UUUUUUUUUUU’UI III. II I I . I I UUI .'Il',II IUUQ'I'U UUU U I. .III I'UUUU IU‘IUI‘;U‘.'3'U'-I‘UUUUUU\UUUU3UU III' I U. I ‘fIUI'UiUU-U, I'U‘LIU..I|'I‘I ,I. “:3. UUIUUUU UIU3UUIU'I EQUUUI'I'IUIIU -I 'UIU'U' 'UUUUUUUU ‘itl l ‘UUU'UI'I'IU‘I' I‘llIl33‘3 .I‘UI I '33 ‘UUfUUUUUIUU :IUI‘U" IIIIIII Hm“ '3'3'UIUUUU I“ IUUIIIII .II. .UII IUUUUU33U UUUIUUU‘UUUUUU - I.IIIIII 333M II 3“ I|I 3UUUU‘3IJU 6",3U: I. 3'" 3'3; hU‘U U‘3U' UUUUI'UU 13:3 ‘IUIUUUIUUUf'I In 33., .U“ 'IUUI WI .I\IIIII.I.I 3333' ”III ,U.,. U" JUIUUZI'II'UU 3 U' ,UU III}; 'IUUUUU UU U 3" 'U' 'I U UU U3 IUU‘UUUU 'UIUUIU' 3'2 fiIIIII-IUUI'U ,‘II'U U.'|Ul'. UUU U3 ' | 'UUU '(U’UURIUU (“'33. I U: ' I UUUI UUUUUU I UU .0 UUUU.UUUUI U'UUUUUMI 3UU'U3UU I UUIUIUUUU 'U‘II '3'.UU U ' UU3UUU'U‘ U'UUUIIUUUUU Ur'IUU 3"" . U Uh. .33 "' . UU'IUi ‘I I3 UUU II ... IIIIIU ""‘I'UI'UUUl 'UJII'II' UUI .I'U I 3- UI I .UIUUU .UUII I'...'UI:I UUUUU vUU'lUUUIUU . UUU.U"U 33.33 .- UU U U). II‘IUUU'UUI UJIII UUU."‘U‘ U-l .12. £2.57 ;".'I U IUU IUU 'EIU'I IIU. "UI' AA —— . , LIBRARY Michigan Stat: I University rat! 1: This is to certify that the thesis entitled THE RELATIONSHIP BETWEEN UNITED STATES AND FOREIGN INTEREST RATES: A STUDY OF FINANCIAL MARKET INTEGRATION presented by Anthony Q. Chua has been accepted towards fulfillment of the requirements for Ph.D. degreein Business Administrat1on 11111111111 3 1293 10443 lllllllll lllllllllllllll JJ: Fir‘g‘ IA" .H p 1331,“ ‘_Z ‘_:;.:;.~v »-.. ——— .._é___ win“ , P‘sage i“ ' )OOD. HEL'JYT. {O "El-Vt: “‘53 2‘ fl" . m J“ Cl ““5“: I...“ 'fii'W‘q 13¢ per la, fit, ,. 1:3; g". ‘r; THE RELATIONSHIP BETWEEN UNITED STATES AND FOREIGN INTEREST RATES: A STUDY OF FINANCIAL MARKET INTEGRATION By Anthony Q. Chua A DISSERTATION Submitted to Michigan State University in partial filfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Accounting and Financial Administration 1981 é//é 68’; @9 Copyright A"th°“y Q. Chua 198] ABSTRACT THE RELATIONSHIP BETWEEN UNITED STATES AND FOREIGN INTEREST RATES: A STUDY OF FINANCIAL MARKET INTEGRATION By Anthony Q. Chua Research on financial market integration has concentrated on the existence of integration rather than on the structure of any relationship. With the exception of Hendershott (45) and Kwack (55), there has been no test of the speed of adjustment of interest rate changes among financial markets. This study will examine the temporal and structural relation- ships involved in the adjustment of interest rates to one another. Instantaneous adjustment of foreign interest rates (holding-period returns) to changes in the United States interest rates is hypothe- sized. Invoking the concepts underlying the relative purchasing power parity theory, the interest rate theory of exchange rate expectation, and the interest rate parity theory, the fOreign inter- est rates were adjusted to account fbr external factors such as the level of international trade and movements, the level of foreign exchange rates, and the relative price level. Because the adjust- ments were done such that a complete turnaround transaction was Anthony Q. Chua obtained, the adjusted foreign interest rates were effectively holding-period returns. The relationships examined include those of the United States interest rates with the Eurodollar rate, and the short-term and long-term interest rates of Canada, West Germany, United Kingdom, France, Netherlands, and Switzerland. The relationships were exam- ined over the time periods from l97l-72, 1973-75, and 1975-78, using the methods developed by Box and Jenkins. The results of the study provide evidence for the nonrejec- tion of the hypothesis that financial markets are integrated. That there is instantaneous adjustment of foreign interest rates (holding- period returns) to changes in the United States interest rates cannot be rejected. This statement is supported by the magnitudes of the cross correlation coefficients depicting the relationships between the United States interest rates and foreign adjusted inter- est rates. However, financial market integration characterizes the May 1975-November 1978 period, but not the January 197l-April 1975 period. The exceptions are Canada and France. The Canadian adjusted interest rates are found to be consistently correlated to the United States interest rates over the entire sample period, while the French adjusted interest rates are found to be consistently unrelated to the United States interest rates over the entire sample period. The implications of the results are twofold. 0n the micro- economic level, the lead-lag relationships imply little benefit to be derived from the transfer of funds across national boundaries. Anthony Q. Chua This statement is based on the assumption that the transfer of funds takes place as a reaction to a change in the level of interest rates in the United States. Also, moving across boundaries to take advan- tage of lower financial costs provides little benefit to the firm. 0n the macroeconomic level, the study implies that, on the one hand, governments have had some success in reducing foreign influences on their domestic financial rates, as shown by the somewhat weak coefficients found in the cross correlation analysis. On the other hand, the existence of lead-lag relationships suggests that governments have not been able to completely isolate their economies from events in the United States financial market. Hence, the use of interest rates as a tool of monetary policy has not necessarily been weakened. ACKNOWLEDGMENTS I wish to express my appreciation to the members of my dissertation committee, Dr. John R. Brick (chairman), Dr. Richard R. Simonds, and Dr. Dennis Warner, for their guidance, encourage- ment, and aid in completing this dissertation. To Dr. Myles S. Delano and Dr. Alden C. Olson, I express my gratitude for their advice and understanding during the period of my doctoral program. Thanks are also extended to Dr. Gardner M. Jones and Dr. Harold M. Sollenberger, who served as Chairman of the Department of Accounting and Financial Administration, for marshalling a good measure of academic and financial support during my stay at Michigan State University. Finally, to my parents, for their support, kindness, and encouragement, I dedicate this dissertation. ii TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . . . . xv LIST OF FIGURES . . . . . . . . . . . . . . . xii LIST OF APPENDICES . . . . . . . . . . . . . . xiii Chapter I. INTRODUCTION . 1 Objective of the Study . 2 Importance of the Study . 5 Financial Integration Defined 6 Background of the Study 7 Literature Review . . 9 Capital FLows as a Measure of Integration . 9 Interest Rates as a Measure of Integration 15 Swmmy. . . . . . . . . . fl II. THEORETICAL CONSIDERATIONS 23 Factors Relevant to the Study 24 Expected Price Level Change 25 Expected Exchange Rate Change . 29 Summary. . . 33 III. RESEARCH HYPOTHESES AND DESIGN . 35 Research Hypotheses . 35 Research Design 37 Sample . . . 37 Data Source . . . . . . . . 37 Time Period. . . . . . 38 Data Adjustment Procedure 40 Methodology. 42 Summary. 50 IV. SHORT-TERM INTEREST RATE RELATIONSHIPS 52 Univariate Analysis . 52 Cross Correlation Analysis 60 iii Chapter VI. VII. VIII. IX. Multivariate Analysis Summary. LONG-TERM INTEREST RATE RELATIONSHIPS THREE- MONTH HOLDING PERIOD Univariate Analysis . Cross Correlation Analysis Multivariate Analysis Summary. . . . . LONG-TERM INTEREST RATES SIX-MONTH HOLDING PERIOD Univariate Analysis . Cross Correlation Analysis Multivariate AnalySis Sumnary. . . . . LONG-TERM INTEREST RATES ONE-YEAR HOLDING PERIOD . Univariate Analysis . Cross Correlation Analysis Multivariate Analysis Summary. . . . . LONG-TERM INTEREST RATE RELATIONSHIPS OVERALL EVALUATION . . Univariate Analysis . . Cross Correlation AnalySis Multivariate Analysis CONCLUSION . . . . . . . , Summary of Findings--Short-Term Interest Rates . Summary of Findings--Long-Term Interest Rates Results of the Study .v . . . . . . Implications of the Result Suggestions for Future Research . iv Page 68 77 79 80 91 98 106 108 108 115 122 132 134 134 140 144 152 153 153 162 166 170 171 173 174 176 178 Table 10. 11. 12. 13. LIST OF TABLES Studies on capital movements in response to interest rate changes . . . . . . . . . Studies on interest rate parity theory Univariate time series analysis, short-term interest rates . . . . . . . . . . . . . . . Comparison of ARIMA (1,l,0) models fitted to the second and third subperiods of the Eurodollar time series . . . . . . . . . . . . . Theil's U coefficient for the univariate time series models, short-term interest rates . . . Cross correlations of weekly interest rates, January 29, 1971 - December 8, 1978 . . . . Multivariate time series models, short-term interest rates . . . . . . . . . . . . . Mean squared error of the forecast of the output series with and without the input series,short-term interest rates . . . . . . . . . Variance of the forecast errors made with and without the input series, short-term interest rates . Theil's U coefficient, multivariate time series models, short-term interest rates . Univariate time series models, long- -term interest ra tes O O O O O O O O O O 0 Comparison of ARIMA (1,1,0) models fitted to the first and second subperiods of the U. S. 10-year constant maturity bond yield. . . . . . . . . Comparison of ARIMA (O,l,l) models fitted to the West German long-term public authority loan rate . . Page 11 18 54 56 59 62 7O 71 72 73 81 85 86 Table Page 14. Comparison of ARIMA (1,1,0) models fitted to the first and third subperiods of the Dutch long-term government loan rate . . . . . . . . . . . . . . . 89 15. Comparison of ARIMA (1,1,0) models fitted to the first and third subperiods of the Swiss long- term confedera- tion bond yield . . . . . . . . . . . . . 90 16. Theil's U coefficient for the univariate time series models, long-term interest rates . . . . . . . 91 17. Cross correlations of weekly interest rates January 29, 1971 - November 23, 1978 . . . . . . . . . 92 18. Multivariate time series models, long-term interest rates . . . . . . . . . . . . . . . . 99 19. Mean squared error of the forecast of the output series with and without the input series, long-term interest rates . . . . . . . . . . . . . 101 20. Variance of the forecast errors made with and without the input series, long-term interest rates . . . . 102 21. Theil's U coefficient, multivariate time series models, long-term interest rates . . . . . . . 103 22. Univariate time series models, long-term interest rates . . . . . . . . . . . . . . . . 109 23. Theil's U coefficient for the univariate time series models long-term interest rates . . . . . . . . 116 24. Cross correlations of weekly interest rates January 29, 1971 - November 23, 1978 . . . . . . . . . 117 25. Multivariate time series models, long-term interest rates . . . . . . . . . . . . . . . . 123 26. Mean squared error of the forecast of the output series with and without the input series, long-term interest rates . . . . . . . . . . . . 125 27. Variance of the forecast errors made with and without the input series, long-term interest rates . . . . 126 28. Theil's U coefficient, multivariate time series models, long-term interest rates . . . . . . . 128 vi Table Page 29. Univariate time series models, long-term interest rates . . . . . . . . . . . . . . . . . 135 30. Theil's U coefficient for the univariate time series models, long-term interest rates . . . . . . . . 139 31. Cross correlations of weekly interest rates January 5, 1973 - November 23, 1978 . . . . . . . . . 141 32. Multivariate time series models, long-term interest rates . . . . . . . . . . . . . . . . . 145 33. Mean squared error of the forecast of the output series with and without the input series, long-term interest rates . . . . . . . . . . . . . . 147 34. Variance of the forecast errors made with and without the input series, long-term interest rates . . . . 148 35. Theil's U coefficient, multivariate time series models, long-term interest rates . . . . . . . . . . 149 36. Univariate time series models, long-term interest rates . . . . . . . . . . . . . . . . . 154 37. Means, standard deviations, and coefficients of varia- tion of each interest rate series . . . . . . . 161 38. Cross correlation coefficients, long-term interest rates, holding-period returns . . . . . . . . . 163 39. Multivariate time series models, long-term interest rates . . . . . . . . . . . . . . . . . 167 A-l. Segression statistics for equation: st+n = a + th,n + 186 t o o o o o o o o o o o o o o o o o 8-1. Forecasts of the Canadian three-month finance paper rate September 27, 1974 - April 16, 1975 . . . . . 190 8-2. Forecasts of the three-month Eurodollar rate February 2, 1978 - December 8, 1978 . . . . . . . . . . 191 8-3. Forecasts of West German three-month interbank rate February 3, 1978 - December 8, 1978 . . . . . . . 193 B-4. Forecasts of the Canadian three-month finance paper rate February 3, 1978 - December 8, 1978 . . . . . 195 vii Table Page B-5. Forecasts of the United Kingdom three-month interbank rate February 2, 1978 - December 8, 1978 . . . . 197 B-6. Forecasts of the Dutch three-month interbank rate February 3,1978-Decenber 8,1978 . . . . . . 199 B-7. Forecasts of the Swiss three-month interbank rate February 3, 1978 - December 8, 1978 . . . . . . 201 Cel. Forecast of the Canadian long-term government bond yield Jualy 14, 1972 - December 29, 1972 . . . . 204 C-2. Forecasts of the Swiss long-term confederation bond yield July 14, 1972 - December 29, 1972 . . . . . 205 C-3. Forecasts of the Canadian long-term government bond yield October 11, 1974 - May 2, 1975 . . . . . 206 C-4. Forecasts of the West German long- term public authority loan rate January 20, 1978 - November 23, 1978 . . . . . 207 C-5. Forecasts of the Canadian long-term government bond yield January 20, 1979 - November 23, 1978 . . 209 C-6. Forecasts of the Dutch long-term government loan rate January 20, 1978 - November 24, 1978 . . . . 211 C—7. Forecasts of the Swiss long-term confederation bond yield January 20, 1978 - November 23, 1978 . . . . 213 D-l. Forecasts of the West German long-term public authority loan rate July 14, 1972 - December 29, 1972 216 0-2. Forecasts of the Canadian long-term government bond yield July 14, 1972 - December 29, 1972 . . . . . 