NBER WORKING PAPER SERIES

THE TERM STRUCTURE OF INTEREST RATES

Robert 3. Shiller

3. Huston McCulloch

Working Paper No. 2341

NATIONAL BUREAU OF ECONOMIC RESEARCH1050 Massachusetts Avenue

Cambridge, MA 02138August 1987

For the Handbook of Monetary Economics. The research reported here is partof the NBER's research program in Financial Markets and Monetary Economics.Any opinions expressed are those of the authors and not those of the NationalBureau of Economic Research.

NBER Working Paper #2341August 1987

The Term Structure of Interest Rates

ABSTRACT

This paper consolidates and interprets the literature on the term

structure, as it stands today. Definitions of rates of return, forward

rates and holding returns for all time intervals are treated here in a

uniform manner and their interrelations, exact or approximate, delineated.

The concept of duration is used throughout to simplify mathematical expres-

sions. Continuous compounding is used where possible, to avoid arbitrary

distinctions based on compounding assumptions. Both the theoretical and the

empirical literature are treated.

The attached tables by J. Huston McCulloch give term structure data for

U. S. government securities 1946-1987. The tables give discount bond yields,

forward rates and par bond yields as defined in the paper. The data relate

to the concepts in the paper more precisely than does any previously

published data series.

Robert J. Shiller J. Huston McCullochCowles Foundation Department of Economics

Yale University Ohio State UniversityBox 2125 Yale Station 410 Arps Hall

New Haven, CT 06520-2125 1945 N. High St.

(203) 432-3798 Columbus, OH 43210-1172(614) 292-0382

The Term Structure of Interest Rates1

Robert J. Shiller

The term of a debt instrument with a fixed maturity date is the time

until the maturity date. The term structure of interest rates at any time

is the function relating interest rate to term. Figure 1. shows the U.S.

term structure of nominal interest rates according to one definition for

each year since 1948. Usually the term structure is upward sloping;

long-term interest rates are higher than short-term interest rates and

interest rate rises with term. Sometimes the term structure is downward

sloping. Sometimes it is hump shaped, with intermediate terms having

highest interest rates.

The study of the term structure inquires what market forces are

responsible for the varying shapes of the term structure. In its purest

form, this study considers only bonds for which we can disregard default

risk (that interest or principal will not be paid by the issuer of the

bond), convertibility provisions (an option to convert the bond to another

financial instrument), call provisions (an option of the issuer to pay off

the debt before the maturity date), floating rate provisions (provisions

that change the interest payments according to some rule) or other special

1The author is indebted to John Campbell, Benjamin Friedman, Jonathan

Ingersoll, Edward Kane, Stephen LeRoy, Jeffrey Miron, and J. HustonMcCulloch for helpful comments and discussions, and to Sejin Kim,Plutarchos Sakellaris, and James Robinson for research assistance. Researchwas supported by the National Science Foundation.

I

mo nths

Figure 1. The Term Structure of Interest Rates. Data plotted are par bondyields to maturity, r (t,t+in), against time t and term in, annual data. enof June, 1948-85. Curses on surface parallel to in axis show the termstructure for various years. Curves on surface parallel to t axis show pa:through time of interest rates of various maturities. Maturities shown are0, 1, 2, 3, 4, 5, 6, and 9 months and 1, 2, 3, 4, 5, and 10 years. Note t:longer maturities are at the left, the reverse of the usual plot of termstructures, so an 'upward sloping' term structure slopes up to the left.So'irce of data: See Appendix by J. Huston McCulloch

2

1980

4years 3

term: m6

5

time tyear

1960

4

21950

0

features.2 Thus, the study of the term structure may be regarded as the

study of the market price of time, over various intervals, itself.

What follows is an effort to consolidate and interpret the literature

on the term structure, as it stands today. The notation adopted is a little

more complicated than usual, to allow diverse studies to be treated in a

uniform notation. Definitions of rates of return, forward rates and holding

returns for all time intervals are treated here in a uniform manner and

their interrelations, exact or approximate, delineated. The concept of

duration is used throughout to simplify mathematical expressions. Continu-

ous compounding is used where possible, to avoid arbitrary distinctions

based on compounding assumptions. The relations described here can be

applied approximately to conventionally defined interest rates or exactly to

the continuously compounded McCulloch data in the appendix. The McCulloch

data, published here for the first time, are the cleanest interest rate data

available, in that they are based on a broad spectrum of government bond

prices, and are corrected for coupon and special tax effects.

Section II below is a brief introduction to some key concepts in the

simplest case, that of pure discount bonds. Section III sets forth the full

definitions and concepts and their interrelations. Section IV sets forth

theories of the term structure, and Section V the empirical work on the term

structure. Section VI is an overview and interpretation of the literature.

2The U. S. government bonds used to produce Figure 1 are in somedimensions good approximations to such bonds: default risk must be consi-dered very low, the bonds are not convertible, and there are no floatingrate provisions. However, many long-term U. S. bonds are callable five yearsbefore maturity, and some bonds are given special treatment in estate taxlaw.

3

II. Simple Analytics of the Term Structure: Discount Bonds

A discount bond is a promise by the issuer of the bond of a single

fixed payment (the "principal") to the holder of the bond at a given date

(the "maturity"). There are no intervening interest payments; thus the bond

sells for less than the principal before the maturity date, i. e., it is

expected to sell at a discount. The issuer of the bond has no other obliga-

tion than to pay the principal on the maturity date. An investment in a

discount bond is not illiquid because the holder can sell it at any time to

another investor. Let us denote by Pd(t,T) the market price at time t of a

discount bond whose principal is one dollar and whose maturity date is T,

t � T. The subscript d denotes discount bond, to contrast this price from

the par bond price to be defined below. The "term" of the bond (which will

be represented here by the letter in) is the time to maturity, in — T-t. Thus,

the term of any given bond steadily shrinks through time, a three-month bond

becoming a two-month bond after one month and a one-month bond after two

months.

