Indian Money Market: Market Structure, Covered Parity and Term Structure

Jayanth R. Varma Indian Institute of Management, Ahmedabad Abstract In the context of the relatively recent deregulation of interest rates in India, this paper analyses the structure and inter-relationships of money market interest rates and studies the extent to which covered interest parity holds in India. The paper shows that there was a major structural break in September 1995 when in the wake of turmoil in the foreign exchange markets, covered interest arbitrage came into play in a big way for the first time. Even after September 1995, the forward premia continue to violate covered parity in systematic ways. These violations are shown to be related partly to the distortions in the foreign exchange market as measured by the premium in the unofficial foreign exchange market. Partly, however, covered parity violations also reflect distortions in the money market rates and in the formation of expectations. Though the money market is free from interest rate ceilings, structural barriers and institutional factors continue to create distortions in the market. Apart from the overnight inter-bank (call market) rate, the other interest rates in the money market are sticky and appear to be set in customer markets rather than auction markets. A well defined yield curve does not therefore exist in the Indian money market

4. 3. and traded in markets of varying liquidity. 1997). By comparison. This makes it difficult to carry out an empirical study about long term interest rates. Market structure differences between continuously clearing auction markets and sticky price customer markets. and there have therefore been few studies about the behaviour of interest rates in the country.It is only in the early nineties that interest rates were progressively deregulated in India. Since the foreign exchange market in India is not entirely free. issued by the government and by banks and corporates of varying credit rating. it is also useful to look at the unofficial (black market or havala) exchange rate of the rupee and the premium of the havala rate over the official rate. The money market is also intimately linked with the foreign exchange market through the process of covered interest arbitrage in which the forward premium acts as a bridge between domestic and foreign interest rates. Commercial Paper (CP). This paper seeks to study the structure and inter-relationships of money market interest rates in India. Treasury Bills (TBills). the secondary market for long term debt is highly illiquid and underdeveloped. The covered interest differential between international interest rates adjusted for the forward premium and domestic interest rates of comparable maturity and default risk. Even today. The money market encompasses a wide range of instruments with maturities ranging from one day to a year. the money market is relatively more liquid making it a better candidate for empirical research (for example. and Inter Corporate Deposits (ICD). 2. 1996. Data The money market instruments covered in this study include the call market. Varma. Certificates of Deposits (CD). Official publications of the Reserve Bank of India (RBI) provide data on the call market and on . Thus an analysis of money market interest rates covers four elements: 1. the paper focuses on interest differentials based on dollar LIBOR and the forward premium of the dollar at three month and six month maturities. The term structure of interest rates (the segment of the yield curve up to a maturity of one year). The credit spread between instruments of similar maturity but differing credit risk. On the foreign exchange side.

Traders are known to quote forward premia by looking at interest differentials (Deardorff.the dollar interest rate which it pays and the forward premium that it pays for covering the exchange risk. The study includes data up to the end of 1996. Covered interest parity holds very well when there are no restrictions on capital movements. but are often published in the business press on the basis of quotations provided by market participants (brokers or dealers). the implicit overnight Euro rupee rate is defined to be the US federal funds rate plus the cash/spot forward premium (annualized) of the dollar against the rupee. forward premiums are virtually identical to interest differentials in the Euro-currency markets which are completely free. In order to have a consistent source of data on these variables. this paper refers to the covered dollar LIBOR as the implicit Euro-rupee rates. forward premia and havala exchange rate (as quoted in Dubai) required for this study. this study relies on the data published every quarter in the Economic and Political Weekly. In the absence of a readily available overnight rate in the Euro dollar markets. For this reason. Most of the study however uses data only from September 1995 because of a significant structural break that was detected during the course of the study as explained later in this paper. A bank operating in India which has access to borrowing abroad in dollars can convert this dollar borrowing into rupee borrowing by taking a forward cover in the foreign exchange markets. This means that. it is the rate that would prevail in the Euro-rupee market if such a market existed. the (implicit) three month Euro-rupee rate is defined to be the three month dollar LIBOR plus the three month forward premium (annualized) of the dollar against the rupee. for example. For example. If enough banks make this shift. Data for most of the variables were available from July 1994 onwards. 1979). If this total cost (the covered interest rate) is less than the domestic rate at which the bank can borrow. the forward premium acts as a bridge between domestic and foreign interest rates through the process of covered interest arbitrage. It is worth pointing out that since the LIBOR and federal funds rates have been quite stable during this period. their action would push up the forward premia (as they all seek forward cover) and would also bring down the domestic interest rate by reducing the borrowing pressure in the domestic market. it would prefer to borrow abroad than in India. regression results using the implicit Euro rates are very similar to the results using the . the Euro rupee rates behave very much like the forward premia with a constant added. The reverse shift would take place if the covered interest rate were higher than the domestic rate. The US Federal Reserve Board publishes data on the dollar LIBOR rates of interest. except for the intercept terms.primary market auctions of T-Bills. Its total borrowing cost measured in rupees has two components . The principle of covered interest parity says that the covered interest rate equals the domestic interest rate for comparable maturity and default risk. Data on most of the others are not available in official publications. while forward premia have been highly volatile. This includes weekly data on all the interest rates. but some key interest rates were available only from October 1994. As pointed out earlier.