217 D-3. Forecasts of the West German long-term public author- ity loan rate October 11, 1974 - May 2, 1975 . . . 218 D-4. Forecasts of the Canadian long-term government bond yield October 11, 1974 - May 2, 1975 . . . . . . 219 D-5. Forecasts of the United Kingdom 39% war loan yield October 11,1974 - May 2,1975. . . . 220 D-6. Forecasts of the French long-term public sector bond yield October 11, 1974 - May 2, 1975 . . . . . . 221 viii Table Page D-6. Forecasts of the French long-term public sector bond yield October 11, 1974 - May 2, 1975 . . . . . . 221 D-7. Forecasts of the West German long-term public author- ity loan rate January 20, 1978 - November 23, 1978 . 222 D-8. Forecasts of the Canadian long-term government bond yield January 20, 1978 - November 23, 1978 . . . . 224 D-9. Forecasts of the Dutch long-term government loan rate January 29, 1978 - November 24, 1978 . . . . . . 226 D-lO. Forecasts of the Swiss long-term confederation bond yield January 20, 1978 - November 24, 1978 . . . . 228 E-l. Forecasts of the West German long-term public author- ity loan rate October 11, 1974 - May 2, 1975 . . . 231 E-Z. Forecasts of the Canadian long-term government bond yield October 11, 1974 - May 2, 1975 . . . . . . 232 E-3. Forecasts of the French long-term public sector bond yield October 11, 1974 - May 2, 1975 . . . . . . 233 E-4. Forecasts of the Dutch long-term government loan rate October 11, 1974 - May 2, 1975 . . . . . . . . 234 E-S. Forecasts of the West German long-term public author- ity loan rate January 20, 1978 - November 23, 1978 . 235 E-6. Forecasts of the Canadian long-term government bond yield January 20, 1978 - November 23, 1978 . . . . 237 E-7. Forecast of the Dutch long-term government loan rate January 20, 1978 - November 23, 1978 .. . . . 239 F-l. United States three-month treasury bill rates: weekly January 29, 1971 - December 8, 1978 . . . . 242 F-2. Three month Eurodollar rates: Weekly, January 29, 1971 - December 8, 1978 243 F-3. West German three-month interbank rates: Weekly, January 29, 1971 - December 8, 1978 . . . . . 244 F-4. Canadian three-month finance paper rates: Weekly, January 29, 1971 - December 8, 1978 . . 245 ix Table F-S. F-6. F-10. F-11. F-12. F-13. F-14. F-15. F-16. F-17. F-18. F~19. United Kingdom three-month interbank rates: weekly, August 31, 1973 - December 8, 1978 . . . . French three-month interbank rates: weekly, April 25, 1975 - December 8, 1978 . . . . . Dutch three-month interbank rates: weekly, April 25, 1975 - December 8, 1978 . . . . . . . Swiss three-month interbank rates: weekly, August;29, 1975 - December 8, 1978 . . . . . . . United States ten-year constant maturity bond yield: weekly,January 29, 1971 - November 23, 1978 West German long-term public authority loan rate adjusted: weekly, January 29, 1971 - November 23, 1978 . . . . . . . . . . . . . . . Canadian long-term government bond yield adjusted: weekly, January 29, 1971 - November 23, 1978 . United Kingdom government 31% war loan yield adjusted: weekly, January 29, 1971 - November 23, 1978 . . . . . . . . . . . . . . French long-term public sector bond yield adjusted: weekly, January 29, 1971 - November 23, 1978 . Dutch long-term government loan rates adjusted: weekly, January 29, 1971 - November 23, 1978 Swiss long-term confederation bond yield adjusted: weekly, January 29, 1971 - November 23, 1978 West German long-term public authority loan rates adjusted: weekly, January 29, 1971 - November 23, 1978 . . . . . . . . . . . . . . Canadian long-term government bond yield adjusted: weekly, January 29, 1971 - November 23, 1978 . . United Kingdom long-term 3&% war loan yield adjusted: weekly, August 28, 1971 - November 23, 1978 French long-term public sector bond yield adjusted: weekly, January 5, 1973 - November 24, 1978 Page 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 Table Page F-20. Dutch long-term government loan rate adjusted: weekly, January 5, 1973 - November 23, 1978 . . . 261 F-21. Swiss long-term confederation bond yield adjusted: weekly, January 29, 1971 - November 23, 1978 . . . 262 F-22. West German long-term public authority loan rates adjusted: weekly, January 5, 1973 - November 23, 1978 . . . . . . . . . . . . . . . . 263 F-23. Canadian long-term government bond yield adjusted: weekly, January 5, 1973 - November 23, 1978 . . . 264 F-24. United Kingdom government 31% war loan yield adjusted: weekly, January 5, 1973 — November 23, 1978 . . . . . . . . . . . . . . . . 265 F-25. French long-term public sector bond yield adjusted: weekly, January 5,1973 - April 25, 1975; March 26, 1976 - November 24,1978. . . . . . 266 F-26. Dutch long-term government loan rates adjusted: weekly, January 5, 1973 - November 24, 1978 . . . 267 F-27. Swiss long-term confederation bond yield adjusted: weekly, January 5, 1973 - November 23, 1978 . . . 268 xi LIST OF FIGURES Figure Page 1. A schematic illustration of the time series analysis . 48 xii LIST OF APPENDICES Appendix A. ‘11 rn U n W o o o o o The Relationship Between the Forward Exchange Rate and the Future Spot Exchange Rate . . . . Time Series Forecasts--Short-Term Interest Rates Time Series Forecasts--Long-Term Interest Rates Time Series Forecasts--Long-Term Interest Rates Time Series Forecasts--Long-Term Interest Rates Interest Rate Series xiii Page 181 189 203 215 230 241 CHAPTER I INTRODUCTION The transfer of funds between savings-surplus and savings- deficit units occurs primarily in the domestic financial markets of a country. However, most financial markets have links abroad: domes- tic investors purchase foreign securities and may invest in foreign financial institutions. Conversely, domestic banks make loans to foreign residents who may issue securities or deposit funds with domestic banks. Such foreign borrowing and lending transactions have traditionally been directly subject to the public policy governing transactions in a particular market (6). Since the early 19605, the development of the Eurocurrency markets has removed both domestic and international borrowing and lending from the sole jurisdiction and influence of domestic finan- cial markets. Particularly, the growth of the Eurodollar market in the late 19605 and the interaction of domestic financial markets with the Eurodollar market have increased the range of borrowing and lending alternatives available to economic entities. The Euro- dollar market has become a link between domestic and foreign financial markets. The phenomenon of the Eurocurrency markets has raised such questions as: What is the nature of the connection between financial 1 markets? How sensitive are domestic financial markets to changes in external financial markets? Such questions have led to numerous studies on financial market integration. The studies fbcus primar- ily on the relationships between domestic and fbreign financial markets, using interest rate relationships and capital movements as measures of integration. Although the evidence is conflicting, the results tend to support the view that the financial markets are interrelated or integrated. Most of the studies, however, deal with the existence of integration rather than on the structure of any relationship. With the exception of Hendershott (45) and Kwack (55), there has been no test of the speed of adjustment of interest rate changes among financial markets. There is no evidence regarding the degree of integration that characterized the financial markets under investigation. Objective of the Study This research will examine the temporal and structural rela- tionships involved in the adjustment of interest rates to one another. Instantaneous adjustment of fbreign interest rates (holding-period returns) to United States interest rate changes is hypothesized. The hypothesis is set ferth after taking into consideration the find- ings of studies based on the interest rate parity theory and the concept of interest rate covariation. However, the hypothesis is contrary to the findings of Hendershott (45) and Kwack (55), who fbund some time lag in the adjustment of Eurodollar interest rates to domestic interest rates. However, the data employed in their studies were from the period of 1957-64, during which time the size of the Eurocurrency market was not comparable to that of the United States financial markets. In addition, the activities in foreign and international financial markets had not attracted the attention they have since the early 19605. Finally, the international monetary system is characterized by a system of fixed but adjustable exchange rates. The Bretton Woods Agreement of 1944 was enforced during the period before 1973. Specifically, under the Agreement, each country agreed that the rates at which current foreign exchange transactions were carried out within its territory were not to vary by more than plus or minus 1 percent from an established par value. A country could not change the par value of its currency except to correct a fundamental disequilibrium. The current study employs data from the l9705--a period char- acterized by conditions different from those existing in the 19505 and 19605. In 1971, the United States refused further convertibility of dollars into gold, which prompted some observers to say that the Bretton Woods system ended in 1971. The year 1973 marked the transi- tion to a floating exchange rate system. The need fer a transition to a different monetary system was found to be more urgent when a huge increase in the price of oil in late 1973 led to major shifts in the balance-of-payments positions for almost all countries. The oil crisis of 1973 precipitated the inflation rate spiral that also characterized the 19705. The large amount of current account sur- pluses in the oil-producing countries was an impetus to the development of the Eurocurrency markets as members of the Organiza- tion of Petroleum Exporting Countries invested their surplus funds in Eurocurrencies in European banks and dollar deposits in United States banks. Because of the dramatic fundamental changes that occurred in the 19705, this study will not necessarily present support fbr or refute earlier analyses of financial market integration. Rather, the study will provide eVidence of integration for the decade of the 19705. The study examines the speed with which arbitrage activi- ties minimize or eliminate interest rate differentials between financial markets. If the data show evidence of instantaneous adjustment of adjusted interest rates to changeS'hiother markets, this will serve to support the contention that financial markets are integrated. The remaining parts of this chapter discuss the importance and implications of the study, the background of the study, the definition of financial market integration, and summarize the pre- vious studies undertaken in the area. Chapter 11 presents the theories and concepts relevant to the study. Specifically, the interest rate theory of exchange rate expectations, the interest rate parity theory, and the relative purchasing power parity theory will be discussed. Chapter III outlines the research hypotheses and the pro- cedures to be applied in the testing of the hypotheses. Chapter IV discusses and evaluates the results of the tests perfbrmed on short-term interest rates. Chapters V, IV, and VII discuss and evaluate the findings of the tests performed on long- term interest rates assuming holding periods of three months, six months, and one year, respectively. Chapter VIII presents an overall evaluation of long-term interest rates. Finally, Chapter IX presents the summary of findings, con- clusions and implications, and suggestions fer future research. Importance of the Study The importance of the study is twofbld. First, on the micro- economic 1evel,the adjustment mechanism of interest rates to one another is a factor in decision-making processes. That is, the adjustment relationship provides investors with infbrmation on the potential risks resulting from the transfer of funds across boundar- ies. Moreover, an understanding of the adjustment mechanism will provide insight regarding the role of foreign investments in inves- tors' portfolios. With respect to borrowers, the nature of the adjustment process may provide infbrmation on the timing of the firm's financing. On the macroeconomic level, the research may have repercus- sions on the affected countries' autonomous pursuit of domestic economic policies. If the findings demonstrate a lead-lag relation- ship between United States interest rate and foreign adjusted inter- est rates, this will have an effect on the use of interest rates as a tool of monetary policy. However, should this study reveal instantaneous adjustment, it would not necessarily negate the use of interest rates altogether. On the one hand, the effectiveness of interest rates as a tool fbr changing domestic monetary conditions may be weakened. Such a weakening of monetary policy poses a serious stabilization problem for governments. On the other hand, monetary policy may be used to regulate the balance of payments, thereby freeing other instruments of economic policy fbr domestic economic objectives. If it is found that there exists a lead-lag relationship between interest rate changes, then the use of interest rates may be more effectively used as an instrument of monetary policy. The longer the lead or lag of the adjustment of interest rates to one another, the more effective interest rates will be in influencing domestic economic conditions. Financial Integration Defined The definition of financial integration adopted by this study fellows basically that of Kenen (11). That is, financial . integration is defined by the extent to which markets are connected.