All discount bonds maturing at date T for which there is no risk of

default by the issuer ought to be perfectly interchangeable, and to sell at

time t for pd(t,T) times the principal. The price Pd(t,T) is thus determined

by the economy-wide supply and demand at time t for credit to be repaid at

time T. The determination of Pd(t.T) is thus macroeconomic in nature, and is

not at the discretion of any individual issuer or investor.

The price Pd(t,T) of a discount bond may be generally expected to

increase gradually with time t until the maturity date T, when it reaches

its maximum, equal to one dollar. The increase in price for any holder of

the bond over the period of time that he or she holds it is the return to

4

holding it. The actual increase in price, since it is determined by market

forces, may not be steady, and may vary from time to time. It is useful to

have some measure of the prospective increase in price that is implicit in

the price pd(t,T). The yield to maturity (or interest rate) rd(t,T) at time

t on the discount bond maturing at time T can be defined, given Pd(t,T), as

the steady rate at which the price should increase if the bond is to be

worth one dollar at time T. If the growth of price is to be steady, then the

price at time t', t � t' T, should be given by pd(e,T)e(ttd(tT).

Setting this price equal to one dollar where t' — T, and solving for

rd(t,T), we find that the yield to maturity is given by:

rd(t,T) — log(p(t,T))/(Tt)

The term structure of interest rates, for discount bonds, is the function

relating rd(t,t+m) to m. We may also refer to r(t,t+m) as the "rn-period

rate," and if m is very small as the "short rate," if m is very large as the

"long rate."

Note that the term structure at any given date is determined exclu-

sively by bond prices quoted on that day; there is a term structure in every

daily newspaper. Those making plans on any day night well consult the term

structure on that day. We can all lend (that is, invest) at the rates shown

in the paper, and while we cannot all borrow (that is, issue bonds) at these

rates, the rates shown are likely to be indicative of the rates at which we

can borrow. If the one-year interest rate is high, arid the two-year interest

rate is low (i. e., if there is a descending term structure in this range)

then individuals firms, or governments who plan to borrow for one year may

5

be rather discouraged, and inclined to defer their borrowing plans for

another year. Those who plan to lend this year rather than next would be

encouraged. The reverse would happen if the term structure were ascending.

Most individuals, of course, do not pay close attention to the term struc-

ture, but many do, and firms and governments do as well. The term structure

on any day is determined by those who enter their preferences in the market

on that day. A descending term structure on that day means that if the term

structure had been flat there would be an excess supply of one-year bonds or

an excess demand for two-year bonds. The descending term structure arises,

of course, to choke off this excess demand.

In making plans using the term structure, it is helpful to realize that

the term structure on any given date has in it implicit future interest

rates, called forward rates. In the above example, where the term structure

is descending between one and two years it is implicit in the term structure

that the one-year interest rate can be guaranteed to be lower next year than

it is this year. To guarantee the forward rate one must be able both to buy

and to issue bonds at quoted prices. One achieves this by trading in bonds

of different maturities available today. One buys a discount bond at time

t maturing at time T at price Pd(t,T) and issues an amount of discount

bonds maturing at t' at price Pd(t,t) , where t < t' < T. If the number

of bonds issued equals pd(t,T)/pd(t,t') , then one will have broken even,

at time t . That is, one will not have acquired or lost any cash today in

the transaction. However, at time t' one must pay the principal on the

bonds issued, equal to pd(t,T)/pd(t,t') . At time T one will receive the

principal on the (T-t)-period bond, equal to 1. Thus, the outcome of the

transaction is in effect that one is committing oneself at time t to buy a

6

discount bond at time t' maturing at time T with price Pd(t,T)/Pd(t,t).

The forward rate fd(t,t' ,T) at time t applying to the time interval t' to T

is the yield to maturity on this contract:

fa(t,t',T) — log(p(t,T)/p(t,t'))/(Tt') t < t' < T

This may also be called the t'-t period ahead forward rate of maturity T-

t'. One can as well guarantee that one can borrow at the forward rate

fd(t,t' ,T) by buying discount bonds maturing at t' and issuing bonds

maturing at T.

One might thus consider, in deciding whether or not to defer borrowing

or lending plans, a comparison of the spot rate today rd(t,T) with the

forward rate of corresponding maturity k periods in the future,

There is also another margin to consider. One might hold

one's borrowing or lending plans fixed, deciding, let us say, to invest at

time t+k, but to consider whether to tie down the interest rate today at

or to wait and take one's chances with regard to the future

spot rate rd(t+k,T+k).

The subject of the literature surveyed here is how people who are

making decisions at the various margins interact to determine the term

structure. Before embarking on this, it is important to broaden our

definitions and concepts.

III. Fundamental Concepts

111.1 Bonds: Their Definition

The term "bond" will be used here for any debt instrument, whether

7

technically bond, bill, note, commercial paper, etc., and whether or not

payments are defined in nominal (money) terms or in real terms (that is,

tied to a commodity price index).

A bond represents a claim on a prespecified sequence of payments. A

bond which is issued at time I and matures at time T is defined by a

w-element vector of payment dates (t1, t2, ..., t,,,1, T) where I < t � T

for all i, and by a w-element vector of corresponding positive payments

(s1, s2, s3, .. . , s) . In theoretical treatments of the term structure,

payments may be assumed to be made continually in time, so that the payment

stream is represented by a positive function of time s(t), I < t � T.

Two kinds of payment sequences are common. For the discount bond

referred to above the vector of payment dates contains a single element T

and the vector of payments contains the single element called the principal.