as argued below. The interest rates used in the study are the following: CALL The weekly average (trade weighted) call market rate published by the RBI TBPRIM The implicit cut-off yield in the weekly auctions of 91 day T-Bills TBSEC The yield quoted by the Discount and Finance House of India (DFHI) for secondary market transactions in T-Bills CDPRIM The mid rate of yields on CDs in the primary market. but a significant number are for three and six month maturities. CP and ICD rates while the second factor included the call rate and the (implicit) Euro rupee rates. Since the interest rates used in this study included a wide range of risk classes. It was therefore quite surprising to find that the two factors thrown up by the factor analysis (Table 1) had nothing to do with either maturity or risk. CD. CPs are typically for a maturity of six months CPSEC The yield quoted by the Discount and Finance House of India (DFHI) for secondary market transactions in CPs. The second factor includes rates drawn from the call market and the foreign exchange markets which are readily classified as auction markets. it was expected that factor analysis may partition the rates on the risk dimension in addition to the time dimension. ICDs are typically for a maturity of three months ER3MO The implicit Euro rupee rate for maturity of three months ER6MO The implicit Euro rupee rate for maturity of six months ERFUNDS The implicit Euro rupee rate for overnight maturity HAVALAPR The percentage premium of the havala exchange rate over the RBI reference rate Market Structure To gain an understanding of the overall structure of the money market. CDSEC The yield quoted by the Discount and Finance House of India (DFHI) for secondary market transactions in CDs. the intercept term (or themean of the interest differential) is quite important. On the other hand. the . The first factor included the T-Bills. CDs range in maturity up to one year. The first factor spans the entire spectrum of the risk dimension from sovereign to corporate. in an analysis of covered parities. 1983).forward premia themselves. Nelson & Schaefer. However. Using the implicit Euro rates is therefore preferable to working with the raw forward premia. ICDPRIM The mid rate of yields on ICDs in the primary market. 1991. But these two seemingly inexplicable factors have a very simple and intuitive interpretation in terms of market structure. It is well established in international markets that factor analysis of interest rates on government securities typically yields two major factors (representing the term structure of interest rates) which explain almost all the variation in yields (Litterman & Scheinkman. CPPRIM The mid rate of yields on CPs in the primary market. The interest rates in this factor are therefore continuously market clearing rates. a factor analysis was carried out of the various interest rates. The second factor spans the time dimension ranging from overnight to six months.