‘ This definition can be employed to describe the degree to which participants in any financial market are able and obliged to take notice of events occurring in other markets (11). Market partici- pants are able and obliged to do so in order to achieve their wealth maximizing objective. Although each country has its own particular set of domestic regulations, structure of interest rates, domestic currency, and cost structure of raising funds, the circumstances do not preclude the importance of interest rates in other countries as a significant contributor to the determination of domestic interest rates. The study will proceed on the assumption that different Inarkets have different specific risk levels, financial market struc- tures, and cost structures. But the characteristics specific to each financial market do not necessarily insulate the markets from external influences occurring in other markets. To the extent that external occurrences also affect internal economic conditions, the domestic and fbreign markets may be said to be integrated. Background of the Study The differences in timing of an economic entity's savings and investment behavior give rise to the creation of financial assets and liabilities. To meet an investment need for which it does not have ready capital, the economic entity can issue either equity or debt securities. In order for this economic entity to borrow or issue equity, some other economic entity must be willing to lend or provide capital. Such an economic entity would hold a financial asset because it had been able to accumulate savings in excess of its own investment needs. The determination of its willingness to hold a financial asset will depend on the perceived risk and expected return of the financial asset. The interaction of the provider and user of funds can be either direct or indirect. A direct interaction between the user and the provider of funds occurs when savers purchase the securities issued‘by ultimate users of funds. Financial liabilities such as bonds, commercial papers, and equity securities would fall under ‘this type of interaction. An indirect interaction between the user and the provider of funds involves a financial intermediary. Pro- viders of funds can invest in the obligations of financial inter- mediaries, who in turn lend the funds to users of funds. Whereas in direct financing the provider of funds is confronted directly with the credit risk of the user of funds, in indirect financing a financial institution acting as the intermediary interjects itself between users and providers of funds. The existence of financial intermediaries contributes to the fermation of financial markets, the existence of which helps in the process of funds allocation. Financial intermediaries, by bringing together economic units to satisfy their savings-investment surplus or deficit needs in one market, enhance the efficient allocation of excess available funds. Funds are allocated only to those deficit units offering the highest investment return for the level of risk. This interaction of the savings-deficit unit, the savings-surplus unit, and a financial intermediary contributes to the determination of the interest rate. Although the allocation process is affected somewhat by capital rationing and government restrictions, interest rates are the primary mechanism whereby supply and demand are brought into balance for a particular instrument across financial markets (10). The linking of the user and the provider of funds takes place not only in the domestic market, but also in the external financial markets. The interaction of users and providers of funds in foreign financial markets, however, is subject to the rules and institutional arrangements of the respective domestic markets. With the rapid growth of the Eurocurrency markets, the transition to floating exchange rates, the imposition and removal of flow of funds control programs (such as the Voluntary Restraint Program and the Interest Equalization Tax), and the increasing amount of capital flow movements, the question of the relationship of financial mar- kets to each other has been raised. This question has led to stud- ies on the integration of financial markets, which have mainly been focused in two directions: integration via interest rates and inte— gration via capital movements. The next section of this chapter summarizes the research done on the issue of financial market inte- gration. Literature Review Studies on financial market integration have focused mainly on the relationships between domestic and fbreign financial markets. The studies employed capital flows and interest rate relationships as measures of integration. The remaining parts of this section provide a discussion of the studies undertaken. Capital Flows as a Measure of Integration The rationale behind the use of capital flows as a measure of financial market integration is that, as interest rates in differ- ent countries change in response to monetary policy, capital will flow among the countries. The greater the sensitivity of capital flows to interest rate differentials and the greater the magnitude K 10 of those flows, the lower the degree of a country's independence (88). The studies employing capital flow measurements involve two dimensions--magnitude and time lag adjustment. Magnitude of capital movements.--Numerous studies have attempted to measure the sensitivity of capital flows to changes in interest rates between countries (90, 84, 68, 4, 89). A 1 percentage point covered interest differential in favor of the United States1 is estimated to have induced inward capital movements from as low as $210 million to as high as $1.25 billion over a six-month period. The estimates vary depending on the time period studied and the interest rates involved. Table 1 provides a brief summary of the studies undertaken, all of which were conducted using regression analysis. Explanations fer the varying estimates have also been attempted. Stein (68) suggests that the inconsistency arises from the nonindependence of the independent variable--the covered inter- est differentia1--from the disturbance term. Moreover, the existence of speculative pressure confbunds the interest rate-induced move- ments. Floyd (34) suggests that the relationship between interest rates and capital movements is a simultaneous one. Interest rates 1Interest rates are said to be covered when the exchange rate risk inherent in any transaction involving two different cur- rencies is eliminated by means of a forward exchange contract. Such contracts are bought or sold fbr future delivery against payment on delivery in national currency at a prearranged exchange rate. A covered interest differential is said to be “in favor of the United States" when the United States interest rate is higher than the fOreign interest rate after adjustment is made for the foreign exchange risk via a fbrward exchange contract. 11 TABLE l.--Studies on capital movements in response to interest rate changes Period Data Results Bell 1959-61 Short-term claims on A 1 percentage point (1962) fbreigners and covered increase in the United interest rate differ- States interest rate entials (monthly and induces: uarterly) (1) United Kingdom (1) Total movement of about $175 million. (2) Canada (2) No effect. Kenen 1959-61 Short-term capital flows A 1 percentage point (1963) and interest rate dif- increase in the United ferentials (quarterly Kingdom interest rate data)fbr the United reduces inflows of $260 States and United million per quarter Kingdom Stein 1958-62 Private fereign short- A 1 percentage point (1965) term capital and United increase in the United States-United Kingdom Kingdom interest rate Treasury bill rate dif- induces: ferentials (monthly data) (1) no speculative flows (1) Net inflow reduction of $462 million per year. (2) speculative flows (2) Net inflow reduction of $2.5 billion per year Branson 1959-64 Short-term claims on, A 1 percentage point (1968) and liabilities to for- increase in the United eigners and Treasury States interest rate bill rate differentials induces (1) United States- (1) Increase of $449 Canada million in fereign deposits over a six-month period. (2) Reduction in claims on fbreigners of $210 million over a six- month period. (2) United States- Eurodollar 12 TABLE l.--Continued Period Data Results Miller 1957-66 Long-term portfolio pri- and vate capital in the Whitman United States and (1970) (1) corporate bond yields in the United States and ten other countries (2) government bond yields in the United States and ten other countries Branson 1960-64 United States short- and term claims and Willett Treasury bill rates. (1972) of United States, United Kingdom, and Canada A 1 percentage point increase in the United States rate results in a one-time adjustment of $1,073 billion and a continuing flow of $21 million fer each succeed- ing quarter. A 1 percentage point increase in the fereign interest rate induces a one-time adjustment of $912 million and a con- tinuous flow of $18 million fbr each succeed- ing quarter. A 1 percentage point increase in the United States interest rate induces a reduction in the outstanding stock of short-term claims on fbr- eigners of about $300- $400 million in three quarters, and a reduction in subsequent flows of about $20-$30 million per year. he at l 1011: of 1 PW. 1161.3 nut in: My Quart ITEas Eumd 13 have an influence only insofar as they determine investments. Therefore, the relationship is also an indirect one. Using the magnitude of capital movements to measure financial integration does not provide a clear explanation of the effect capital flows have on the countries involved. Supposedly, capital movements will equate the marginal benefit of a dollar in the United States to the marginal benefit of a dollar in a foreign country. However, the length of time required for capital movements to induce interest rate equilibrium was not considered. Furthermore, the approach does not directly confront the issue of what ultimately happens in the countries involved (88). Time lag of capital flow adjustment.--Investigating the speed at which capital movements react to interest rate differentials fol- lows basically the same rationale as the measurement of the magnitude of capital movements discussed above. The speed of capital movements provides, however, an additional dimension. It suggests a direct measure of the degree of sensitivity that magnitude of capital move- ments does not provide. Miller and Whitman (89) fOund that one-third of the adjust- ment between the desired level of foreign asset holdings and the actual level existing at the beginning of the period takes place within one quarter. Branson and Willett (78) found the same time lag using Treasury bill rates and short-term claims on foreigners. Hendershott (45) used supply and demand schedules of the Eurodollar deposit and loan markets to study the relationship 14 between the United States Treasury bill rates and Eurodollar rates. It is posited that the increase in the United States Treasury bill rate will shift the equilibrium Eurodollar deposit rate in the same direction immediately, but the market Eurodollar deposit rate is expected to respond to the higher equilibrium rate only over time due to the gradual shift of the existing demand and supply schedules. The relationship was empirically tested with monthly data from 1957 to 1964. The findings are: (1) the Eurodollar rate will increase by .14 percentage point during the current month in response to a 1 percentage point rise in the mean United States rate; (2) the monthly speed of adjustment is .131 percentage point, implying that adjustment takes about a year: and (3) changes in the United States Treasury bill rates have a geometrically declining impact on Euro- dollar deposit rates. Kwack (55) extended Hendershott's study by including three European interest rates (namely, the Frankfurt interbank loan rate and the Canadian and United Kingdom three-month Treasury bill rates) as determinants of Eurodollar deposit rates. Kwack estimated that a period of twelve quarters is required for a 1 percentage point change in Eurodollar rates in response to a 1 percentage point rise in United States Treasury bill rates. One problem with the study is that foreign interest rates may also be influenced by United States Treasury bill rates and, hence, multicollinearity may exist among the variables. In the Hendershott and Kwack studies, the direction of the relationship was from the United States and/or fereign interest 15 rates to the Eurodollar market. This same approach was taken by Mills (97) and Argy and Hodjera (23). The latter studies were different, however, in that Mills included Regulation 0, the reserve requirement, ferward premiums, and the quantity of certificates of deposit as explanatory variables, whereas Argy and Rodjera included only Regulation 0. The studies undertaken by Hendershott and Kwack proceeded one step further than the previous capital movement studies. They examined the ultimate effect in one market due to interest rate changes in the other markets. This was done by an analysis of demand and supply adjustments to interest rate changes. Interest Rates as a Measure of’Integration Interest rates can be used to measure how two or more finan- cial markets are connected. The logic supporting the use of interest rates as a measure of integration is that as the interest rate level in one country changes, other countries will be affected in several ways. First, capital may flow in reaction to higher or lower inter- est rates. The greater the sensitivity of capital flows to interest rate changes, the less independent a country is in pursuing its economic goals. This paint has been discussed in the previous sec- tion. Second, exchange rates may adjust to counteract the effects of higher costs of trade and capital. This reaction to interest rate changes is discussed in the interest rate parity theory section that follows. Third, interest rates themselves may adjust in reaction to an external interest rate change. This occurrence is discussed 16 under the section on the convergence and covariation of interest rates. Interest rate parity theory.--Under the interest rate parity theory, interest rate differentials on assets of comparable risk are related to the forward premium or discount on one of the two currencies of asset denominations compared to the other. If the interest rate parity theory holds and international currency markets are efficient, the return on a domestic investment and the return on a hedged2 foreign investment should be the same. 