A coupon bond, in contrast, promises a payment at regular intervals of an

amount c called the coupon and a payment of the last coupon and principal

(the latter normalized here at 1) at the maturity date. Thus, for example, a

coupon bond that will mature in an integer number of periods and whose

coupons are paid at integer intervals has vector of payment dates (1+1,

1+2, . .., I+w-l, I+w) , and vector of payments (c, c, . .. c, c+l) . A

perpetuity or consol is a special case of a coupon bond for which T , the

maturity date, is infinity.

The purchaser at time t of a bond maturing at time T pays price

p(t,T) and is entitled to receive those payments corresponding to the t.

that are greater than t , so long as the purchaser continues to hold the

8

3bond. A coupon bond is said to be selling at par at tune t if p(t,T)is equal to the value of the principal, by our convention equal to 1.00.

A coupon bond may be regarded as a portfolio of discount bonds. If

coupons are paid once per time period, for example, then the portfolio

consists of an amount c of discount bonds maturing at time 1+1 , an

amount c discount bonds maturing at time 1+2 , etc. , and an amount c+l

of discount bonds maturing at time T . Should all such discount bonds be

traded, we would expect, by the law of one price, that (disregarding

discrepancies allowed by taxes, transactions costs and other market imper-

fections) the price of the portfolio of discount bonds should be the same as

the price of the coupon bond.4

There is thus (abstracting from market imperfections) a redundancy in

bond prices, and if both discount and coupon bonds existed for all maturi-

ties, we could arbitrarily confine our attention to discount bonds only or

coupon bonds only. In practice, we do not generally have prices on both

kinds of bonds for the same maturities. In the United States, for example,

discount bonds were until recently available only for time to maturity of

31n the United States coupon bonds are typically traded "and accruedinterest" (rather than "flat") which means that the price p(t,T) actuallypaid for a coupon bond between coupon dates is equal to its quoted priceplus accrued interest which is a fraction of the next coupon. The fractionis the time elapsed since the last coupon payment divided by the timeinterval between coupons.

4Conversely, a discount bond may be considered a portfolio of couponbonds, though in this case the portfolio involves negative quantities. Forexample, a two-period discount bond may be regarded as a portfolio of one-and two-period coupon bonds whose coupons are c1 and c2 respectively.

The portfolio would consist of -(c2)/[(c1+l)(c2+1)] of the one-period

coupon bonds and l/(c2+1) of the two-period coupon bonds.

9

one year or less. There is also redundancy among coupon bonds, in that one

can find coupon bonds of differing coupon for the same maturity date.

111.2. Interest Rates: Their Definition

The yield to maturity (or, loosely, interest rate) at time t of a bond

maturing at time T is defined implicitly as the rate r(t,T) that discounts

its vector of payments s to the price p(t,T)

_(tit)r(t,T)(1) p(t,T) — E s.e

1

1

The right-hand side of this expression is just the present value, discounted

at rate r(t,T) , of the remaining payments accruing to bond holders. For

discount bonds, this expression reduces to the expression given in Section

II. above. The yield to maturity may also be given an interpretation as

above. Given the price p(t,T), r(t,T) is that steady rate of appreciation of

price between payment dates so that if the price falls by the amount s at

each t. before T, the price equals ST at time T. In theoretical treatments

of the term structure in which the payments are assumed made continually in

time, the summation in (1) is replaced by an integral.

The expression (1) gives the continuously compounded yield to maturity

r(t,T) . One can define a yield to maturity with any compounding interval

h: r(t,T,h) — (etT) - 1)/h . In the United States, where coupons are

traditionally paid semiannually, it is customary to express yields to

10

maturity at annual rates with semiannual compounding.5 Continuous compound-

ing will be assumed here for consistency, as we do not wish to allow such

things as the interval between coupon dates to dictate the compounding

6interval.

For coupon bonds, it is customary to define the current yield as the

total coupons paid per year divided by the price. Current yield is not used

to represent the interest rate and should not be confused with the yield to

maturity.

If coupon payments are made once per period, then equation (1) is a

(T-t)-order polynomial equation in e1tT) which therefore has T-t

roots. However, given that s 0 for all i, there is only one real

positive root, and this is taken for the purpose of computing the yield to

maturity r(t,T)

Roots of polynomials of order n can be given an explicit formula in

terms of the coefficients of the polynomial only if n is less than five.

Thus, yields to maturity for T-t greater than or equal to five can be

determined from price only by iterative or other approximation procedures,

or with the use of bond tables.

The term structure of interest rates at time t is the function

5Thus, computing yield by solving (1) and converting to semiannualcompounding (using h — .5 ) gives us exactly the yields in bond valuetables, as in Financial Publishing Company, (1970), so long as the term Inis an integer multiple of h — .5 . Whether or not m is an integermultiple of h , this also gives exactly yields to maturity as presented inStigum (1981) page 111 if p(t,T) is represented as price plus accruedinterest.

6Continuously compounded yield to maturity has also been referred to as"instantaneous compound interest," "force of interest,' or "nominal rateconvertible instantaneously." See for example Skinner (1913).

11

relating yield to maturity r(t,t+m) to term m. A plot of r(t,t+m)

against m is also known as a yield curve at time t . There is a term

structure for discount bonds and a term structure for coupon bonds. If we

assume the law of one price as described in the preceding section, then,

given the coupons, there is a relation between the different term structur-

es.