The primary T-Bills rate is set by auction and would at first sight appear to be a market clearing rate. as argued below.458 CPSEC 0. during periods of tight money.836 TBPRIM 0.318 TBSEC 0. This is direct evidence on the stickiness of the latter.418 0.118 0. The CD and CP markets show clear traits of customer markets. the first factor contains sticky interest rates characteristic of customer markets.804 ERFUNDS 0.331 CDPRIM 0.868 ER6MO 0. The CD and CP markets show clear traits of customer markets.185 CPPRIM 0. The secondary rates quoted by the DFHI are changed slowly and infrequently. Another piece of evidence in this direction is provided by the standard deviations of the different rates (Table 2).766 0.529 CDSEC 0.856 0.294 0.first factor contains sticky interest rates characteristic of customer markets. But these two seemingly inexplicable factors have a very simple and intuitive interpretation in terms of market structure. All the rates in the second factor (auction markets) show much higher variability than the rates in the first factor (customer markets).959 0.186 ICDAVG 0. The secondary rates quoted by the DFHI are Table 1: Rotated Factor Loadings from Principal Component Analysis Factor Loadings Variables First Factor Second Factor CALL 0. In fact.426 0. Moreover. Thus all the rates included in the first factor are set in customer markets and show evidence of stickiness. the description of the primary T-Bill market as a customer market is not totally inappropriate if we regard the Government as a favoured customer of the RBI. The second factor includes rates drawn from the call market and the foreign exchange markets which are readily classified as auction markets. The interest rates in this factor are therefore continuously market clearing rates. a significant portion of the T-Bills are sold through non competitive bids. Banks often subscribe to the CP of a large and valued customer in order to maintain the relationship.810 0. It is also possible that some of the auction market interest rates display the excess volatility . Banks often subscribe to the CP of a large and valued customer in order to maintain the relationship.913 overnight to six months.766 0.879 0. The same is true when a large customer wishes to place a CD with the bank.369 ER3MO 0. The same is true when a large customer wishes to place a CD with the bank. The volume of trading done at these rates is often negligible indicating that they are quite far from being market clearing rates.957 0. the RBI has often let large amounts devolve upon it to prevent a rise in the cut-off yield. On the other hand. However. These characteristics make the T-Bills rate as sticky as it would be if it were set in a customer market.

36 CPPRIM 1.11 0.17 TBSEC 2.27 CPSEC 1. this paper uses data only from this point onward.27 0.89 0.68 CDSEC 1. 1990).22 0.35 0.22 0.12 ER6MO 4.88 ICDPRIM 3.21 Covered Interest Arbitrage: The Structural Break in September 1995 As indicated at the outset.08 ERFUNDS 17.35 0.74 0.74 0.09 ER3MO 6. This is a topic for future research.21 Covered Interest Arbitrage: The Structural Break in September 1995 . 1994. Table 2: Volatility of Interest Rates in Customer Markets and Auction Markets First Factor (Customer Markets) Second Factor(Auction Markets) Variable Standard Deviation Coefficient of Variation Variable Standard Deviation Coefficient of Variation CDPRIM 2.88 ICDPRIM 3.93 0.08 ERFUNDS 17.83 0.83 0.12 ER6MO 4.23 0. there was a major structural break in the money market in September 1995 and.23 0.68 CDSEC 1.11 0.15 TBPRIM 1. Shiller. therefore.93 0. Table 2: Volatility of Interest Rates in Customer Markets and Auction Markets First Factor (Customer Markets) Second Factor(Auction Markets) Variable Standard Deviation Coefficient of Variation Variable Standard Deviation Coefficient of Variation CDPRIM 2.40 0.09 ER3MO 6.19 CALL 9.45 0.89 0.27 0.19 CALL 9.45 0. The crux of this structural break was in the inter-relationship between the money market and the foreign exchange market.17 TBSEC 2.36 CPPRIM 1. thought there were some shifts in the relationships between the domestic interest rate themselves.27 CPSEC 1.40 0.15 TBPRIM 1.that has been detected in other financial markets in India and elsewhere (Barua and Varma.