'fl/,z-”"If financial markets are integrated, a rise in the interest Ifjrate differential in favor of one country will induce the movement of arbitrage funds, which will have the effect of increasing/reducing 3 on the domestic currency (23). The the forward discount/premium greater the sensitivity of arbitrage funds to changes in the interest rate differentials, the more integrated the financial markets involved may be. Studies by Stoll (70), and by Kesselman (51) indicate that the ferward premium adjusts by about 80 to 100 percent of the change in the uncovered interest rate differential. This finding implies that 2Hedging is defined as buying or selling forward currency so as to eliminate or exchange risk due to (a) normal international com- mercial transactions or (b) foreign investment of short-term capital funds 9 . 3A forward premium/discount exists if the forward exchange rate of a domestic currency is greater/less than the spot exchange rate of the currency. 17 a change in the uncovered interest differential is impacted in the): ; ferward rate via the ferward market. However, it is argued that "2\ induced capital flows partially offset the initial change in inter- est rates. Hence, the full impact of the interest rate change is I not shown in the forward rate. To account for the effect of induced capital flows, Caves \ j and Reuber (5) set up simultaneous equations determining ferward and \ spot exchange rates and the uncovered interest rate differential. The result of the study was contrary to the assertion of the interest rate parity theory. K\~, ‘\ Researchers have tried to provide explanations fbr such devia- tions of empirical data from the interest rate parity theory. 59" . Table 2 summarizes the explanations as set fbrth by each researcher ...—— t and the outcome of studies undertaken. ; The interest rate parity theory was found to hold true only}; in the Eurocurrency markets. This is not surprising, as the assump- tions of the theory are satisfied in such market environments. Convergence of interest rates. The second interest rate approach to the study of financial market integration fecuses on the convergence of interest rates. The less the divergence among inter- est rates, the greater the degree of integration will be. Investigations undertaken with this approach (79, 80, 23) mainly employ standard deviations and coefficients of variation of average domestic interest rates. Countries examined included the United States, Canada, European countries and Japan. Generally, the 18 .5: 2... .8 tags 252 3...... 2.03:8 -ocam veg .uucosu «n.3nuocau .uc—pcoam-ocam uc—ua 78.3.85 as: .3230 s: .ha¢~ as» voucoanam .mo—ucoccauogam as» ou ace—uu. -mcocu conga—abs cocosou sec .099: ac.v~cu - ac.»—aa< u_og a» has. as» case. .aumou m:¢.uu~uc.cu so. ac.um:nv~ ac. mo.u.cauom 832.389.2223“ 2.. 8:53-328 9.3: .pgx— ogu vooeoaaam uu—auoc veg nouns u—«onoa asp—Loamlocau we. «cap—aucsau eon: .»a¢_ use voucocaan napaaac .mo¢u.c:uou cacao ca—_I,m v:- m*cna c. toamm— mo.u,c:uoa voua:.-ueovcvcaoa can gonzo; c. toga“. «0.».Lauou coo-:.socoouco——oe co moo-L can: .ucaooa.us-:_nuca engage» o>.uuoauos 5.0:» uc- nouns 35323 3 32:2. 5:. 22 3 2.8.. a! S: ..§.§8§ as: .3238 s: .coocoa can ago» so: cognac; un.xu woe—caucoaao oo-Lu.ac~ ope-u.uocn as case» .moagc ac_umosc_ sea a:— -noccoa .a.a:osouu.u we. muuou ace—unancocu so; toumanu< .ucaoun’uxusvnuga uchcoe we. mouse ——.a scam-och ca-u-eau use ..x.= ..m.: new: con: saga .uucau ua-cu.aca co aculu>ol cg» qua—:I_um cu cause ea Laos can uc—.o voouxo «was po.ucoco~c.e nuns amocouc. «5 o... “gunfignra 22:8 «5 6.38 32% 2: Agog-«cove: avaun pau.c.alu oz. .33.... 533 3...: 23:5 5 «save new mace-osol paa.a~u ago» ouggm co 23:92:05 pact-538 .u—aavov be munou town—u -onma as» we. man's he oucouu.xu .mco.u.c.uoe «engage-c. we. mega-ucamuul c. acoccu .8933..- o:.u:uoxo :. was. ecu .muosL-n «so.c~> c. mo.u*u.umo—o apaaam was veg-ac .uumou meo.uu-mcacp .uposo— xn_c u:¢c~u~.o we as»: no.uaum mao.>oca c. can: nouns amuse»:— .u—oso— xn—L aeoco~»_u we use: no.uaum «so.>oca c. you: moans unaccus— ..vov.soca co.a-=-—axu as. .m:o_u¢auoaxu ¢,.u-—aooam 3:28:25 “or... .338 .numou n:o_uu-mc-L» «was; ounce-95¢ yo hu——.a -~—.a»- age so nu:.acun=¢u pace—uau.uu=_ Ga: :38... Ga: aging veg copuc $3: 288.... we» coco< .msopv gu..os van ~0aeocm :8: “as: as: he: as: 55. :3: 28m 88: 5:28.... 38: 82.2.. 38: shuffled $2 813 :8. 22: .—— .c— :53..." «5 co 352.: :o_u~ca—axm nears» mu.can can; «masque. co no.uaum--.~ m4mmo vcmucaum pmuoz mmrcmm mwumL HmmLmucr Emulfihozmllmpwvos mvawm 02:. wflmwsm>wcall1m M153? 55 NNF 1 :11 ounces a F1» N «Noam; xcmncmch cum. N 1 N 1 N :pcOE1mmcgu mszm .m p F1» N mum; xcmncmucF mNN. a 1 N 1 N ebcos1mmeeb guess .N 1 mum; xceacmucF oFm. we 1 F NN 1 FN gpcos1mwcsp cucmcu .o a F1» u can; xcmncwucF specs mam. a 1 N - N 1mmeeb soumch umecs .m 1 man; Luann oucocFF emu. pm 1 F FN 1 FN sacos1mmggu :NFuNcou .e N N1» F1» a mans xcmncmacF CNN. a 1 N mFN. + N NFN.F 1 N =N=o513megb eascma Nam: .m mmF. be 1 NLN eeN. + F.NN eeN.F 1 HN acme LNFFoeoeam eb=051mmegF .N . p F1u u mFFFm zeammmcF NFmF. a 1 N - N =N=051mmcgb mmbacm umeea .F FmeF 1 av mNmF .NN Ngaaeea 1 mNaF .mN FFea< "m vowema .0 mFmaumem mo quo: . mochm eoFbermo ecaeeacm emser=OU11.m NSNou ELNF1mcoN guNso .m uFmF> vcom couumm mmF.F F1NN mom. + NN 1 F1NN 1 NN uFFnza ELNF1mcoN gunmen .m N 1N NFNFF eNoN cN3 NON. N 1 N N 1 NN Nmm NNNENNN>ON .e .3 .e 1 NFNFF Ncom Newsccm>ow omN. NN 1 F NN 1 NN ELNF1meoN cNFNNeNu .m NNNm F1N N F1N N :NoN aNFcogN=< NFFNNN Nam. N NNN. + N 1 N 1 N ELNF1meoN cNEme Nmmz .N u. Flu. Fla 0 Upmp> tcom hflszflwz NmoF. N 1 N mNN. + N NNN.F 1 N NNNNmeou NNNN1NF .m .= .F ANN 1 av NNNF .N NFaN 1 FNNF .NN NNN==NN "F NoNema .< NFNNNFNNN No . coFNNF>No NLNNNNNm FNNoz mNFcNm chczNNc qucmg mcFNFo; sNeoE1NNchV mNNNL NmmgmNcF ELNN1mcoN11mFNNoE mchNm NEFN NNNFLN>FN=11.FF mNmou ELNF1mcoN NNNNQ .o N F1N N NFNFF Ncom coNumm NNN.F N 1 N 1 N NFFnaa ELNF1N=8N NNNNLN .N N F1N N NFNFF cNoN ch mmm. N 1 N 1 N mam N=Ne=2N>ou .x .3 .N 1 1 1 NFNF> Ncom Newsccm>ow NON. FF NN Noe. 1 NN 1 N NN NNN. + F NN NNN.F 1 NN ELNF1N=8N NNFNN=NN .N NNNN F1N N F1N N cNoN NNFNNNN=< NFFNNN omm. N mNF. + N 1 N 1 N ELNF1mcoN :Nscmu Nmm: .N N N1N F1N N NFNFF Neom NNFNNNNz NNNo. N 1 N NNN. + N NNN.F 1 N NcNNmeoo L.NNNNANF .w .3 .F FNm 1 cv NNNF .e LNNoNuo 1 man .m ngscNN "N Nocha .m NFNNNFNNN No :oFNNF>No NNNNcNNm anoz NNchm emscFN:OU11.FF mNmoo ELNF1mcoN NNNao .m NFNF> econ N F1N N coNuNm uFFasa mam. N 1 N N ENNF1mcoN suemem .m N F1N N NFNF> Neom wa. N 1 N N «mm Newsccm>ow .x .2 .N N N1N F1N N NFNF> Ncom Ncmsecm>oo mmN. N 1 N NmF. + N NmF.F N ENNF1mcoN :NFNNcNu .m NNNm F1N N F1N N :NoN aNFcogN=< NFFNNN mnw. N oFo. 1 N 1 N N ELNF1m=oN achLmo Nam: .N N F1N N NFNF> Neon NNFNNNNz mooF. N 1 N N NcNchou cme1oF .m .2 .F FFNF 1 :v wNmF .mF NNNNNNN 1 E: .N .3: "N 828.. .N NFNNNFNNN No FNNoz NNFcNm coFNNF>Nn NNNNNNNm vmscFNzou11.FF m4m)* = 1 .0605 a (1) g (1.17gé.173))1 . 1 .0470 O (1) = (.39056.10))I . 1 .0411 87 deviation of residuals as expected, with the second subperiod exhibit- ing more volatility. In the models characterizing the process generating the Canadian goverment bond yield, there is a strong difference in the models over the three subperiods. The first subperiod is a random walk model. The third subperiod is characterized by an autoregressive integrated model. The parameter coefficient in this model may be considered insignificant in the economic sense although significant statistically. This implies a "noisy" random walk process. The model for the second subperiod is very different from many of the models in the study. It can be described as an autoregressive inte- grated moving average model with first order regular differencing. The moving average parameter in the model is an eleven order seasonal moving average parameter which is very unusual. The magnitude of the parameter, however, is relatively small. Hence, the model may be considered an autoregressive integrated model. The behavior of the standard deviation of residuals turned out as expected, with the second subperiod showing a higher value. However, the magnitude of the statistic are close to one another, lying in the range of .230 percent and .310 percent. The United Kingdom war loan yield can be described by random walk processes with the second subperiod exhibiting more volatility than the first and third subperiods. Among all the time series con- sidered in this section, the French public sector bond series has the largest standard deviation of residuals, hovering at about the 88 1 percent level. The French long-term rate series are random walk models in the second and third subperiods, but is an integrated moving average model in the first subperiod. For the Dutch long-term government loan rate, the models found to describe the series are autoregressive integrated model for 1971-72 and 1975-78 and a random walk model for 1973-75. The auto- regressive parameter values are tested to determine statistical difference and, as shown in Table 14, the autoregressive parameter values are found to be different. The first subperiod is essentially characterized by a "noisy" random walk model. As expected, the 1973- 75 period showed more volatility than the other time periods. Finally, the Swiss long-term confederation bond yields are examined. Note that the first and third subperiods are characterized by autoregressive integrated models. The significant parameter coeffi- cients in the first and third time periods were determined to be insignificantly different from each other as shown in Table 15. However, the values are rather small to have any economic meaning. The underlying stochastic process in the second subperiod is a random walk model. The standard deviations of residuals confirmed the expected higher volatility during the second subperiod. A general observation that can be made from Table 11 is that the standard deviations of residuals in the first subperiod are higher in comparison to those of the third subperiod. The exceptions are West Germany and the United Kingdom. In many instances, the magni- tudes of the differences are quite small. 89 TABLE 14.--Comparison of ARIMA (1,1,0) models fitted to the first and third supberiods of the Dutch long-term overnment loan rate (Three-month holding-period return) Period n o A = l-¢ o(A)* 0 df First Subperiod 1971-72 76 .138 .862 H- .0396 42.4 35 Third Subperiod 1975-78 141 .442 .558 H- .0418 32.2 35 Standard error of A(]) - A(2) = 1/(.O396)2 + (.0418)2 = 1 .0576 .304 1“) - 1(2) = .862 - .558 Therefore, there is a real change in A. van-(313,411).i 6(1) = (Ewifi = 1 .0396 6(A)==(e§§§5=55§l)* = 1 .0418 90 TABLE 15.--Comparison of ARIMA (1,1,0) models fitted to the first and third subperiods of the Swiss long-term confederation bond yield (Three-month holding-period return) Period n o A = 1-0 o(A)* 0 df First Subperiod 1971-72 76 .225 .775 1.0479 30.4 35 Third Subperiod 1975-78 141 .269 .731 1.0373 35.7 35 Standard error of 1(‘) - 1(2) = /(.0479)2 + (.0373)2 = 1 .0607 1(1) - 1(2) = .775 - .731 = .0440 Therefore, there is no real change in A. 110(4) = (19121)); 0(A) = (:Zlgéeggélli = 1 .0479 0(A) = (.731(.269))§ 14] 1 .0373 91 Table 16 provides the Theil's U Coefficients for each of the time series models. All the models can forecast better or at least as well as a naive, no-change extrapolative model except for the West German model for 1975-78 and the Canadian model for 1973-75. Nevertheless, these two models are selected because they best char- acterize the time series. TABLE 16.--Theil's U Coefficient for the univariate time series models--Long-term interest rates (three-month holding period return) Period 1 Period 2 Period 3 l. U.S. lO-year Constant Maturity Bond Yield .830 .900 1.000 2. West German Long-Term Public Authority Loan Rate .991 .962 1.533 3. Canadian Long-Term Government Bond Yield 1.000 1.164 .990 4. U. K. Government 31% War Loan Yield 1.000 1.000 1.000 5. French Long-Term Public Sector Bond Yield .782 1.000 1.000 6. Dutch Long-Term Government Loan Rate .992 1.000 .911 7. Swiss Long-Term Confederation Bond Yield .956 1.000 .975 Cross Correlation Analysis The findings from the cross correlation analysis are tabu- lated in Table 17. Although the time series are cross correlated 92 TABLE l7.--Cross correlations of weekly interest rate5--January 29, 1971 - November 24. 1978 (Three-month holding period returns) (Input is U. S. lO-Year constant maturity bond yield) mtput West German Canadian United Kingdom French Dutch ' Swiss La Long-Term Long-Term Government 33% Long-Term Long-Term Long-Term 9 Public Authority Government War Loan Public Sector Government Confederation Loan Rate Loan Rate Yield Bond Yield Loan Rate Bond Yield A. Period 1: January 29, 1971 ~ July 7, 1972 (n I 76) ~13 .138 -.103 ~.118 ~.128 .135 .030 -12 .007 -.001 -.088 .042 ~.080 .063 -ll ~.033 .013 .009 .198 .012 .100 -10 .032 ~.032 .021 ~.063 -.015 -.130 - 9 -.121 ~.097 ' .068 ~.085 .005 -.084 - 8 ~.006 .091 .062 .021 -.011 .219* ~ 7 -.l94 -.050 ~.017 .131 .015 ~.04O ~ 6 .000 -.073 .019 .012 .060 .087 ~ 5 -.021 .013 .007 .058 .156 ~.135 - 4 .031 ~.072 .007 ~.150 .008 ~.007 ~ 3 .110 -.003 .013 .050 .081 -.019 - 2 .193 ~.021 -.048 .015 ~.018 .066 - l .000 129 .039 ~.081 ~ 077 - 023 0 -.095 467' ~ 008 .115 ~ 125 150 + l ~.196 ~ 209 -.047 -.151 .051 - 357* + 2 .065 147 .075 113 056 + 3 .165 .085 .070 .003 - 008 103 + 4 ~.103 ~ 098 005 -.182 ~ 040 ~ 136 + 5 .178 .068 - 018 ~.016 083 .085 + 6 .001 044 025 .025 219‘ 009 + 7 -.144 .028 -.059 ~.036 ~.089 .053 + 8 .124 .150 .089 .014 ~.164 ~1041 + 9 .051 ~.030 ~.072 ~.034 .110 -.096 +10 .025 .045 ~.106 ~.016 ~.033 .021 +11 .111 .216* ~.006 .013 .071 .197 +12 ' ~.287' -.167 ~.055 ~.088 ~.059 ~.169 +13 .204 .127 .015 .041 .051 .112 8. Period 2: January 5, 1973 - October 4, 1974 (n I 92) ~13 .023 .060 .014 ~.086 .026 .010 -12 -.052 -.137 .142 -.083 ~.042 .028 ~11 .020 .055 .068 -.012 .072 ~1054 ~10 .024 ~.012 .083 .016 ~.039 .028 ~ 9 .070 .066 .218‘ -.106 ~.002 .003 ~ 8 .077 .088 .187 ~.122 .105 .098 ~ 7 .071 ~.129 ~.013 .026 .065 .160 ~ 6 .062 ~.045 ~.045 .115 ~.029 ~.074 - 5 .026 0093 .0052 0028 0M2 02%. - 4 .080 .041 .148 -.022 .140 .240. - 3 .029 .181 .142 .076 .156 .146 ~ 2 .124 .051 .154 .077 .090 -.000 ~ 1 .182 .085 .122 1219‘ .073 .162 o .159 0204. .00“ e017 .092 ‘007‘ + l ~.112 .189 ~.004 .017 .092 -.074 + 2 .037 -.053 -.003 , .105 -.081 .180 + 3 .001 ~.091 ~.020 ~.063 ~.053 ~.243‘ 93 TABLE 17.