111.3. Par Bonds

Consider a bond that pays coupons continuously at rate c per period

until the maturity date T when a lump-sum payment of 1 is made. If we

disregard taxes and other market imperfections, the law of one price implies

that the price of this bond in terms of pd(t,T) is given by:

(2) P(tT) — 1T + pd(t,T)

The yield r(t,T) of a par bond is found from P(t1T) by setting the

left-hand-side of this expression to 1 and solving for c:7

1 -

(3) r (t,T) — ___________P rT

I

Jt111.4 Instantaneous and Perpetuity Rates

The interest rate of term zero is r(t,t), defined as the limit of

r(t,T) as T - t or as rd(t,t) defined as the limit of rd(t,T) as T

approaches t . It is the instantaneous interest rate, which is of course

7Note that for a par bond the yield to maturity equals the coupon. Notealso that in the presence of taxes the law of one price need not imply (2)or (3). McCulloch's [1975b] formula for r (t,T) collapses to (3) if theincome tax rate is zero. p

12

not directly observed in any market. Since r(t.t) — rd(t,t) we can adopt

the simpler notation rt to refer to this instantaneous rate of interest. L:

the other extreme is, r(t1a) the limit of r(tT) as T approaches . This

is the consol or perpetuity yield, which is just the inverse of the integra.

of pd(t,5) from s—t to s—.8

111.4. Estimates of the Term Structure

At any point of time t , there will be an array of outstanding bnth

differing by term, m — T-t , and by payment streams. Of course, not all

possible times to maturity will be observed on available bonds at any givefl

time t , and for some terms there will be more than one bond available.

There has long been interest in estimates of rates of interest on stand.ard

bonds in terms of a standard list of times to maturity, interpolated from

the rates of interest on bonds of those maturities that are actively trade.

The U. S. Treasury reports constant maturity yields for its own

securities, that appear regularly in the Federal Reserve Bulletin. SaThmoL

Brothers (1983) provides yield curve data for government bonds of a wid-e

range of maturities. Durand (1942, and updated) provides yield curve d.ata

8Corresponding to the consol yield, we may also define the yield of adiscount bond of infinite term, r,(t,infinity), defined as the limit ofrd(t,T) as T goes to infinity. Dbvig Ingersoll and Ross (1986) have acurious result concerning rd(t,infinity) in the context of a state pricedensity model. They show that if r1(t,infinity) exists for all t, thenr(t,infinity) � r(sinfinity) with probability one when t < s. Othersearbitrage profits would obtain. Thus, the long term interest rate so defincan never fall. Intuitively, this seemingly strange result follows froa thefact that for large enough T the price p,(t,T) is virtually zero and hencecannot decline, but will rise dramaticalTy if there is any decline in rd(t

13

for corporate bonds. These data are interpolated judgznentally.9

McCulloch used a spline interpolation method (1971),(1975b), that deals

statistically with the redundancy of bonds and deals systematically with

some differences among bonds, such as tax provisions pertaining to them.

His method produced an estimate of an after-tax discount function and from

that the price pd(t,T) of a taxable discount bond as a continuous function

of T . Expression (1) was then used to convert this estimated function

into a function rd(t,T) . Values of his estimated continuous function for

various values of t and T appear in Table A-l. His method allows for

the fact that, in the U. S. personal income tax law, capital gains are not

taxable until the bond is sold and that, until the 1986 tax act, capital

gains on bonds originally issued at par were taxed at a rate which was lower

than the income tax rate. He describes his function in the appendix. Other

functional forms for estimation of the term structure have been discussed by

Chambers, Carlton and Waldman (1984), Jordan (1984), Nelson and Siegel

(1985), Schaefer (1981), Shea (1984), (1985), and Vasicek and Fong (1982).

McCulloch used his estimated Pd(t,T) to produce an estimate of the

term structure of par bond yields, using an equation differing from (3)

above only for tax effects.

111.5. Duration

The term m — T-t of a bond is the time to the last payment, and is

9Other important sources of historical data may be noted. Homer (1977)and Macaulay (1938) provide long historical time series. Amsler (1984) has

provided a series of high quality preferred stock yields that might proxyfor a perpetuity yield in the United States, a series which is much longerthan that supplied by Salomon Brothers (1983).

14

unrelated to the times or magnitudes of intervening payments s1, 2'

S1 . Since bonds can be regarded as portfolios of discount bonds, it may

be more useful to describe bonds by a weighted average of the terms of the

constituent bonds rather than by the term of the longest bond in the

portfolio. The duration of a bond, as defined by Macaulay (1938), is such a

weighted average of the terms of the constituent discount bonds, where the

weights correspond to the amount of the payments times a corresponding

discount factor)0 The use of the discount factor in the definition implies

that terms of very long-term constituent discount bonds will tend to have

relatively little weight in the duration formula. Thus, 30-year coupon

bonds and 40-year coupon bonds have similar durations; indeed they are

similar instruments, as the payments beyond 30 years into the future are

heavily discounted and not important today relative to the coupons that come

much sooner.

Macaulay actually gave two different definitions of duration that

differed in the specification of the discount factor. The first definition

of the duration of a bond of term m at time t , uses the yield to

11maturity of the bond:

E (t - t)s.et>t 1

(4) D(m,t) — -(t -t)r(t,t+m)Es.e

ti>t1

10 . . . -Hicks (1946) independently defined Naverage period," which is equival-ent to Macaulay's first definition of duration.

11The second argument, t , of duratior. will be dropped below incontexts where the interest rate r(t,t+m) is replaced by a constant.

15

The second definition of duration of a bond of term m at time t

uses prices of discount bonds as discount factors:

E (t. - t)sipd(tti)t >t

(4') D'(m,t) —E s.p (tt.)1

1

By either definition of duration, if the bond is a discount bond, then

the duration equals the term m, that is, we shall write

Otherwise, (since payments s are positive) the duration is less than the

time to maturity.

If a bond is selling at par and coupons are paid continually,then

duration using (4) is:

1-mr (t,t+m)

(5) D(mt) — r(t,t+m)

Thus, the duration of a perpetuity whose term is infinite is l/r(t,).