but once they have come into existence. Simultaneously. Fig 1: Covered Interest Parity before & after September 1995 0.1% level post-break. but insignificant at even the 5% level pre-break.00 45.00 35.00 15. After holding steady for nearly three years. the forward premia shot up to unprecedented levels. The interesting fact was that even after calm was restored in the markets.00 20. the rupee started depreciating rapidly against the dollar. This section studies the change that took place in September 1995 in the process of covered interest arbitrage. The regression coefficient is significant at the 0. the new-found linkage persisted.10 pre-break. This was thus a classic case of hysterisis in financial markets: markets and arbitrage mechanisms come into being in response to abnormally high profit opportunities.00 Fig 1 shows the call market interest rate and the three month implicit Euro-rupee rates plotted on the same graph. the two interest rates go their respective ways with little relationship to each other.00 10. It may be seen that before September 1995. After September 1995 however. The Indian foreign exchange and money markets went through a period of turmoil in the last quarter of 1995. the two rates track each other fairly well going through the same peaks and troughs.00 30. The well known theoretical linkage between the money market and the foreign exchange markets was witnessed in practice for the first time. The crux of this structural break was in the inter-relationship between the money market and the foreign exchange market. . this paper uses data only from this point onward. These developments had a immediate impact on the money markets too and the call market interest rate too rose sharply. thought there were some shifts in the relationships between the domestic interest rate themselves.58 after the break as against only 0.00 5.00 40. therefore. The learning process that markets and market participants have gone through leaves a permanent impact. they do not disappear when the profit opportunity returns to normal levels.00 25. there was a major structural break in the money market in September 1995 and. Statistical tests confirm the break. A regression of the Euro rupee on the call market had an R-Square of 0. A structural break is visually evident from the graph.As indicated at the outset.

The results were similar. The regression coefficients were significant at the 0. but only for short term funds (overnight to a fortnight). the mean is significantly different from zero (t = 5. A strong case can be made for testing covered parity using a longer maturity rate like the CD rate (CDPRIM) at which banks borrow from large non-bank lenders. covered parity holds between instruments of similar maturity.60% during 1980-82 and 0. the mean is again significantly different from zero (t = 7. The structural break was therefore tested using CDPRIM as well as other rates like TBPRIM and CPPRIM. One of the difficulties in studying covered interest parity in India is in identifying a domestic instruments of similar maturity. P < 0. Germany.35%.96. These tests also confirmed the structural break. a strong case can be made for using the CD rate (CDPRIM) at which banks borrow from large non-bank lenders.86.26% during 1983-88 (Abaruchis.89% during 1974-79.57%. It is however necessary to examine how effective this arbitrage has been even in this period in bringing the markets closer to covered interest parity. risk and liquidity characteristics as LIBOR. Japan. The differential in India is therefore several times what prevailed in the developed economies even .001). the comparison should ideally be with a three month or six month inter-bank market in India. (ER6MO-CDPRIM) had a mean of 3. To put these magnitudes in proper perspective. Similarly. P < 0. risk and liquidity characteristics.1% post-break.001). Since LIBOR is an inter-bank market. United Kingdom and the United States were 0. as independent variables instead of CALL.15% and a standard deviation of 3. 1993). As stated earlier. there is no simple solution to this because of the absence of an inter-bank term market in India.25% and a standard deviation of 4. On intuitive grounds. 0. Anecdotal evidence suggests however that in deciding whether to carry out the arbitrage. it is worth noting that the mean absolute deviations from covered interest parity of quarterly average domestic money market (three month) interest rates in Canada. Covered parity was also examined using the six month rate (ER6MO) instead of the three month rate (ER3MO) as the dependent variable.The use of the call market rate in the above analysis is not quite appropriate. The differential (ER3MOCDPRIM) had a mean = 3. One possible comparison is with the call market which is an inter-bank market. an inter bank term lending market did not exist in India during the period of the study due to excessive reserve requirements on term borrowing. Covered Arbitrage After September 1995 The above discussion indicates that covered interest arbitrage was practically non existent before September 1995. but has become a significant factor since then. This comparison showed a large and highly volatile differential between the implicit Euro rupee rate and the CD rate. Unfortunately. but not pre-break. Strictly speaking. banks try to forecast the likely average of the call rate over the next three/six months and compare that forecasted average with the implicit Euro rupee rate.