-~Continued Nest German Canadian United Kingdom French Dutch Swiss La Long-Tenn Long-Term Government Long-Term Long-Term Long-Term 9 Public Authority Government 31: Her Loan Public Sector Government Confederation Loan Rate Loan Rate Yield 80nd Yield Loan Rate Bond Yield + 4 ~.002 .180 .022 ~.089 ~.057 .127 + 5 .016 ~.001 .097 .002 ~.034 ~.017 + 6 .000 .191 .030 ~.010 .082 ~.142 + 7 ~.162 .007 ~.074 .048 .091 .102 + 8 ~.022 .030 .049 ~.170 ~.001 ~.Oll + 9 .099 .043 .003 .125 .183 ~.001 +10 .056 ~.087 .009 ~.06l ~.lll ~.033 +11 .085 ~.ll4 .053 ~.037 ~.015 .109 +12 ~.064 .031 ~.053 .147 .001 ~.023 +13 .124 ~.046 .051 ~.206‘ .009 .096 C. Period 3: May 9. 1975 - January 13. 1978 (n I 141) ~13 .016 .099 .132 .111 ~.018 .100 ~12 .029 ~.008 .028 .084 ~.001 ~.008 ~11 .043 .078 ~.l47 .004 .105 .021 ~10 .026 .032 .210' .011 .016 ~.065 ~ 9 ~.021 ~.056 .000 ~.012 ~.100 ~.063 - 8 .013 ~.024 .011 .031 .003 .083 ~ 7 .030 ~.038 ~.029 ~.Ol7 ~.066 ~.009 ~ 6 ~.048 ~.081 ~.016 .038 .074 .109 - 5 ~.050 .125 ~.153 .043 ~.093 ~.142 ~ 4 ~.005 032 ~.058 015 015 ~.007 ~ 3 .047 053 .097 .104 ~ 151 038 ~ 2 .133 .000 139 ~.053 ~.03l ~ 040 ~ 1 ~.015 .208* 120 .093 .167* ~ 009 0 .532‘ .262' 110 .065 216* 188‘ w 1 ~.l70* .011 ~ 143 .057 192* 135 + 2 .016 .181* ~.115 ~.060 .003 .068 + 3 .064 .115 ~.066 ~.032 ~.088 .148 + 4 ~.033 ~ 169 .038 .030 ~ 133 143 + 5 .005 ~.014 .052 .144 ~.06l .054 + 6 .062 .067 .023 .083 .043 ~.019 + 7 .001 ~.047 ~.051 ~.075 085 - 041 0 8 ~.064 .020 .139 ~.114 093 ~.052 + 9 .083 .067 .107 ~.09l ~.014 ~.102 +10 ~.129 .054 .106 ~.066 ~ 021 ~.012 +11 .028 .028 .002 .042 145 .125 +12 ~.049 .058 .022 ~.022 .086 .051 +13 .091 ~.151 .031 ~.003 .133 .044 *lndicates that the correlation coefficient is statistically significant at the 95 percent level. 94 over an interval of plus thirty-six and minus thirty-six weeks, only the interval from minus thirteen to plus thirteen weeks is included in the table. In the 1971-72 period, only three time series show some relationship with the United States bond yield. These are the West German, Swiss, and Canadian time series. In the United States-Nest German cross correlation analysis, a significant coefficient exists at the plus twelve lag. This implies almost a three-month lag relationship.. However, the sign of the coefficient is contrary to expectation. The impulse response weights do not exhibit any discernible pattern. Moreover, the para- meter estimates based on the impulse response weights are small, suggesting little economic significance. Hence, the coefficient is most likely spurious. In the United States-Switzerland cross correlation, the only significant correlation is at the plus one lag. However, the sign of the coefficient is negative which implies an inverse relationship between the United States bond rate and the Swiss confederation bond rate. Although the significant coefficient at the plus one lag appears to be spurious, the impulse response weights are quite siz- able. Hence, the United States-Swiss relationship will be examined further in the next section on transfer functions. In the case of the United States-Canada relationship, the coefficients at the zero and plus eleven lags are significant. They also bear the correct sign. Since the coefficient at the plus eleven lag and the corresponding impulse response weights are small, the 95 -coefficient is likely spurious. The coefficient at zero lag implies a contemporaneous relationship between the two time series. The second panel of Table 17 shows the lead-lag relation- ships for 1973-75. The United States bond yield and the French public sector bond yield are found to be significantly correlated at the minus one and the plus thirteen lags. The significant coefficient at plus thirteen lag suggests a three-month lag. Since the coefficient bears the wrong sign, the coefficient is most likely spurious. The sig- nicant coefficient at minus one lag suggests that the French public sector bond yield leads the United States bond rate by one week. The magnitude of the coefficient is small and barely significant at the 95 percent confidence level. Moreover, when the impulse response weights are examined, the pattern cannot be interpreted to be of any possible structure and the magnitudes of the weights are rather small. This implies that the estimates for the transfer function parameters are small, indicating little economic significance. In the United States-Swiss cross correlation analysis, signifi- cant coefficients are at the minus four and the plus three lags. The plus three lag coefficient carries a negative sign and makes no economic sense. The coefficient at minus four lag implies that the bond rate lags behind the Swiss confederation bond yield. The coef- ficients, however, is small. The impulse response weights are also small and exhibit no discernible pattern. Hence, the coefficient is likely spurious. 96 The behavior of the United States-Canada long-term interest rate relationship should be noted. The only significant coefficient appears at the zero lag. Although the coefficient is small and barely significant at the 95 percent confidence level, the coeffi- cient implies a contemporaneous relationship between the two time series. The impulse response weights are sizable. Therefore, in the next section, the relationship will be examined fUrther to determine if the coefficient can be employed for forecasting and decision-making purposes. The analysis of the third subperiod, 1975-78, is found in Panel C of Table 17. Except for France and the United Kingdom, the table shows some relationships existing between the time series examined. Note that most of the significant cross correlation coeffi- cients are at either minus one lag, zero lag, or plus one lag. This suggests that the relationships are mostly of a contemporaneous nature. Also, the significant coefficients are mostly of the correct sign. The significant coefficients in the United States-West German relationship are at the zero and plus one lags. The plus one lag coefficient does not have the expected sign. Moreover, the magnitude of the coefficient is rather small. These suggest that the coeffi- cient is most likely spurious. The significant coefficient at the zero lag implies that the West German public authority loan rate is contemporaneously related to the United States bond yield. The significant cross correlation coefficient at the minus ten lag fer the United States-United Kingdom relationship suggests 97 that the United Kingdom rate leads the United States rate by ten weeks. The coefficient is rather small and is barely significant at the 95 percent confidence level. Moreover, a lead of ten weeks has little economic meaning. The United States-Netherlands cross correlation analysis shows significant coefficients at the minus one, zero, and plus one lags. Even though the coefficients are not large, the implication is that a feedback-feedforward relationship exists between the two time series. The same situation exists for the United States-Canada cross correlation. The existence of significant coefficients at the minus one, zero, and plus two lags signify a feedback-feedforward relationship. Finally, examine the cross correlation analysis for the United States and Swiss long-term interest rates. The only signifi- cant coefficient appears at the zero lag. The coefficient bears the correct sign and is significant at the 95 percent confidence level, thereby indicating a contemporaneous relationship. From an evaluation of the coefficients, it appears that the significant relationships are not strong except for the United States- west German relationship. The magnitudes of the coefficients are relatively small. This raises doubt as to whether the relationships have economic meaning. Similar to the relationships found on the short-term interest rates, there seems to be a change in the corre- lations from one time period to another. In the first subperiod, the lead-lag relationships were of a very low level. In the second 98 subperiod is found either a low level relationship or none at all. The third subperiod shows the increase in the interaction of the different financial markets as countries start experiencing more flows of funds with the easing of restrictions and controls between countries. This is evidenced by the larger number of significant cross correlation coefficients at the zero lag. The cross correlation results in this study do not provide clear evidence of high comovement between holding-period returns. Although the analysis indicates some correlation between the various time series, the comovement is not of a sufficient magnitude to justify a conclusion that the holding-period returns exhibit a strong relationship. Moreover, the degree of comovement is not consistent over the entire time period and for the sample under study. The only time series consistently showing comovement with the bond yield over the entire time period from 1971-78 is the Canadian long-term govern~ ment bond yield. However, the degree of comovement is not stable over the period. The first subperiod is found to have the highest cross correlation coefficient. Multivariate Analysis This section will discuss the transfer functions suggested by the cross correlation analysis performed in the previous section. The transfer functions modeled are shown in Table 18. The inclusion of the input series in the estimation of the 1971-72 Canadian long-term government bond rate reduces the standard deviation of residuals of the output series. The use of the United 99 um Ammwmw . Fv u.x mm~.- u a» 8pm?» econ :ovumgmumwcou scmhumcob mmwzm cam. on Ammpep - Pv F5.x mmm. + px pom. u a» mama smog “cmsccm>oo ecmh-aco4 guuzo P-8a emu. + mNN. F-» . a u c—ww> x «mm. + x ope. u a ucom pcmsccm>ow scmhumcob :mPcmcmu Fupm mam. pa + Flux oou.m com. a up a mama cue; + x o-.~- n F x «mm. u » xuwcospa< uppnsa Ecmh one; eczema 9mm: ».e_u cv mum. . .m_ acaacaq - msmp .m .m nopcaa .u a akmmnmwnulrv .A m_~m. - 2v mam. +8 __ a x mNm. uF » mmm. . u» upmw> ucom acmsccm>oo Ecmkumcog curvacmu Amm . cv chap .5 aanouoo - memp .m zeaacaa "N cocoa; .m mom. um . mmmm. ax own. u a» quw> ucom cowumcmvwmcou Egmhumcog mmwzm uuem. Pups Pmm. + as + u.x on~.p n a» 8.6?» ucom pawsccm>oo seaplane; cmpumcmu A85 u av Nump .N »_=a - .KmP .mN zgaacaa up costaa .< mpmavwmmm mo . cowpmw>oo ucmucmum —muoz mmwcom Amzcauwc copgma mcpvpo; cacoEImmcgbv moans ummcmpcv Ecmpnmcodaampmuoe mwpcom we?» mumwcm>wupaziu. mp u4m ccom coppmcmcmmcou mme. an n F.aN - NN Ecmh-mco4 mmvzm .N mama coo; acme -- uccm>ou sgmhimco; sense .0 upmw> ucom couumm .. oppose enmeumcod gunmen .m N p-» a «u—mr> smog co: awn Nmo.P a u N - N newsccw>ow .g .2 .e a p-» a upmw> econ acme=Lm>ow mmm. m u N u N Ecmhumcob newnocmu .m a p-“ a mama :mog »N_cozps< oansm omm. N u N u N Esmeumcog custom paw: .N p N-» P.» a 8P8?» ucom xuvcsumz Nmop. a u N mmm. + N mm~.~ . N penumzou cmmzuop .m .3 .F AoN u cv NNm. .N Npaa - :2 .8 29:2. x 828.. .< mpoauvmmm co pane: mmwcmm copuaw>mo vcmucmam Amcgzumc newton mcpupo; genealxpmv--mmaac umocmucr Ecmpumcop mpmuos mawgom we?» mumvcm>vcauu.- u4m - acam =o_pacaumccou no u m-uN + euaN - p NN - 8N Eco»-mco4 mmwzm .N . muam cue; acme P “N . NN -ccm>oc ecmp-mco4 gonzo .m - u_av> econ coaoam _ HN - HN uvpnaa stap-m=og negate .m pip Rpm. a mam. I one. “a Nmm. m-ua N_N. - as new. viva new. I an n muuN me. + mauN mom. + upmw> cue; ca: N - N-5N Fan. - .-8N mom. - u.N Nmm geesecasou .x .3 .8 u—mm> ucom Newsccm>ou u FuuN . pN Ecmhamcog cawumcnu .m mung u coo; auwcogu=< owpaza P 9N . NN Ecmhumcog cascmw umm: .N -8 upaw> neom praapaz N mme._ . N penumcou cmm>uop .m .2 .p “No u av mNm_ .4 canopoc . mmmp .m xgmzcmu ”N nopcma .m «mm. 46 ll mun. Rumo. an n NuHN mud. + P mpmauwmmm co =o_um_>mo ucmucmum Paco: mmvcwm vmacrucouun.- m4mo2 scmpumcom guano .2 2pmw> 2:02 couumm uwpazm scmpumco2 202mg; .m 2pmr> cmom cm: 2mm ucmsccm>ou .x .2 .2 2pm?» 2:02 u2m522m>ou Ecmpumcom cmwumcmo .m mpum =mo2 zpvcosu2< umpnam Esmeumcom :mELmo umm: .N 2pmw> ucom »u_c=umz ucmpmcoo cmmanop .m .2 .P APeF u :2 mum? .mp xgmacma u msmp .2 an: “m vopsmm .2 mpmauwmmm mo cowumm>m2 coaucmpm Pmuoz mmpcmm um==.p=ou--.NN m2m ucom couomm uppnza scmhumcoa zucmca e,m_> caou an: 2mm Hausaaasom .a .2 2pmp> ucom acme -ccm>o2 EsmeumcoA cmwuucmu mama :mom Aupco222< owpnza ecmhumcom cmecmm 2mm: “Na n =2 eNmF .5 2828882 - mNm_ .m agascaa "N coeaaa .m 2pm?» ucom acme lecm>o2 scmpumco2 cmwumcmu mama cm02 zuwcozu2< oppnza Ecmhumco2 2552mm «mm: A25 n :2 «amp .5 ap22 1 pump .mm zum2cm2 up vowgma .< mpmscwmma mo :owumw>m2 vcmucmum Page: mmwcmm Amcszpmc newsmaumcpupoc gases-xwmv mmpmc ammcmucw Ecmuumco2uimpm2oe mmvcmm ms?» mumrcm>wupzzuu.m~ 2224* 124 upma> 2:02 coaa mom. am + ax mac. n a» -mcmmmacou Egmaumcoa amazm Nua a _ a a mama cmoa acme mma. m amm. + m + x omm.a u » -ccm>oo Ecmhnacoa amaze a a a 2—ma> mcom acme 2mm. m + x om~.a u a -cam>o2 Egmhuucoa :mammcmu _-a . l a m can m + mama Nua Fua a a cmoa aaaco2a2< oaaaaa mmm. x oaN.N - x o~o.m + x omm.a- u » Emma-m=oa custom ammz ape, n =2 mNma .m. aaazcao . mama .m 2m: um moacma .u mamzmamma 2o coaama>m2 mmmozmam ammo: mmacmm 2m2cwacoonn.mm mamOw ELwPIQCOA guano .m 2pma> mama moaomm uaansa Egmhumcoa gamma; .m mama» zoom cm: umm acmsmcm>o2 .a .2 .e 2pma> 2:02 acme uccm>o2 Emmhamcoa cmammcmo .m mama :moa 2aa2o2a2< uaamaa Emmaumcoa :mecmw ammz .N mama» mcom xaaczamz acmamcoo cmmxnop .m .2 .P ANm n =2 eNma .e ammoauo - «Nap .m aamscmm "N moaama .m AmN n =2 NNm. .N aaza - _Nma .mN hamscam "a uoaama .< mamamamma mo coaama>m2 mammamam Paco: mmagmm Amcmsamg moagmaumcaupo; mmm21mcov mmamm ammcmaca Egmaumcoaunmamuos mmacmm meaa mamacm>ac2uu.m~ mam<~ 136 o¢p u cc vamp» econ coaaaaa - aN -mmacoo sampumcoa mmazm .N . mama cmoa acme moa. am u a aN - aN -=2m>om Eama-m=om amaze .m .uamaa econ Loauam Nam. m n mme. a-am Nme. + an n N-aN mmm. - a-aN Nmm. - am uaaaaa saaa-m=om gucmaa .m a pua a 2Fma> :moa em: ace. m n N u N mam acmecmm>ow .a .2 .c .a 2Fma> 2:02 acmscam>o2 New. am n a N . aN Ecmhumcoa cmaumcmo .m mama a pua a :moa a«$35.2 oapmza mam. m u N u N Ecmhimcoa cmecmu ammz .N a _.a a mama> 2:02 aaacmamz cop. m n N u N acmamcou mmmauo— .m .2 .p :3 u 5 £2 .2 25.2. . mama .2 am: an moacma .u mamamamma mo ammo: . mmammm :0 3b tawn— Pam—2cm...“ mmacaacouuu.m~ mam3 Ema-2.8.. 28:2 coma-p .-a a a-» N 58.3 2.; 2.8 .388 amp. m Nvm. + m a N - N gamma-n magma; snob-2:62 gunman amen-p . a . p-a - a - ~22 p am p-aN aN gamma 2 2—ma» :moa em: NoN. m a N - N sauce-n «an aemeccm>mu .x .2 Emma-— . a a —-a - a - 8N a. 3N N 58.. 