The duration using yields to maturity to discount D(m,t) is the

derivative of the log of p(t,T) , using (1), with respect to the yield to

maturity r(t,T)12

Thus, duration may be used as an index of the "risk"

of a bond. The concept of duration has thus played a role in the literature

on 'immunization' from interest rate risk of portfolios of financial

intermediaries. A portfolio is fully immunized if there is complete cash-

flow matching, that is, if the payments received on assets exactly equal

12This fact was used by. Hicks and was rediscovered by Samuelson (1945),Fisher (1966), Hopewell and Kaufman (1973) and others.

16

payments paid on liabilities. When such cash-flow matching is infeasible,

portfolio managers may instead try to match the overall duration of their

assets with the duration of their liabilities. As long as the term struc-

ture makes only parallel shifts, the yields on bonds of all terms being

increased or decreased by the same amount, then duration matching will

perfectly immunize the portfolio and there is no uncertainty about net

worth. However, the term structure rarely makes a parallel shift, long- -

term interest rates being more stable than short-term interest rates, and so

duration tends to overstate the relative riskiness of long-term bonds.

Other methods of immunization have been proposed that take this into account

(see Ingersoll, Skelton and Weil (1978)).

13111.6. Forward Rates

The time t discount bond forward rate applying to the interval from

t' to T , fd(t,t',T), alluded to in Section II. above, is defined in

terms yields to maturity and duration in Table 1, expression 1. Using (1)

one verifies that this expression is the same as the expression given in

13The earliest use of the term 'forward rate,' and the first indicationthat it can be thought of as a rate on a forward contract that can becomputed from the term structure appears to be in Hicks (1946) (firstpublished (1939). Kaldor [1939] speaks of forward rates and their inter-pretation as rates in forward contracts, but attributes the idea to Hicks.Macaulay (1938) speaks of computing "implicit interest rates" without makingan analogy to forward contracts (p. 30). Of course, the notion that longrates are averages of future short rates has a longer history; the earlierauthors appear not to have written of computing forward rates from the longrates, or of showing an analogy of such rates to rates in forward contracts.

I wrote Sir John Hicks asking if he had coined the term forward rate inthe term structure. He replied that he only remembers being influenced by a1930 paper in Swedish by Lindahl, later published in English (1939), whichwhich is couched, Hicks writes, "in terms of expected rates rather thanforward rates; it is likely that the change from one to the other is my owncontribution."

17

section II. above. The forward rate compounded once per h periods is

fd(t,t',T,h) — (exp(hfd(t,t',T))l)/hThe limit of expression (1) of Table 1 as t' approaches T

denoted fd(t,T,T) or just f(t,T) is the instantaneous forward rate:14

(6) f(t,T) rd(t,T) + (T-t)drd(t,T)/dT

or:

(7) f(t,t+m) rd(t,t+m) + mdrd(t,t+m)/dxn

It follows that the instantaneous forward rate follows the same relation to

the spot rate as does marginal cost to average cost. To see this relation,

think of in as output produced and rd(t,t+m) as price of a unit of

output. As with the familiar cost curves, the instantaneous forward rate

(marginal) equals the instantaneous spot rate (average) when in equals

zero, that is, f(t,t) — r . The forward rate is less than the spot rate

where the slope of the term structure is negative and is greater than the

spot rate where the slope of the term structure is positive. An example

'4McCulloch, who was concerned with the effects of taxation, writes((1975b), page 823) what appears to be a different expression for theinstantaneous forward rate. If we adopt some of the notation of this paperthis is:

f(t,T) — -{88(t,T)/8T)/((1-z)6(t,Tflwhere 6(t,T) is the price at time t of an after-tax dollar at time Tand z is the marginal tax rate. However, since &(t,T) —

exp((lz)(Tt)rd(t,T)) , his formula is identical to the one shown here,

i.e., the tax rate drops out of the formula expressed in terms of rd(t,T)

The tax rate should drop out because both the interest rate rd(t,T) andthe forward rate are taxable.

18

showing a term structure and forward rate curve is shown in Figure 2.15

Solving the differential equation (7) we can show:

-l(8) fd(t,t',T) — (T-t')

If(t,s)ds

J ti

Thus, the forward rate fd(t,t',T) is a simple average of the instantaneous

forward rates between t' and T

Par bond forward rates can also be computed. These are especially

useful if one wishes to make comparisons with spot interest rates as

commonly quoted, since longer-term bonds usually trade near par. At time t

one can guarantee for oneself a par bond issued at time t' and maturing at

time T (t t' � T) by buying at time t one discount bond maturing at

time T , buying discount bonds maturing continually between t' and T

whose principal accrues at rate c , and selling a discount bond maturing

at date t' such that the proceeds of the sale exactly equal the total

purchases made. If one then chooses c such that the number of bonds

maturing at time t' sold is 1, one will have guaranteed for oneself, in

effect, the rate of interest on a par bond at time t' maturing at T . The

par forward rate F(t,t',T) equals c or:'6

15lnstantaneous forward rates computed using McCulloch's data for largeT-t seem to be very erratic. Vasicek and Fong (1982) have suggested thatthe problem would be eliminated if McCulloch had used exponential splinesinstead of the ordinary splines of his procedure, however McCulloch (1984)has disputed whether this would solve the problem.

16This formula for the forward rate differs slightly from that in

McCulloch ((1975b) page 825). His formula replaces with 8(x,y)

— P(xy)' and divides by (1-z) where z is the marginal tax rate. His

formula is quite identical to the one shown here if z > 0. However, theVolterra-Taylor linearization (like that which follows immediately in thetext) of his expression in terms of instantaneous forward rates j identical

19

pd(t,t) - pd(tPT)(9) F (t,t',T) — __________________

P rT

J pd,dst,

Note the similarity between this expression and expression (3) for the par

spot rate, above.

The limit of expression (9) as t' approaches T is the same as the

values given by expressions (6) or (7) above for the instantaneous discount

forward rate, hence the omission in those expressions of the d or p sub-

script.