A similar interpretation would follow from the expectations hypothesis about interest rates also. The regression equation was therefore estimated using only the term structure slope and the havala premium as explanatory varaibles. then the interest differential would reflect a risk premium. it would be strongly correlated with the havala premium (HAVALAPR) which is driven by the same factors. The intercept term (representing that part of the mean interest differential which is not accounted for by the independent variables) is not statistically significant at even the 5% level.The Havala Premium: If the interest differential is driven by the restrictions on capital flows. This independent variable has an alternative interpretation in terms of the difference between sticky price customer markets and continuously clearing auction markets. it is likely to come down and vice versa. while the second represents the distortions in the foreign exchange market due to exchange controls. Mean reversion implies that when the call rate is very high.in the early days of floating exchange rates. . we must have an independent variable which contains a similar component. one can identify several possible determinants of this differential: The slope of the term structure: It is possible that while forecasting the future average call rate. These factors are therefore seen to be the principal impediments to the process of covered interest arbitrage in India. In trying to understand this phenomenon. Of the two explanatory variables. It may be seen that the R-square is quite high (0. the slope of the term structure is a measure of the likely mean reversion in the call rate. the first captures the distortions in the process of expectations formation and/ or the stickiness of prices in customer markets. and the difference between ICDPRIM and CDPRIM (the credit spread between banks and corporates). Since the dependent variable includes a component representing the spread between the two kinds of markets. the “normal” rate to which the call rate mean reverts is itself changing over time. the residual mean covered interest differential is not significant. banks may place too much reliance on the current call rate. the Euro rates and the call rate are market clearing rates. and that after these have been accounted for. the call rate would revert to the mean or “normal” rate. while the T-Bills and the CD rate are sticky. As shown earlier in this paper. Assuming that the term rate is a proxy for the normal rate.57) and both the term structure slope and the havala premium are highly significant factors determining the covered interest differential. The slope of the term structure was defined as the difference between the T-Bills rate and the call rate (TBPRIM-CALL). Two measures of the credit spread were tried: the difference between CDPRIM and TBPRIM (the credit spread between the government and the banks). The credit spread: If the risks involved in the Euro rupee borrowing are different from that of the CD borrowing. If this risk premium were time varying. This would be inappropriate because it would ignore the high degree of mean reversion of the call rate (see Varma 1997). These results indicate that the covered interest differential is strongly influenced by the term structure slope and the havala premium. As shown in Varma (1997). it is likely to be correlated with other credit spreads. In both cases. Regression analysis indicated that the credit spread was not a significant explanatory variable in explaining the interest differential (ER3MO-CDPRIM).

637) R-Square = 0. If the three/six month rate is excessively sticky then 1.53. In each week we can therefore compute the actual realized average of the call rate over the three months.34 which is not too low. This reflects the fact (discussed earlier) that the Euro rates also reflect pressures in the foreign exchange market unrelated to interest rates. and the hypothesis that the coefficient is unity is strongly rejected (t = 3. the regression coefficient is close to 3.295) (2.349.025): CALL26FA = -93. Conclusion This paper has presented evidence on some major distortions in the Indian money markets and . In the case of CALL13FA.001) CALL13FA = -20. we can compute the six month (26 week) forward moving average (CALL26FA). The regression coefficient would be well in excess of unity The empirical results are mixed. P < 0. the best predictor turned out to be TBPRIM with an R-square of about 0. The realized average would be more volatile than the three/six month rate 3. it was decided to investigate whether the Euro rates (which are set in auction markets) are a better predictor of the realized average call rate than the domestic rates. this is a thirteen week forward moving average (CALL13FA). Similarly. Under the expectations hypothesis.012) Since the domestic interest rates other than the call rate are sticky. The results provide statistical evidence of the often lamented absence of a well defined yield curve in the Indian money market despite a plethora of interest rates of different maturity. The results showed that the Euro rates have negligible explanatory power though their standard deviations are comparable to those of the realized averages. F(1.669 + 8.427 TBPRIM (-2. the three month rate represents the market expectation of the average short term rate over the next three months. P = 0. F(1.55) = 29.32. The expectations hypothesis of the term structure provides the possibility of resolving this question.952 (P = 0.44) = 6. But CALL13FA is much more volatile than TBPRIM.142. the fit was poorer and the hypothesis that the slope is equal to unity is somewhat weakly rejected (t = 2.512 +2. This regression would have little explanatory power 2.001) When CALL26FA was regressed against TBPRIM.434) R-Square = 0.849 TBPRIM (-3.215) (5. We then regress these realized averages on a three/six month rate.524 (P < 0.Term Structure: Expectations Hypothesis The results earlier in this paper on stickiness of many of the interest rates raises the perturbing possibility that much of what appears to be a term structure in interest rates only reflects the fact that the sticky three/six month rates are adjusting slowly to changed market conditions while the market clearing call rate has adjusted instantly. risk and liquidity characteristics.