8 Ba; 28 cow. m u N - N gamma-n aemiccm>ma scab-2:64 cmammcmu am - P-aN - aN cmma-p onm. am - —-aN - aN ga295r2 p-a a p-a a mama cmoa aaamogaa< can. m NNN. + m u N - N gamma-n capasa Emma-2202 cmssmw amm: . v No» piu 0 «Us I 5 th— as 8...... «22p m a N now. + N m2~._ - N - pan. .uN sumaema N— 26.5mm .< m—mamamma mm c coaama>m2 mammemam ammo: onnwpwn “macmm «mama ammcmaca Emma-meoa-mpmmoa «magma meaa mamacm>a:2-.om uaa econ ca.» . N p-N N «3 a a N N 5.3....” -2885”. 23-98.. 3:.» mac. No a .-NN NN sum»-— . N ..N N .. cmo N u u N N sagas 0 can: coo; acme «as. as u — NN NN gueosin ucsu>ou snob-ucoa gauge mom. a. u .-NN NN sans.— . a-» . - N n .-N N 5 N :N N. .-NN NN 5:2va 3o: 28 .338 one.— a a N N saga-um u_pa:¢ use»-ocoa sagas; 8'. Pin“ @8- U a” I "0“” 8.. + NluaN ”FF. I Flu" ”Pm. P “N L.g-P N3. WNN 8n. + .-NN .. N-NN an. .. TNN a8. NN 58.3 EN; 53 O a P-» u 0 I «an N u N N gacosrn so: n.n Ncuszgo>oo x a new. a. u puuN NN Lama-p . N .. ..N N ..8. Nun —_-N N N-» p-” N NN g» o v.0.» scam ago: com. a now. - a a N «on. + N con.p N suco:-n -cgu>om snob-acoa an_oacau —Ne. a. - ~-NN NN unusup . N . ..N N - NNN 7N N. .-NN NN 55.. N 35. 53 3.352 amo. a «up. + a a N N suco¢.m NN—aaa ssoh-ocoa cassow «no: . N N-N .-N N ANN .. 5 :2 .N ..338 =8 N . N at + N 8... N - 22 .... ENS... "N 822. .N upoau.mo¢ No pave: no.5»; 83.28 2.2.3. 9:29. vozcvucouuu.on ma¢ No muco.u.uyoou NcN .Nco.uN.»uo szchuu .nanz--.Nn uau NNN. N.N. N.- N.N. N - .Ne-N .Nz N.N ENNN=.¥ NN..NN NNN. N ..N. N NNN. N - .<.z ...-. N... ..+ N.N. N NNN. N NNN. N .Ns-N .m.. N + NNN. N .NN. ..+ N.N.. NNNN News NNN. . - NNN. N .NN. N .NE-N -N.N>NN ENN.-NNN. NN.NN=NN mmm. . - NNN. N NNN.- N - ..N. . - .<.z ...-. N.N. ..+ NNN. N mm... . - NNN. N NNN.- N.+ NN..- N - .Ne-N N...- . + NNNN NNN. ....NNNNN NNN. N .NN.- N.+ .Ns-N N..NNN ENN.-NNN. NNE.NN .mmz .NN.N...NNN NN. .NN.N...NNN NN. .NN.N...NNN NNN NN..NN N NN..NN N NN..NN . NN..NN N=.N.N: Aupm.» Neon Nuwgapme accumcou Lama-o. .m .3 m. pan:.v .mcgaum. NowgmN-mcpupozv Noam. ummgmpcw Egon-acog-umucm.ow$$mou cowuapmgsou mmogu-.mm u4m.|_. upm_> acom cowamgmummcou sLmh-NNNN mm.zm mama cue. News ucgm>ow snob-mac. guano upww> ccom Loaomm uwpazm Egmh-m=04 scams. .N.. N NN.. N.- NNN. N.- .Ne-N NNN.- N - NN.. N NNN. N - .NN.- . + .Ns-N NNN. N NNN.- N - .NN. . - .<.z ...-. NN.. N.+ .NN. N NNN.- N - .<.z .Ns-N Nap. . + N.N. N .N.. . - .Ns-N NNN.+ N .<.z ...-. NNN: N + ...... .N.....N NNN.- N.+ N.N. . - .Ns-N .NN.N...NNN NN. .NN.N...NNN NN. NNN.N...NNN NN. NN..NN N NN..NN N NN..NN . NN..NN .N=.N.N= NNNN..NNN--.NN NNNN. 165 In the l973-75 period, the West German and Canadian rates exhibit some consistency of lead-lag relationships with the United States bond rate. Note, however, that the magnitudes of the coeffi- cients are not large and, hence, may signify weak relationships. The consistency is seen to be at zero lag and carrying the expected sign for the coefficient. The third panel of Table 38 shows the significant coeffi- cients for the l975-78 period. Two time series show a consistency in their relationships with the United States bond rate: the Cana- dian and Dutch long-term rates. The consistent relationships for these two series are contemporaneous ones. Compared to the first and second subperiod, the coefficients in the l975-78 period are larger in magnitude. Moreover, there is an increase in the number of significant coefficients, indicating a closer relationship during this time period as governments begin to reduce the restrictions that prevented free interactions between markets. The above evaluation of the cross correlations between the time series indicates an increase in the comovement of interest rates as the time period progresses toward 1978, regardless of a change in the length of the holding period. However, there is no conclusive evidence of any type of consistency describing the lead- lag relationships. Any consistency found is peculiar to each indi- vidual time series only. Neither can there be any general conclusion with respect to the time series being consistently cross correlated with the United States bond rate in the entire time period included 166 in the study. The only exception is the United States-Canadian relationship. The lead-lag structure describing this relationship is characterized by significant zero lag coefficients in all sub- periods and lengths of holding period. Multivariate Analysis Table 39 consolidates the transfer functions fbrmulated according to the time periods across different holding periods. The transfer functions applicable to the West German long term rates show some similarity in the third subperiod. The only factors that caused some differences are the yt_1 term in the three-month holding period and the xt_2 term in the six-month holding period. The stan- dard deviations of residuals of these three transfer functions are virtually the same. 0f the six transfer functions modeled, only that which assumes a three-month holding period during the third sub- period is acceptable because it was able to utilize effectively the lead-lag structure characterizing the United States-Nest German rela- tionship. This transfer function is a better model than its univari- ate model in describing the process generating the West German public authority loan rate. It is also a better model fbr fbrecasting and decision-making purposes. The other transfer functions showing similarity in their structures are those of the Canadian long-term government bond yield. With the exception of the transfer functions for the second and third period assuming a three-month holding period, the models are alike in structure and also in their standard deviations of residuals. However, 167 TABLE 39.--Multivariate time series models--Long-term interest rates (Holding-period returns) Standard Deviation of Residuals West German Long-Term Public Authority Loan Rate‘ Period 1: 3-mo. 6-mo.* y - .299 y = .999 x _ + a .561% l-yr. t t-l t 13 t Period 2: 3-mo. 6-mo. yt - .234 yt_1 = .341 Xt + at .676 1 yr.* yt = 2.550 Xt + at .436 Period 3: 3-mo. yt + .322 yt_1 = -2.270 Xt + 5.700 Xt_1 + at -.5499 at_1 .666 6-mo. yt = -l.330 Xt + 6.020 Xt_] -2.710 xt_2 + at - .626 at“1 l-yr.* yt = -l.590 Xt + 3.990 Xt_1 + at - .660 at”1 .696 Canadian Long-Term Government Bond Yield Period l: 3-mo.* yt = 1.270 Xt + at + .381 at-l .247 6-mo.* yt = .793 Xt + at + .253 at_1 .213 ll . . _ {ll-.3718 1 ' Period 2. 3-mo. yt - .655 yt_] - .825 Xti- _. at .295 6-mo.* yt = 1.690 Xt + at .310 l-yr.* yt = 1.330 Xt + at .224 Period 3: 3-mo.* yt = .610 Xt + .592 Xt_]+at + .237 at_1 .223 6-mo. yt = 1.230 xt + at .256 l-yr. yt = 1.100 Xt + at .221 United Kingdom Government 3;% War Loan Yield Period 1: 3-mo. 6-mo. l-yr. Period 2: 3-mo. 4 , 1 + .2418 6-mo.* yt - 1.240 xt + f.. B at .583 l-yr. 168 TABLE 39.--Continued Standard Deviation of Residuals Period 3: 3-mo. 6-mo. 1-yr. French Long-Term Public Sector Bond Yield Period 1: 3-mo. 6-mo. layr. Period 2: 3-mo. 6-mo. 1 yr. Period 3: 3-mo. 6-mo. l-yr. * ‘< H I - .557 Xt_4 + at 1.720 Xt + a at- '~< 1+ II t Dutch Long-Term Government Loan Rate Period 1: 3-mo. 6-mo. l-yr. Period 2: 3-mo. 6-mo. - * - 1 yr. yt .824 Xt + a t .960 .595 .960 Period 3: 3-mo. y6 = .86] Xt 4' .958 Xt_-| "Rm at 6-mo. yt l-yr. yt 1.680 Xt - at Swiss Long-Term Confederation Bond Yield Period 1: 3-mo.* yt 6-mo. 1-yr. Period 2: 3-mo. 6-mo. l-yr. Period 3: 3-mo.* yt 6-mo. y l-yr. t *Rejected .612 Xt + at 1.990 xt + .207 at_2 1 - .2558 ”80 xt " it‘rTTms); at l .488 .370 .520 .280 .307 169 the higher degree of volatility of the second time period is not clearly reflected in the statistic. Three transfer functions are acceptable for forecasting. These are the models with a three-month holding period in the 1973-75 period and the models with a six-month and one-year holding period in the 1975-78 period. The Dutch and the Swiss transfer functions do not show any similarity in their structures. Three transfer functions for the Dutch government loan rate are considered better than their univari- ate counterparts for forecasting purposes. 0f the three transfer functions formulated for the Swiss confederation bond yie1ds, only one is found to be acceptable. None of the transfer functions modeled for the United Kingdom government loan yield and the French public sector bond yield is found to reflect effectively the relationships of the time series to the United States bond rate. Hence, the respective univariate models are better models for forecasting and decision-making pur- poses. CHAPTER IX CONCLUSION The relationships between United States and foreign interest rates were examined in this study. There are two major sets of inter- est rates, the short-term and the long-term. Short-term interest rates consist of the three-month United States Treasury bill rate, the three-month Eurodollar rate, the three-month Canadian finance paper rate, and the three-month interbank rates of Nest Germany, the United Kingdom, the Netherlands, France, and Switzerland. The long- term interest rates include the United States constant maturity bond yield and the government bond yields of West Germany, Canada, the United Kingdom, the Netherlands, France, and Switzerland. The hypothesis for both the short-term and long-term interest rates states that there flsinstantaneous adjustment of foreign adjusted interest rates in reaction to changes in the United States interest rates. The hypothesis implies comovement of interest rates across national boundaries. Due to the different monetary systems existent over the 1971-78 period, the hypotheses are examined over three separate time periods: 1971-72, 1973-75, and 1975-78. 170 171 Summary of Findings--Short-Term Interest Rates The underlying models characterizing each time series were mostly of a random walk nature. The time series were found to exhibit most volatility in the second subperiod, 1973-75. The third period, 1975-78, showed the least volatility. The behavior of the time series as described by the standard deviation of residuals reflect the state of the world economy in each time period. Several events contributed to the volatility in the 1971-72 period, including the refusal of the United States to allow further convertibility of dollars into gold, the devaluation of the dollar, and the consequent exchange realignment of practically all major currencies. Many major currencies stopped pegging their currency to the dollar and permitted them to float. However, this was a "dirty" float because monetary authorities of each country were intervening in the foreign exchange markets by buying and selling their own currencies to control the exchange rate. In addition, the United States imposed an import sur- charge that obscured the true exchange rates by artificially keeping imports out of the United States. The f1uctuations in the exchange rates were viewed by most countries as undesirable and transitory rather than as a basic change in the system. The year 1973 marked the second devaluation of the dollar and the officia1 transition to a floating exchange rate system. The huge oil price increase led to major shifts in the balance-of-payments positions for many countries. The transition to a different monetary system together with shifts in balance-of—payments positions were 172 major factors giving rise to the volatility characterizing the 1973- 75 period. In the 1975-78 period, countries began to adjust to major changes in the world economy. Governments gradually took steps designed to reduce the adverse impacts of world events. To determine the lead-lag relationships between the United States Treasury bill rate and the interest rate of other countries, the time series were cross correlated. The analysis indicates that the markets show closer movement in the third subperiod, 1975-78, and an increasing number of significant cross correlation coeffi- cients increasing in magnitude as the time period moves toward 1978. None of the foreign interest rate series was found to be consistently correlated with theUnited States rate in all three periods except the Canadian finance paper rate. Although the lag zero coefficients for all series, except France, were significant, the existence of some significant coefficients at both the plus and minus lags suggests a feedback-feedforward relationship between the time series. The lead-lag relationships of the time series fbrm the bases upon which transfer fUnctions were modeled. 0f the models set forth in Table 7 on page 70, only that of the Canadian finance paper rate for the 1973-75 period was considered acceptable, based on an evalua- tion of the estimation and forecasting ability of the model. The rejection of the other transfer functions implies that the univariate models describing the time series perform better than the transfer functions. 173 Summary of Findings--Long-Term Interest Rates Three different holding periods were assumed in the examina- tion of the long-term interest rates. Adjustments of the interest rates were done employing three-month, six-month, and one-year for- ward exchange rates. Hence, when adjusted, the long-term interest rates are effectively three-month, six-month, and one-year holding- period returns. About three-quarters of the underlying models characterizing each time series were of a random walk nature. The time series were found to exhibit a decrease in the degree of volatility as the hold- ing period was lengthened. That is, one-year holding-period returns show lower standard deviations of residuals than the six-month holding-period returns, and six-month holding-period returns have lower standard deviations of residuals than three-month holding- period returns. . The U. S. ten-year constant maturity bond yield was cross correlated with the holding-period returns of other countries to determine the lead-lag relationship between the time series. Similar to the findings in the short-term interest rate cross correlation analysis, an increase in the comovement of interest rates is evident as the time period progresses toward 1978. This increase can be attributed to the reduction by governments of restrictions that pre- vented free interactions between markets. Only one time series exhibited a consistent lead-lag relationship with the United States bond rate. This is the Canadian long-term government bond yield. 