It can be shown that the par bond forward rate is a weighted average of

instantaneous forward rates, where the weights are proportional to the

prices of the discount bond maturing at the date to which the instantaneous

forward rate applies:17

rT

J Pd(t,f(t,5)ds(9') F (t,t',T) — t'

P rT

J pd(t5

This expression may be compared with the corresponding expression for

discount bonds, expression (8) above. This expression gives more weight to

instantaneous forward rates in the near future, rather than equal weight to

all forward rates as in expression (8).

to equation (10) belowand the tax rate drops out of that.

17See McCulloch (1977).

20

Using (8) for t — t' and substituting for pd(t,s) in the above

expression makes Fd(t,t' ,T) a functional of f(t,s) considered as a

function of $ . Following Campbell (1984), (1986), this functional can be

linearized around f(t,s) — R using a Volterra-Taylor expansion (Volterra,

(1959). We will refer to the linear approximation to the forward rate

F(t,t',T) as f(t,t',T) . This is:18

(10) f(ttT) et't)R - eTt JTe(t)f(ts)ds

Using expression (10) and that f(t,t,T) r(t,T) gives us again

expression 1 in Table 1, for this forward rate in terms of par interest

19rates only. The expression for forward rates on par bonds is the same as

that on discount bonds except that duration on par bonds D(s-t) replaces

the duration on discount bonds Dd(st) , s-t, s t', T . It might also

be noted that expression 1, Table 1 gives the true par forward rate

F(t,t' ,T) exactly if a slightly different, but less convenient, definition

of duration is used.2°

'8The lower case f here denotes a linear approximation to the uppercase F above. In contrast, for discount bonds the linear approximation isno different from the true forward rate, so lower case f is used for both.A discrete time version of (10) appears in Shiller (1979).

'9The quality of this approximation to F(t,t' ,T) is discussed in

Shiller, Campbell and Schoenholtz (1983). The approximation is good exceptwhen both t'-t and T-t' are large.

20If we use Macaulay's definition of duration using prices of discountbonds as discount factors, (4'), and compute duration at time t of a bondwhose continuous coupon is not constant through time but at time t+i

always equals f(t,t+i) , then one finds that duration is given by:

21

111.7. Holding Period Rates

A holding period rate is a rate of return to buying a bond (or sequence

of bonds) and selling at a later date. The simple discount bond holding

period rate hd(t,t',T), where t � t' T, is the rate of return from

buying at time t a discount bond maturing at date T and selling it at

date t'; see Table 1, expression (2). The rollover discount bond holding

period rate hd(t,t',T) where t < T < t' is the rate of return from

buying at time t a discount bond of term in — T-t , reinvesting ("rolling

over") the proceeds in another in-period discount bond at time t+m , and

continuing until time t' when the last in-period discount bond is sold

(Table 1, expression (3)).

The par bond holding period rate H(t,t',T) where t � t' � T is the

yield to maturity on the stream of payments accruing to someone who buys at

time t a par bond maturing at T , receives the stream of coupons between

t' and T , and sells the bond at time T . This holding period rate can

be defined as an implicit function of the coupon on the bond r(t,T) and

the selling price, which in turn is a function of r(t' ,T) as well as the

coupon r(tT) . This implicit function may be linearized around

H(t,t',T) — r(t,T) — r(t',T) — R to yield the approximate h(t,t',T)

t+m

D'(m) — p (t,s)dsJ dt

Using expression (9) (and the fact that F (t,t,T) — r (t,T) ) one findsthat F (t,t',T) equals the right hand sine of Table ?, expression 1 wherethe aboe D'(m) replaces D(in)

22

shown in Table 1, expression (2).21 The rollover par bond holding period

rate H(t,t' ,T) , t < T < t' is the yield to maturity on the stream of

payments accruing to someone who buys at time t a par bond maturing at

time T — t+m , reinvests proceeds in a par bond maturing at time t+2m

and continues until time t' when the last par bond is sold. The linear

approximation h(t,t',T) appears in Table 1, expression (3).

IV. Theories of the Term Structure

IV.l Expectations Theories of the Term Structure

The expectations hypothesis, in the broadest terms, asserts that the

slope of the term structure has something to do with expectations about

future interest rates. The hypothesis is certainly very old, although it

apparently did not receive an academic discussion until Fisher (l896).22

Other important early discussions were in Fisher (1930), Williams (1938)

Lutz (1940) and Hicks (1946). The expectations hypothesis probably derives

from observing the way people commonly discuss choices between long and

short debt as investments. They commonly speak of the outlook for future

interest rates in deciding whether to purchase a long-term bond rather than

a short-term bond as an investment. If interest rates are expected to

decline, people may advise "locking in" the high long term interest rate by

21The quality of this linear approximation to H(t,t',T) is generally

quite good; see Shiller, Campbell and Schoenholtz (1983) and Campbell(1986).

22Fisher (1896) appears to say that the market has perfect foresight(p. 91). I have been unable to find any earlier discussion of the expecta-tions theory of the term structure. Bohm-Bawerk (1891) discussed howexpectations of future short rates affect today's long rate, but appears toconclude that the term structure is always flat (p. 280). Perhaps there is ahint of the expectations theory in Clark (1895), p. 395. Malkiel (1966)claims (p. 17) that "one can find anticipations of the expectations theory"in Sidgwick (1887) and Say (1853). In reading any of these works, one is ledto conclude that the hint of the expectations theory is very slight.

23

buying a long-term bond. If everyone behaves this way, it is plausible that

the market yield on long-term interest rates would be depressed in times

when the short rate is expected to decline until the high demand for

long-term interest rates is eliminated. Thus, relatively downward sloping

term structures are indicative of expectations of a decline in interest

rates, and relatively upward sloping term structures of a rise.