351-364. it is equally important to carry forward the process of structural deregulation in the money market. “The Dynamics of the Term Structure and Alternative Immunization Strategies”. ed. “Market Volatility and Investor Behaviour”. but the evidence shows that interest rate deregulation by itself has not been sufficient to produce well functioning markets. Shiller. forthcoming. J. and Varma. Systematic violations of covered parity prevent an implicit yield curve from being constructed through that rate. R. Blackwell. (1996). CT. A well defined yield curve does not exist in the Indian money market because domestic interest rates other than the call rate are sticky. C.. and Schaefer. the other interest rates in the money market are sticky and appear to be set in customer markets rather than auction markets. et al. Innovations in Bond Portfolio Management: Duration Analysis and Immunization. (1983). (1979). Oxford. 1. (1993).foreign exchange markets. 2. A. S. (1994). T. & Scheinkman. . J. “Common Factors Affecting Bond Returns”. References Abaruchis. American Economic Review. Varma. (1990). Greenwich. CT. R. 58-62. Indian Journal of Applied Economics. International Financial Market Integration. Advances in Financial Planning and Forecasting... R. “The Stochastic Dynamics of the Short Term Interest Rate in India”. in Bierwag. R. “Bond Valuation and the Pricing of Interest rate Options in India”. July 1996. Journal of Political Economy. ICFAI Journal of Applied Finance. Barua. 54-61. 2(2). V. J. J. ed. R. Structural barriers and institutional factors continue to create distortions in the market: 1. in Stansell. 87. S. “Speculative Dynamics: The Case of Mastershares”. G. 161-176. A. Apart from the call rate and the forward premia. 3. (1997). These violations have been shown to be related partly to the distortions in the foreign exchange market as measured by the havala premium. Nelson. Deardorff. The money market is free from interest rate ceilings. O. (1991). covered parity violations also reflect distortions in the money market rates and in the formation of expectations. K. the forward premia continue to violate covered parity in systematic ways. Varma. 80(2). Even after September 1995. Journal of Fixed Income. “International Financial Markets Integration: An Overview”. S. “One Way Arbitrage and its Implications for the Foreign Exchange Markets”. R. JAI Press. M. Litterman. however. It is evident therefore that while interest rate deregulation is useful. JAI Press. Greenwich. J. Partly.

9478.070/3907089.088 41.302.  .083.479089.30.43/705708039890/89479438390147030.250/203989490 574.4.7098 09080..033/.5..7090701470800394-090573.709/:094 0.7-97.439748 %0801.:894207 2.30884157..

7894-0. %072$97:.9700.550.98901.3088412.9438549088 %0708:980.908437010.5074389.9.9.808905079:7-3 5488-99.9072897:.99089..9:7050.3419039070897.9:70339070897.9 9.92:.9087.41.7073985.

97900300147.904.0.908.425:9090./:890/389.07 90970024398988..73.9438549088 909700 24397.43/9438090 2./:8938494.9:.9700.824397..907057080398902..3986:08943 &3/0790050.90...39 %0050.3..904.7.30/2..0847990727.0  $2.07.425:909082439 00 147.9434190.943854908841909072897:.709.0843.70.3.709050..70.9:70 574.7/24..7.0  09037070889080 70.07.0/.07.9..390701470./08905488-9417084.9.07..000.07..0/.9.709.7/24.8.7 0 .04190.3 .07903099700 24398 30.