174 The lead-lag relationships between the time series form the bases upon which transfer functions were modeled. Based upon an evaluation of the estimation and forecasting ability of the model, eight transfer functions were found to be acceptable: the West German three-month holding-period return (1975-78); the Canadian three-month holding-period return (1973-75); the Canadian six-month and one-year holding-period returns (1975-78); the Dutch three-month, six-month, and one-year holding-period returns (1975-78); and the Swiss six-month holding-period return (1975-78). The rejection of the other transfer functions implies that the univariate models describing the time series are superior to the transfer functions. Results of the Study The findings of the study as summarized in the previous sections of this chapter provide evidence for the nonrejection of the hypothesis that financial markets are integrated. The magnitude of the cross correlation coefficients depicting the relationship between the United States interest rates and foreign interest rates supports the nonrejection of the hypothesis. However, financial market integration characterizes the May 1975-1978 period, but not the 1971-April 1975 period, with two exceptions. The first is the lead-lag structure describing the United States-Canadian relation- ship. The Canadian interest rate, both the short-term and the long- term, is found to be consistently correlated to the United States interest rate at lag zero, implying a contemporaneous relationship. The second exception is the lead-lag structure describing the United 175 States-French relationship. In both the short-term and the long- term cross correlation analyses, the French interest rate is found not to be significantly correlated with the United States interest rate. The length of the holding period in the analysis of the long- term interest rates did not make any significant difference in the lead-lag relationship characterizing the time series. Although significant coefficients appear at different lags as the length of the holding period is varied, the general observation that can be drawn from it is that if there is a contemporaneous relationship, it is found in all three different holding periods. The study adds new evidence to previous research on financial market integration without necessarily refuting or supporting pre- vious studies. With respect to the interest rate parity theory as proof of financial market integration, the study does not provide additional support. The underlying ideas of the theory suggest that forward rates for longer periods tend to exhibit more fluctuations because of the higher risk inherent in a longer-term contract. In this research, forward rates of three different time period lengths were employed in adjusting the nondollar time series. It was expected that since there is more risk in the one-year holding period, the adjusted time series would exhibit more instability as measured by the coefficient of variation. Surprisingly, however, as the length of the holding period increased, the coefficient of variation 176 decreased. This is contrary to general expectations based on the interest rate parity theory. The current research neither refutes nor supports the con- vergence studies. The means and standard deviations of the indi- vidual time series shown in Table 37 reveals no evidence that the interest rates are converging. However, these time series are adjusted time series. Hence, the idea of convergence as applied in the previous convergence studies may not apply here. Hith respect to the covariation study by Logue et al., this research provides a continuation in the sense that the time period examined follows from that of the previous study. Implications of the Result The implications of the results of the study were twofold. 0n the microeconomic level, the lead-lag relationships imply little benefit to be derived from the transfer of funds across national boundaries, based on the assumption that transactions are undertaken at a minimum period of one week. Also, the transfer of funds is assumed to take place as a reaction to a change in the level of interest rates in the United States. This assumption does not pre- clude an individual or firm-investor from transferring funds for investment overseas because of higher effective (adjusted) interest rates. In fact, holding investors' risk preference constant, the effective long- and short-term interest rates in the Eurodollar market and in the Nest German, Canadian, Dutch and Swiss markets are higher than the United States interest rates. Of course, the 177 decision by an investor to transfer funds will depend on his or her risk-return preference. The results of the study should be taken into consideration in timing a firm's financing. The existence of significant coeffi- cients at the zero lag indicates that moving across boundaries to take advantage of lower financial costs will not provide any bene- fits to the firm. However, if firms can anticipate changes in the interest rate level and produce timely forecasts of the resultant change in the foreign interest rates by employing an appropriate transfer function, the information will be important input in reduc- ing the cost of financing to the firm. Holding risk constant, the firm should consider the French and Canadian financial market as addi- tional funds sources. The effective interest rates in these markets, after taking into account the foreign exchange adjustments, are lower than the United States interest rate. . The French interest rate does not seem to relate to changes in the United States interest rates. As mentioned in a previous chapter, the French financial market is very much under government control. In this perspective, the market may be considered inde- pendent of outside forces. For investment and financing purposes, the investor or the borrower should take the French market inde- pendently, with strong emphasis on how the government exerts its control over the influence of outside forces on its domestic market. On the macroeconomic level, the results of the study imply that governments have had some success in reducing foreign influences 178 on their domestic financial rates. 0ne interpretation of the less than strong relationships found between the interest rate series is that national governments, to some degree, have been able to control their domestic financial condition. 0n the other hand, the existence of lead-lag relationships suggests that governments have not been able to completely isolate their economies from events in the United States financial market. The magnitude of the cross correlation coef- ficients suggests the possibility of government intervention. How- ever, it is difficult to determine what the probability of success of the intervention would be. Because the coefficients do not imply strong reactions of foreign interest rates to changes in the United States interest rates, governments have a certain degree of inde- pendence in effecting certain economic policies. Hence, the use of interest rates as a tool of monetary policy has not necessarily been weakened. The existence of relationships of a feedback-feedforward nature indicates the need to closely monitor the effects on both the domestic and foreign markets by a change in the domestic interest rate level. Suggestions for Future Research This research suggests several possible paths for future studies. The current study employs weekly data in the analysis of relationships. The use of daily data may show a relationship that may be more effectively used for forecasting and decision-making purposes. For instance, foreign interest rates may lag behind the 179 United States rate changes, implying that the timing of the rela- tionship may be employed to advantage. The financial markets studied included only Canada and five European countries. The inclusion of developed markets in Asia, such as Tokyo and Hongkong, would have been desirable. The develop- ment of a good data base for these markets will be a major factor for their future inclusion in studies such as the current research. The study suggests that there may be some benefit to the transfer of funds across national boundaries. The benefits in terms of dollars or some monetary unit may be studied. The Eurodollar market is generally considered an extension of the United States financial market. This is questioned by the results of the study. The United States and the Eurodollar markets are found to be cross correlated at magnitudes not greater than 0.5. If the two markets are extensions of one another, then why are they not more highly correlated? This question suggests another possi- bility for future research. APPENDICES 180 APPENDIX A THE RELATIONSHIP BETWEEN THE FORWARD EXCHANGE RATE AND THE FUTURE SPOT EXCHANGE RATE 181 APPENDIX A THE RELATIONSHIP BETWEEN THE FORWARD EXCHANGE RATE AND THE FUTURE SPOT EXCHANGE RATE The explanations on the forward rate as an unbiased predictor of the future spot rate is the subject of this appendix. Various viewpoints and the corresponding studies will be discussed. In order to determine which viewpoint is applicable in this study, a regres- sion analysis was perfbrmed on the foreign exchange time series avail- able from the Harris Bank data file. The results of the analysis are presented and discussed. There are three views on the extent to which the forward exchange rate is considered to represent the market's forecast of the future spot exchange rate. The first viewpoint argues that the for- ward rate is an unbiased estimator of the future spot rate because Speculative transactions always bid the forward rate up or down to the point at which it equals the expected spot rate. Frenkel (35), Kohlhagen (53), Aliber (21), and Giddy (40) provide empirical support for this argument. The second approach involves the consideration of transaction costs. This approach argues that any change in the expected spot rate will be accompanied by a change in the forward rate in an amount equal to the change in the forward rate less the transaction costs of 182 183 forward speculation. The hypothesis implies that no change in the fbrward rate will occur if the change in the expected spot rate is not sufficiently large to offset the transaction cost of acting on one's expectations. Consequently, the forward rate will be con- sistently higher than the expected spot rate whenever the latter is falling, and will remain below when it is rising, as long as the change in the expected future rate is larger than the transaction cost.1 Both Kaserman (93) and Levich (95) presented evidence to support this hypothesis. The third approach on exchange rate expectation argues that perceived currency risk may deter investors from profiting fully from a difference in the level of the forward rate and the expected future spot rate. The riskiness of speculative transactions may mean that the demand fer forward currency by investors is not infinitely elas- tic at the forward rate that equals the expected future spot rate.2 Solnick (66) and Roll and Solnick (98) have tried to specify and test the relationship. However, there have been questions on the formulation due to the specification of the currency risk. Levich (86), in a recent study (1979), reports new empirical evidence on the relationship between the forward exchange rate and the future spot exchange rate during the period from January 1967 to May 1978. Levich employed end-of-week bid quotations from the 1Gunter Dufey and Ian H. Giddy, The International Money Mar- kets (Englewood Cliffs, N. J.: PrenticeéHaTl, Inc., 1978), p. 98. 21bid., p. 100. 184 interbank market reported by the Harris Bank weekly Review. He performed the following tests: mean squared forecasting error, fre- quency distribution of forecasting errors, mean ferecasting error, serial correlation of forecast errors, Chi-square test for forecast- ing bias, mean absolute forecast error, and regression analysis. His tests showed that the market is volatile and that large profit Opportunities are possible. However, the evidence does not provide conclusive proof that the market is inefficient. Levich, in his regression analysis, tested the following models. st+n = a + b Ft, n + "t (1) and S b F ‘Em‘ah'si‘hets (2) t t where St+n = spot exchange rate in period t+n, Ft n = forward exchange rate in period t for delivery in the nth period. In both equation models, the joint hypothesis that a = 0 and b = 1 cannot be rejected at the 5 percent significance level. There is some question, however, on how Levich formed his sample for testing. He set up independent samples by taking observations 1, 14, 27, . . . or observations 2, 15, 28, . . ., etc. His preocedure resulted in sample sizes that do not exceed 31 observations. His sampling pro- cedure suggests bias in his test results. 185 In order to verify Levich's conclusions, the current study undertook a regression analysis of the same data set. However, instead of following his sampling procedure, the current test employed available observations from January 1971 to December 1978. The countries examined included the United Kingdom, West Germany, Canada, France, the Netherlands, and Switzerland. The regression statistics for equation (1) are shown in the accompanying table. The tests showed different results than those of Levich. In no instance can it be said that the joint hypothesis of a = 0 and b = l is accept- able. The evidence implies that the forward rate is not an unbiased estimator of the future spot exchange rate. To justify the employment of the forward exchange rate as the proxy for the expected future spot exchange rate, the modern theory of forward exchange is invoked. According to this theory, the forward exchange market is characterized by three types of activi- ties: ,(1) pure interst arbitrage, (2) pure speculation, and (3) commercial hedging. Arbitrageurs enter the forward exchange market to eliminate foreign exchange risk, and purchase or sell fbrward exchange con- tracts in order to cover their foreign exchange positions. However, if there is a difference between the existing forward exchange rate and the spot rate that is expected to prevail in the future, arbi- trageurs will generate excess demand for forward exchange to profit from the difference. Speculators expect ' to profit from fluctuations in the for- eign exchange rate by holding open positions in foreign exchange. 