Early term structure theorists apparently could not think of any formal

representation of the expectations hypothesis other than that forward rates

equalled actual future spot rates (plus possibly a constant).23 Early

empirical work finding fault with the expectations hypothesis for the

inaccuracy of the forecasts (Macaulay (1938), Hickman (1942), and Culbertson

(1957)) were later dismissed by subsequent writers who thought that the

issue should instead be whether the forward rates are in accord with a model

of expectations (Meiselman (1962), Kessel (1965)). Since the efficient

markets revolution in finance in the 1960's, such a model has generally

involved the assumption of rational expectations.

IV.2, Risk Preferences and the Expectations Hypothesis

Suppose economic agents can be characterized by a representative

individual whose utility function has the simple form:

(11) U — u(Ct—0

23Conard (1959) wrote, "I assume not only that expectations concerningfuture rates are held with confidence by all investors, but also that theseexpectations are realized. Only by adding this last assumption is itpossible to build a theory whose predictions can be meaningfully testedempirically." (p. 290)

24

where C is consumption at time t , p is the subjective rate of time

preference, and u(C) is momentary utility or "felicity." Calling v

the real value (value in terms of the consumption good rather than money) of

any asset or portfolio of assets including reinvested coupons or dividends,

a first order condition for maximization of expected utility is that:

(12) u'(C)v(t) — Et{(l+p)ttu(Ctf)v(t*)}, t < t < T

If there is risk neutrality, then u(C) is linear in C, and

u(C) — a + bC . It follows from (12) that:

Ev(t')(13) v(t (l+i)

If the asset is a discount index bond maturing at time T, then v(t') =

Pd(t ,T) and the left-hand side of this expression is one plus the expected

holding return compounded every t'-t periods, i.e., it is one plus

Et(ed(ttT) - l)/x , where x equals t'-t . This means that under

risk neutrality expected holding period returns as computed in the left-

hand-side of (13) will be equalized, i.e., will not depend on T . This in

turn suggests that a particular formal expectations theory of the term

structure follows from risk neutrality. Of course, risk neutrality may not

seem a very attractive assumption, but approximate risk neutrality might be

invoked to justify the intuitive expectations hypothesis described in the

preeding section. Invoking risk neutrality to justify an expectations

theory of the term structure was done by Meiselman (1962), Bierwag and Grove

(1967), Malkiel (1966), Richards (1978) and others.

There are, however, fundamental problems with the expectations hypothe-

25

sis as derived from risk neutrality. It is not possible for all expected

holding period returns as defined in the left-hand-side of (13) to be

equalized if future interest rates are uncertain. This point was emphasized

by Stiglitz (1970), who also attributed it to C. C. von Weizsacker. If

one-period expected holding period returns are equalized for a one-period

compounding interval, then =Eopd(l,2)/pd(O,2) . If two-period

expected holding period returns are equalized for a two-period compounding

interval then 1d0'2 — Eo(l/pd(l,2))/pd(O,l) . It follows that

Eopd(l,2) — l/Eo(l/(pd(l,2)) . This is a contradiction, since Jensen's

inequality states that for any random variable x that is always greater

than zero, unless x is nonstochastic, E(x) > l/E(l/x))

For index bonds, the equation (13) implies that interest rates are

random, and that thus Jensen's inequality does not come into play (Cox,

Ingersoll and Ross (1981), LeRoy (1982a)). This can be easily seen by

substituting pd(t,t') for v(t) in (13). Since t' is the maturity date,

and the real value of the index bond at maturity is specified as v(t') — 1,

it follows that v(t') is not random. Clearly, (13) then implies that

Pd(t,t) is not random either. It will be known with certainty at any date

before t . Thus, while risk neutrality gives us an expectations hypo-

thesis, it gives us a perfect foresight version that is extreme and uninter-

esting. It would be possible to alter the utility function (11) to allow

the subjective rate of time preference i to vary through time, and that

would give us a time varying yield curve. Still, we would have a perfect

foresight model and a model in which preferences alone determine interest

26

24rates.

Risk neutrality is of course not a terribly attractive assumpticri,

given various evidence on human behavior. The theoretical literature does

not appear to contain any argument for appealing simple restrictions on

ferences or technology that singles out for us an attractive versiofl of e

expectations hypothesis; see LeRoy (1982a) for a discussion. Cox, Ingersc.

and Ross (1981) offered two sets of assumptions other than risk neutralitr

that can produce an expectations hypothesis for the term structure: cie

involving locally certain consumption changes, the other involving state-

independent logarithmic utility. 3ut by offering such special cases they

not giving any reason to suspect that the expectations hypothesis shc.ild

taken seriously in applied work.

Applied workers, actually, have rarely taken seriously the risk

neutrality expectations hypothesis as it has been defined in the theceti1

literature, and so the theoretical discussion of this expectations h-poth-

sis may be something of a red herring. The applied literature has defined

the expectations hypothesis to represent constancy through time of difer-

ences in expected holding returns or, constancy through time of the differ-

ence between forward rates and expected spot rates, and not that these

constants are zero. We shall see in the next section that these theores c-.

be described as assuming constancy of the 'term premia.'Campbell (l9E) h

24Cox, Ingersoll and Ross (1981) emphasize that risk neutrality tsefdoes not necessarily imply that interest rates are nonstochastic. Utity snot concave, and investors could be at a corner solution to their maimiz-ation problem in which (13) does not hold. However, they argued that :n tLscase the expectations hypothesis will not generally be valid.

LeRoy (1983) showed (correcting errors in his own papers (l982a(1982b)) a sense in which when there is near risk neutrality," that s,when utility functions are nearly linear, the expectations hypothesis isapproximately satisfied.

27

stressed that some of the important conclusions of this theoretical litera-

ture do not carry over to the definitions of the expectations hypothesis in

the empirical literature.25

IV.3. Definitions of Term premia26

There is little agreement in the empirical literature on definitions

of term preinia, and often term premia are defined only for certain special

cases. Here, some definitions will be adopted which are clarifications and

generalizations of definitions already commonplace. As suggested in the

discussion in the preceding section, economic theory does not give us

guidance as to how to define term premia, and so choices will be made here

to retain essential linearity, which will simplify discussion.