824397.90 1909700.

099005.9475407  %070.909.824397.088.9080.04:/-02470.0/.089.07.3.3909700.4..903  %8707088434:/..

7041.909.708:98.857080390/0./03.7098 .3/ .70       !   $3.3.990:747.8   43.-4:9 .-803.8544707.7.04384202.. 9.947419070..48094 .90/ 9  !       %!#       # $6:.7.9432.709:370..0 3903/.4011.70.90849079.700.07.90/ 9  !        %!#       # $6:..90.3/.908..090/42089.83499444 :982:...:7.389%!# 9019.030-0 05.9:79 78.0.90/9439070897.94:3988420.957088:708390 147030.324302.07.0394:/-0030.4011.0/ .0/0130/0/ .50947.507.7.08 %87010.390/42089.324302.0/ .908 %0708:98840/9.708093.990:747.90 090790:747.3/905490889.990 .90841/11070392..90.0.:.20390/.08841:39 %00257.9479:730/4:994-0 %!#9.425./0/943..7089.990 8450806:.908  %0708:98574.39070897.847010.989.47/894794383903/.089.8/0./089.707 9.947540794:90789.8041 90-089570/.3# 86:..70.70       !   03.7.7.908 .:843 %85.3/ 6:/9.87070880/.9438..824397.4139070897.98901.390.7//0.90  %070708843..9 /8.7098.-0949480419070.4011.302.041.3%!#9070708843.0398:39889743700.709/08590..908./03.:880/0.3/905490889..4.90789.2470 .-09907570/.0398.0419041903.7020/ 390..909.

.84190$479%0723907089#.4.0 8489.90 5943833/.90849079..4342.72.799490/89479438390 147030..94339024302.41550/3.3.94341 050.90839024302..7098 $97:.7/5702. % !7088  0..9343/#09:738 4:73.7/90574.820..3     42243.7098 4:73.08 -.8  .79570.:8942072..07 .3/9825.9033/.3.4.001:3.7093907.9438343/!479144 ..9433 .4.0/4083490893903/..90/070:.9438.9478110..0-00384394-070.390.4342.0107 $     %03. $8902. % !7088  $07 #     .8%0...3/3901472.8041.41550/.79 40.9438:801: 9806:.84190%072$97:.01742-03.147030.  #    $50.550.8 %080.9:7.43897:.0       '.7097.   09.7/4711  '    30.3.7089.4393:094.:7.2.709.3/90!7.070/5./03.3/$.5702:2 !.25479.9.03.7098 %024302.9.02039:7.90.07.9.:7. $  ..97.8  %    39073.9008 307.943.  #    43/'.9:7./070:.34139070897. 90490739070897.3.7:.3.302.3/90147.709'4.9438   0/0130/0/..8349-0038:11.1907$05902-07 90147.4393:094..0 3 $9.9039070897.302..793 88902.9.90.709-0.038 -:9900.1..:.039.79174290.34:73.7098   .9/8947943839024302.:.  #    %0$94..90/5.070/5.3/3899:943.943841.939070897.333.99..3.420     0843  .9:70.90   980.423 3/.7.794-08093.3/470.4342     99072.9433 2.3/9073.893 7003.!.3 #  $./039907014709.90.79 .847010.9438147904703.070/5.3.943-9801.#0.90.  1479..4.032.70983907.943$97.89.4:7 207..89078.4.708  /.2. 410/3.0   :    '.3/22:3.089470..-.90/894794383902.90/974:9.3/.03994574/:.943.0 22:3.9079.324302.7/5702.3.9.:80/42089.  4:73.88.77078..90/070:.70987.908.7.380 $ # 0/ 39073..3 25..709  #010703.3.9478.2.0833.3...9432.41!49.72.9438.7:.709  5.0  .72.7-97.30 .0.03.709 .8:70/-90.943 7003.3.:.943  147/ .7089.08841897:...77 147..4.70. 0/ 334.9433.3/3.70981700174239070897.943. 3907089 7.4..90/.3/'...3994 .