186 .mgaumoocn pusuco-mcmg;uou on» mcwm: umELomcoa mm: cowmmwcmmg mcha Anemo.v Aooao.v mmp mNm.F mmoo. m.eeym_ comm. mmmm. 84am. wa\~_-ma\m Ammoo.v Amomo.v new, mPN mmm._ aeoo. “.mwomp mama. omFo. mean. ma\_-aa\_ -emNBsz Ammop.v Ammmo.v map mom._ mace. a.ammpp «mam. momp. Fame. ma\~fi-ma\P Apemo.v A_amo.v mucap em, awm.p Pmoo. ,.aa~m meam. opoo. mamm. e~\~P-NN\F -amgpaz “upmo.v “Nmm~.v mam New., Paco. a.a_emm Nmmm. mpwo. aapm. m~\~P-F~\F guacau Amamo.v Amoco.v mm_ mo..~ mmoo. ~.mmoPP aemm. amps. mmmm. ma\--ma\~ Ammoo.v “omeo.v seasamu 8_N moa.~ mmoo. m.Pmmm_ memo. a.eo.- meme. ma\P-_a\P “mm: Ammmo.v Aaowo.v Nap moP.N mmoo. m.mmam memo. m¢~_. mmmp. ma\~_-ma\o Ammmo.v Ammmo.v Pep o_m._ mmoo. m.~mmm mmmm. ammo. omNN. ma\m-ma\a moaned A FOmO.v ANmON.v EO—umc PX emm mma._ memo. o.mm~em ammo. «moo. mawm.p wa\~F-_a-m taupe: mcomum>gmmao . . . . maopm accumcou Lo .02 3 a m m a me any . Amy uowama meme acucaou a =+a : + c.uun + m u m "cowpmzam go; «mowum_umum cowmmmcmmmuu.pu< m4mgmmao —pe« mhmnu 005"“ mucua nBOuu flumod mn00 000“” Ommm nh¢m ##mh Omwh 243 0000 mode“ mamou GumOn mmNOn oooou mmm0 0000" Named Om00~ mmouw mm0uu mmowq 000“" Dhumq mh0Nu #050“ 0000“ manna mmumu 000m" Gamma 00mm" muo¢n comma nn¢mu mm0mn 000"" nm0uu 00mm" mmmun mmhuu mmhfiu mwmun om¢uu O¢0ou mn¢on OOnOn mnm0 Nnm0 N000 mmhm N00m mhcm #000 0mnm mmhm 0000 O0¢0 whm0 #560 oonOn numOu o0¢nu #«hOu mh¢0u 0N¢O« oom0 0hn0 0060 manna 0N¢uu Non—n onmuu mmo~w 0500 N050 NON0 ¢u0m 0550 ¢0Nm flo¢m Odom omnm htnh omOh O¢00 0hn0 Nnu0 #000 mmmn nhmn .22 .m .6568 - E: .S .522. 53.8: "means cuppouoczu zucosummgnhuu.mum mam<~ 244 0#00« 0000« «#«0« 0000 5#05 00#0« «000« 000«« #000« 00«0« 0#00« 000#« 0«00« 0«50« 00«0« 000«« 000«« #000 0#50 00«0 000«« #000« 000«« 000«« 0#00« 5005« 0000« 0005« 5000« 5050« 0005« 0000« «0#0« 0000« #005« 0000« .mmuupa Fmswumu 000«« 0000« 0000« 5000« 0000« 5050« «050« «500« 50#0« 0«05« #000« 0000« #000« 0000« 00#0« 5#00« 0050« 0000 0««0« 00«0« 0000 0##0 0#00 5500 «500« 00#0 0500 5#00 0000 0000 00«0 00«5 «000 0000 5005 0500 0000 00«5 0#05 00#5 0050 0000 0000 0005 «#00 50#0 0005 muggy cw .mcomum>comao «000« 0000« 50«5 «555 0505 5000 0000 00«5 #055 0005 0005 5#05 ##05 00«0 50#0 00«0 05«0« 0500« 00««« 5«#0« #000« 5000« 0000« 5000« 0000« #000« 000#« 050#« 000#« 0###« 050#« #000« 5000« 00«0« 0000« #0#«« #00«« 0#00« #50«« 000«« 0000« 5000 00#0 0500 0500 0000« ##00 0#0«« 0000« 0500 050#« 0050« 0000« 0000« 5«00« 00«0« 0#00« «000« 5000« 0005« 0#00« ##05« 5050« «000« 0000« #0«0« 0#00« 5000« #000« #000« 5000« «000« 0000« #000 0000 #0«0 0««0 0000 ««00 0000 5«00 0000 00«0 0000 5000 5«00 0«00 0000 00«0 «000 0000 00#5 «000 «#00 #000 0000 0000 0«00 #000 #000 00#0 0000« 5#0«« 5«0«« 000«« 0000« 5000 0#00 0000 0005 5005 #505 0000 00«5 #505 «050 0000 #005 0000 5005 «000 5000 5000 #000 0000 0000 «000 0000 «050 0««0 0«#0 00#0 0000 «per u «55« .mm xgmacma .xpxmmz «mump .m Lanswumo "Avmamanumv mmaoc xcmacmacv cucosummcgp casgom umozun.num ugmcmmno pp¢« 05#«« 0##«« 000«« 000«« 00««« 5000« «000« 0000 0000 5000 0000 0000 0000 00#0 0000 00#0 0#«0 00«0 0#00 #000 00#0 0000 00#0 0000 0000 0005 0005 «#05 #055 5#05 0005 00#5 0005 «000 «005 0500 5005 0#00 5«05 0005 00«5 0005 0#05 50«5 0505 0005 0005 00«5 00«5 0005 «005 55«5 0000 #050 0000 0000 0500 0000 5000 0000 0000 #550 0000 0000 «000 «0«0 0000 0000 0500 0«00 0000 0#00 0050 #050 0500 «000 0500 0500 0050 5000 00«0 0500 0##0 00«0 000# 0«0# 005# 505# #05# 000# ««0# 005# 505# 000# 000# 0«00 0000 50«0 5000 «00# 000# 0#00 0#0# 00## 000# 000# #05# 50«0 0#0# «00# 5000 000# 0««0 «000 «0«0 050# 05«0 0«00 0000 ##00 #0#0 00«0 5000 #000 0000 5#00 05#0 0000 5000 0000 0500 «000 0000 #000 00#0 00«0 000# 000# 0«0# 000# 000# 000# 005# «000 00«0 000# 00«0 «000 0#«0 0000 #000 000# 00#0 00«0 #000 5#00 0000 0000 0000 0000 0«00 0000 00#0 #000 0050 00#5 0005 0«05 0«#5 00#5 05«5 0005 50«5 #505 0000 0005 00«5 0000 5#«5 0000 «000 5000 0000 «000 0000 0500 0005 0000 0000 0000 0«05 0005 5««5 0«00 0000 0#«5 00«5 0505 05#5 5005 0#05 5005 00«0 00#0 5000 00«0« 0«00« «000« 00#0« 0000« 0000« 0000 «#00« 0000« 5000« 0050« #00«« 000«« 000«« 000«« 0000« 0000« 0«50« «050« 5500« 0000« 5000« 00«0« 0000« 0#00« «000« 0«00« 0000« 0#0«« «00«« 0o0«u 050«« 05#«« 0000« 0000« 0000« 00#0« 00«0« «000 0000 0«00 #000 0«00 0#00 «0«0 50«0 0500 0#00 0500 00«0 0000 0500 «000 0«#0 0#00 #000 0000 0000 0000 0««0 0000 0#00 0000 5000 ««00« 0000« 5500« 5«00« «50«« 000«« «0««« 00««« 050«« 0000« 0500« 0000« 00#0 ##00 0050 0#00 0000 5#«0 0005 0505 0005 0«05 0055 0055 5#05 0005 #005 0«05 0005 0«05 5005 0005 0««5 0««5 5000 0«50 0##0 0000 «000 0000 0550 0000 0000 0«#0 0000 0500 0000 0500 0#00 #«00 0000 0000 0«#0 0000 «500 0«00 0000 #000 ##0# 05«0 000# 0#0# 0500 0000 0000 0«0# 005# 0000 000# 0«0# 0#0# 500# «#0# 50## 000# 050# 005# 050# 000# 505# «00# 0#0# 00## 0000 #00# 0000 0#00 050# 000# 00«# #00# «50# wmmw «000 5304 500# 0#00 0000 000# 0054 0500 05un 0500 «#00 #050 e000 0#00 0000 0000 0000 5#00 0000 0000 #000 0000 00«0 00«0 0500 «0#0 0000 #0«0 «#00 0000 0«00 0«00 5000 0«5# #00# 050# #5## 00«# 0000 0000 0000 0000 0000 0000 0005 005# 500# 500# 505m. .0 cmasmuoa . —5m« .am xgaacaa .apxmmz .Aumpmanuuv mane; Luann ounce,» zucosumwsga cuwvacuuuu. ¢uu mam<fi 246 #000 0005 0000 0005 N000 0#0# 0000 0000 NO¢0 «000 0000 050# 500# «#00 #000 0000 0000 00«0 0000 0000 0000 0000 0#00 00#0 0000 0000 0000 0000 «000 0«00 0000 05«0 0505 0000 0000 #N«# 0000 #0«0l50#«l00#«l«000 0#00 5000 0000 0000 «050 0000 000# #00# 0500 ««00 #0#0 0000 0000 0050 5000 005# 000# «000 000# «000 0000 0«00 5005 0000 0005 «0«0 00«0 0550 0000 0500« 0000« 0000 0000 5050« 0N0«« V00—« 0N0«« N0««« 00¢O« «0#0 0500 0000 5000 0#00 00«0 005m 0005 ««#5 000* 0000 0#00 000# 0¢0¢ 0000 05«N 000¢ ¢«N0 «005 0000 .mmuupn «05,000 000:» :0 .mcowum>gmmao 000* 0005 a mum« .«m umama< .sz003 #050« .0 Lunewumo “Auoumanunv mug“; xcaagmucv gucoe-mmggp soum=_x ump_==--.m-0 04m

gmmno amptr ¢¢50 h0h0 0000 ¢h¢0a "~00 hn¢0 ¢000 Obnfi ¢nnh 00«5 «0hh ¢¢u0 0¢mh 000B 0unh hnh0 nn00 N050 0000 ¢0N0 #000 0000 ¢n00 0000 0n¢n 000m 0000 0050 ¢000 00«0 ¢000 0000 0000 0500 0000 hh¢0 0«00 0000 0¢n¢ 00¢0 0¢NN nub" 00«0 0«00 hnh0 0000 0000 ¢00N ¢000 N000 000m 0000 0000 000¢ ¢0nn 0o¢n n¢00 0000 F000 0000 000¢ nhn¢ «N00 N¢u¢ 0000 n0¢0 0000 0000 NO0N 0000 «00¢ 000¢ ON0¢ 000¢ ¢uN¢ n¢u¢ ¢nfl¢ 000¢ ¢000 0000 000¢ «00¢ 00N¢ 0000 0050 0000 0«00 0000 0000 5050 0000 0000 nn¢m 5000 «000 mnom 0h¢N 0000 n¢0m 0000 «00m 0000 nn¢N 0050 0500 0¢0N "500 50«0 NO¢0 0000 0000 500¢ 0hh0 0000 ¢n0¢ 0000 hn0m 0000 N05" nh—¢ ¢Nm0 0050 ¢000 ¢BO0 000¢ ¢000 0¢00 ¢¢hN "000 000a ¢0¢0l0000105h0 n0¢ulnhhnlnuho ¢000 00h“ hhflm n00N 00¢0 00u0 Noon 0000 Nnnn «00n 0000 0000 0hN0 00¢0 ¢u¢0 0n¢0 0000 0000 N000 ¢n00 0#00 “N00 0000 0000 0«00 ¢005 0505 0000 000¢ 0000 000a 0000 0000 0050 0000 un¢0 0¢00 “00¢ 0000 0nNn n0¢n «05¢ Nu0¢ immmp .m Lmaeoumo - mum. .mN ..ga< .apxma: "Au¢am=nuuv may»; xcungmuc, gu=o5-maggu =0:mgu--.o-m m4m<~ 248 .3qu $533 3...: E .mcofimimmao cm? 00«0u 0000a 00¢nu 00«0" 000w" u00nn 0000a 0000 0000 0¢00 ¢000 «”00“ 0000 ¢000 0000 ¢000 0000 ¢000 u0¢0 0¢00 0000 «A00 0000 0000 0000 0nn0 ¢u00 0000 0000 0000 0000 0000 0000 0000 0000 0«00 0000 n0¢0 0000 0000 0000 0000 0000 n¢¢0 0000 0«00 0¢00 00¢0 0¢00 0000 0000 0000 00«0 0000 0000 00«0 ¢000 0000 0¢00 0000 0000 0000 0000 0000 0«00 00u0 ¢¢00 0000 00¢0 00«0 0¢00 “000 ¢000 0000 ¢¢00 0000 0000 00¢0 0«00 0000 0000 0«00 0000 0000 0000 000¢ 000¢ 0000 0000 000¢ 000¢ ¢00¢ 00«0 000¢ 0¢0¢ 000¢ 0«00 WWW“ WWW“ WWW“ wmmw 000¢ ¢00¢ 0¢0¢ 000¢ 000¢ 000¢ 0¢0¢ 0¢¢¢ 0u0¢ 00¢¢ 0000 0¢¢¢ 000¢ 0000 «000 O00¢ 000¢ 00«0 0000 0000 ¢¢¢0 0¢00 0000 00«0 0¢u0 ¢000 0«00 0000 0000 0000 #000 0000 0«00 ¢000 ¢000 0000 “000 0000 «000 0000 «000 0000 0000 «0n0 00¢0 0000 0u¢0 0000 0«00 ¢000 0000 0¢00 ¢000 0000 0000 0000 0«00 n¢00 0«00 0000 0000 0¢00 0¢00 00¢0 ¢000 0000 0000 00«0 mwmw 0000 0000 0¢00 0nn0 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000 *mnmp .m gmnsmuwo u mump .mm FPLQ< .xpxmmz "Avmumznuwv mwuug xcmngwucw canoe-muss» :uu:ouu.nuu m4mgmmno msp« 0000“ #000u 0#00— 0000u 0000“ 0000a 0onuu oenuu 0000a ~00On 00«0.“ 0000 00¢0 ¢0¢0 0000 0000 0o¢0 0000 0000 0«00 0000 0000 0000 #000 #000 0000 00«0 0000 0¢O0 #000 0¢00 0000 0#00 #000 0¢¢0 0000 OO¢0 0000 0«Q0 0000 00.0 0000 0(00 0«00 ¢0H0 #000 «000 #000 0H00 mum— .m gmasaumo - m~a_ .mm um=a=< .a_xomz "Aumgmznuav mmumg xcuagmucv gazes-umggu vagm--.m-u m4mgmmno mo¢« u pump .mm xgascow .apxwmz «asap .eN gunsm>oz Hypo?» econ zu.g:uas penumcoo staucmu mmumum vauv:=--.mum m4mgmmnc ace; ttflo Assauag newsmauucvu—o; gazes.mmgghv «asap .mu swasw>oz . Fum— .mw xgmzcme .apxmo: "umamancu mums cmop auvgosuan uwpaaa Egmuumcop cmsgmw ummxuu.opum m4moz "umumnnuu rpm.» neon acmscgw>c - _~mp .mm sumscma .»_xomz .mmumpa Pmspumu mugs» cw .mcovua>gmmao moe« Egmpnmco— cavuacmunu.—pnu uamgmmno moe« 0«00 #nu0 0550 5000" Oh¢0« 0000 0000— ¢0¢0 00h0 00cm 000% 0#00 hnhh 0000 000k 00m0 000% 00N0 u pump .mu agaacma .xpxmmz Acgsumg uopgma-m=vupog gacosummgghv «asap .mu smaEmyoz 633:; 2...; $2 .5: fin 222%.,3 289.: 8:5--.24 .39: 254 05¢0u n0¢0~ 5050" 0u¢0u 0000a «000 0#00 0000 0500 0500 «000 0000 «050 0000 ¢0~0 00«0 0«00 00¢0 0055 00a5 055? 0#0# #000 5050 05¢0 0050 500¢ 0000 0050 0«00 0000 0000 0500 0000 0000 500? 0#0# 5u¢0 0000 0«00 5000 000 0000i 0¢50 0000 H000 0000 0500 «000 0000 0¢00 0000 «000 #000 0000 0005 55¢0 0005 0u05 0000 000# 0000 “500 00¢¢ 0u¢¢ 5000 0000 0000 «“50 0000 0«00 ¢0¢0 0#0 005w 000a 000a 000a 0000 «0u0 5n05 0505 500a" 000a" 00¢0u 0#00 ug¢0 00¢0 0000 00u0 0000 5a00 #005 0000 0000 0000 ##00 0000 00¢0 0000 0000 0000 ¢0¢5 #005 00#5 0000 0000 0000 5000 «000 00¢¢u 0v00~ 000uu 0«0uu «000“ numau 0000 u¢55 «000 #000 0000 0000 0000 005mm 0000" 0500a ¢0¢0u 00«0 00#5 0050 . 5a00 .mmuo—a Pmspumu muggy =0 .mcowuu>gmmno mo¢* Acgaumg vowgmauucwupo; sucosummsnuv «mum? .mm smasm>oz "umgmanum upm,» ago; soggy“ u._n=g Egma-m=o_ sucggm--.m~-m u4m

gmmno moe« 00«0n #00nu Acgaumg covgmn-mcwc.oc ;p=02-mmg;hv *msap .m~ gunsm>oz . psmw .mm agmscuc .apxowz "umumsnvu mmumg cmop ucascgm>ou Eguuumcop guaaonu.¢pnm 04mgmmao moe« 0000" 000d“ 00 0? #0 O 0 0 0 0d0fi~0~00 “dc-0 civic-tutu!!!“ 0005 0500a 5#u0u 000#" #000 0500a 0000 0000 0#0#" 500g— 00««« Asgaums vorgonumcrupo: sucosummgzhv «asap .mm sunsw>oz u Pump .mm xgmzcua .xpxmmz "coumanvm upova ucoa copumsoum$=ou Esmuumccp mmwzmuu.mrum ugmgmmno m9:r no¢¢«. «5#0 . «nap .mm xguscma .xpxmm: Acgapug quLog-m=Pcpog gacoe-x_mv «mump .mu gmnsm>oz “umumznva mmamg cuop auvsozazu uvpnan egmuumcop cusgou umwznu.mpuu uam<fi 258 .mmuapa «aswumu mugs» c. .mcopun>gmmno mo¢« « 0000« 0000« 50#0« 0500 #000 0000 00«0 #000 0«00 ««#0 5000 #5«0 00#0 00#0 00#0 0000 00«0 00«0 0500 0500 0#00 0000 0#00 #000 «0#0 0000 0000 005# 00«0 0000 0050 0000 0000 5000 0000 00«0 5«00 5000 0000 0000 «000 0«00 00#0 0#00 0000 -pump .mN xgmscma .xpxmg: Acgaumg uo.gma-ucvvpo; gacoe.x_mv «mum. .m~ gmnew>oz "umpmanvo upmvx anon ucmscgm>om Esmuumcop cmvuacauuu.-um mdmsmmao mnm¢ - P~m_ .mm pm=m=< .xpxmm: Amcgapwg nomgmg-m=_uFo; gazes-x,mv «mum...e~ guasm>oz "augmafium u_m*» snap gm: “an sgma-m=o_ soum=.x www.ca--.mp-u m4mgmmao mom« - mum— .m xgmzcmw .xpxmmz Amcgapmg corgmg-m=wu.o; gu=o5-x,mv «wfimF .¢~ Luaem>oz "umpmawum upuw» unoa gouumm u__n=a agmp-m=op gucwgu--.ap-u u4mgomno mom« «000« #000 #005 5005 #500 «0«0« 0050« 5000 00«0« 5000« 0000« 0050« 0000« 0000« 0050« #000« 00#0« #«00« 00#0« 0«00 0050 0000 5500 0050 0#00 Amcgaamg vowgmaumcpupo; npcosnxpmv «asap .mm LmnEm>oz . mump .m agmacma .xpxmmz "umamanum mum; :uop p=w2=Lm>om Egauumcop guano--.o~-u mamgmmno mo¢« 50«#« Acgaumg nowgmgmmcpu_og gpcos-xwmv *msmp .mw smasm>oz - PNaP .¢N agmacua .apxmmz Hangmanum upmwa ago; co.uagmvm0=ou ngu-m=o. mmwzm--..~-m m4mgmmno mom* «00#« 000#« 000#« 0550« 0«00« 0000« 0000« 5«00« 0#«0« 0000« 000«« 050«« ##5«« #00«« ««0«« 00««« 00««« «0««« 000«« 000«« #0««« 000«« 000«« 0«««« 0000« 0000« 0##0« 0#00« 00«0« 0000« 0000« 0«50 0000 0«00 0000 0«00 0050 0000« 0«00« «0#0« 00#0« 0000« 0000« 0000« 5000« «000« 0000 0«#0« 5000« 00«0« #0«0« 0#00« 0000« 0000« #000« 0000« 0000« #0«0« 0500« 0000« «000« 5500« 0000« #000« 500«« 000«« «#0«« 00««« 000«« «0««« 0«#«« 00««« «0#«« 500«« 00«0« 000«« 0000« 0«0«« 005«« 000«« 00#«« #00«« 000«« «000« 0#00« 0050« 0050« 0000« 0000« «#00« 00«0« #000« 0000« 00«0« 00«0« #050« 0500« 5000« «050« 000«« 5000« 0000« 0500« 00«0« 0000 0000 «#00 0050 «#00 0000 0#«0« 00#0« 5000« 000«« 000«« 000«« #00«« 000«« #50«« #«0«« 505«« 00#«« 000«« 0«0«« «55«« 0#«0« 0000« 0000« 0000« 00#0« 0000« #000« 0000« 0500« 0000« 0000« 0000« 0050« 0000« «000« #000« 5#00« 000«« 000«« 000«« 000«« 000«« «000« 000«« #00«« 00««« 00««« #000« 00«0« 0000 0000 0050 0000 0000 0000 0000 0«00 0050 «000« #000 0«#0 «000 0000 Acgauog newsmatmcwupo; gmm»-mcov «asap .m~ Lwnsm>oz - asap .m agmacaw .apxmmz "vmumafiun mung; :mo« xupgozuam uvpnan sympumco— cusgmc umwzuu.umnm mgmgmmao mom* 0000« 0##0« #000 00#0 #0#0 0000 0050 #000 0000 5000 . mump .m agmacma .xpxmw: Acgsumg workmaumcwupo; Lowxumcov «asap .mm Lansm>oz "umumanum upmva ucoa ucmscgm>om Egmpumcop cmvuacuuuu.mwuu mamgwmno mama .. 22 .m :35, .53: Acgzuwg covgmnumcwcpo; gmmaumcov «wmmp .mN Lansm>oz "umumznum upmva coop so: umm acmscgm>om Ecumcvx ump.c=-u.¢~-g mdmgmmao ~m~* 0«0«« 000«« 0000« 0000« 0050 00#0 0000 00#0 #000 0000 5«00 0#00 #050 0««0 0000 0«50 00#0 «050 0000 «000 0«00 0000 0000 «#05 0005 00«0 «000 «055 #5«0 0005 0«00 0505 0#«0 0«05 05#0 00#5 50#5 0##0 «500 0500 0#00 00## 000# 505# 00«0 00#0 0#0# 00## 0#0# 0050 00«0 050# #000 0000 #050 0000 #000 0005 0#05 0500 05«5 00«0 0050 0000 000# 0#«0 «#5# 000# «50# 00## 005# 0«0# 0000 0#00 0000 0#00 0000 0000 0000 0000 «000 0050 0000 0#00 0«00 0«#0 500# 0«0# 0005 0000 «050 0005 0005 «000 0«00 0000 0005 0500 0050 ##00 0000 00#0 500# 005# 0000 0000 0000 0050 00## 000# 0«00 «00# 00## 0050 0050 5000 0000 #000 5000 vowgmaumcwupo; La¢>smcov - E: .m r252. .58: . 0:38 «asap em L3.532 . asap .om :ugmz “mnmp .mm —_gn< uvmumanuu upwva ucoa gouumm gypsum Egmuumcop sucmgmuu.m~-m uamgomno mom« 0000« 5000« 00«0« gmnsm>oz . nmmp .m agaazmw .xpxmmz «cgzawg wovgmqumcvupo; Lama-ocov +m~m« .em “umamanva mmpag caop «coszsm>om Esmuumco— gonzo--.o~-u m4m

gmmao mom* 0050« 0#0«« 0000« 0000« 0000 000#« 000«« Acgzamg vowgmarmcmupo; Lam»-mcov «asap .mm Lunem>oz "umumanua ume» vcon cowumgmumucou Egmuumcop mmwzmuu.nwum mgm