The forward term premium f.(tt'T) , i — p,d will be defined as

the difference between the forward rate and the expectation of the

corresponding future spot rate. Unless otherwise noted, this expectation

will be defined as a rational expectation, i.e., Et is the mathematical

expectation conditional on information available at time t . Thus we have:

(14) fi(t,t'T) — f.(t,t',T) - Er.(t',T) , t < t' < T , i — p, d

The holding period term premium h.(t,t',T) for t < t' < T will be

Ingersoll and Ross (1981) showed that if there are fewer relevantstate variables in an economy than there are bond maturities outstanding andif bond prices follow Ito processes, then only one version of the rational

expectations hypothesis, what they called the "local expectations hypothe-sis," can obtain in a rational expectations equilibrium. Campbell showedthat this conclusion hinges on the assumption of a zero, not just constant,risk premium. He also showed a sense in which the other versions of theexpectations hypothesis (which they claimed to reject as inconsistent withrational expectations equilibrium) may not be im?ortantly different fromtheir local expectations hypothesis.

of this section follows Campbell arid Shiller (1984) and Campbell(1986).

28

defined as the difference between the conditional expected holding period

yield and the corresponding spot rate:

(15) h,i(t,t,T) — Eh.(t,t',T) - r.(t,t') , t < t'< T , i — p, d

The rollover term premium r .(t,t',m) for t < t+rn < t' will be defined

as the difference between the yield on a bond maturing at time t' and the

conditional expected holding period return from rolling over a sequence of

rn-period bonds:27

(16) (t,t',m) r (t,t') - E h (t,t',t-4-m), t < t+m < t', i — p, d.

r,i i t i

Although earlier authors did not always clearly intend rational expecta-

tions, and often used different conventions about compounding, we can

loosely identify the above definitions with definitions given by earlier

authors. Hicks (1946), who is commonly credited with first defining these

in the terni structure literature, referred to both f(tt' ,T) and

as the "risk premium."28 Because of a subsequent liquidity

theory of interest by Lutz (1940), and analogy with the Keynes' (1936)

liquidity preference theory, the risk premium has also become known as the

"liquidity premium." In this survey, the phrase "term premium" will be used

throughout as synonymous with risk premium and liquidity premium; it is

preferred to these because the phrase does not have an association with a

specific theory of the term structure. The holding period term premium

27Note that this risk premium has the form interest rate minus expected

holding yield,in contrast to expected holding period yield minus interestrate in the preceding expression. This way of defining risk premia seems tobe conventional; the rate on the longer asset comes first with a positive sign.

28See Hicks (1946), p. 147.

29

h,i(t,t,T) is referred to as the expected "excess return" in finance

textbooks.

From the definitions in Table 1, there are simple proportional

relations between holding period term preniia and forward rate term prernia:

(17) hj(t,t',T) — tD.(T-t)/D.(t'-t) - i)tf .(t,t',T) if t < t' < T

We also have the following relations for the rollover term premium, where

t < t+m < t' , t'-t = sm , S integer:

s-i(18) t .(t,t',m) — (i/D.(t'-t)) Z [D.(km+m) - D.(km)] (t,t+km,t+kni+rn)r,i 1

k—O1 f,i

s-i

(19) Zrj(t,t',m) (l/D.(t'-t)) E

IV.4. Early Presumptions Pertaining to the Sign of the Term premium

Hicks (1946) thought that there was a tendency for term premia to be

positive. In this context he referred to the forward rate term premium

f,i(tt ,T) , but if this term premium is always positive then by (17)

above so must the holding period term premium h,i(t,t',T)and by (18) the

rollover term premium rj(t,t',m)

Hicks' reasons to expect that term premia should be positive had their

motivation in the theory of "normal backwardation" in commodity forward

markets of Keynes (1930). Hicks wrote:29

• . . the forward market for loans (like the forwardmarket for commodities) may be expected to have aconstitutional weakness on one side, a weakness which

(1946), p. 146.

30

offers an opportunity for speculation. If no extrareturn is offered for long lending, most people (andinstitutions) would prefer to lend short, at least inthe sense that they would prefer to hold their money ondeposit in some way or other. But this situation wouldleave a large excess demand to borrow long which wouldnot be met. Borrowers would thus tend to offer betterterms in order to persuade lenders to switch over intothe long market.

He offered no evidence (other than that on average risk premia them-

selves) that would support such a "constitutional weakness" on one side of

the forward market.

Lutz (1940) offered a "liquidity theory of interest" that also pre-

dicted positive term premia:

.The most liquid asset, money, does not bear inter-est. Securities, being less liquid than money, bear aninterest rate which is higher the longer the maturity,since the danger of capital loss due to a change in theinterest rate in the market is supposed to be the

greater (and therefore luidity the smaller) the longerthe security has to run.

His theory appears to ascribe term premia to own-variance, contrary to

received wisdom in finance theory today.

Such theories were disputed by Modigliani and Sutch (1966) by merely

pointing out that it is not clearly rational for individuals to prefer to

lend short or to be concerned with short-term capital losses. If one is

saving for a child's college education 10 years ahead, it is least risky to

put one's savings in the form of a (real) 10 year bond rather than roll over

short bonds. They proposed as an alternative to Hicks' theory the "pre-

ferred habitat theory." A trader's habitat is the investment horizon he or

she is most concerned about, and that person will prefer to borrow or lend

at that term. There is a separate supply and demand for loanable funds in

30Lutz (1940), p. 62.

31

each habitat, which could give rise to any pattern of term premia. Trader's

may be "tempted out of their natural habitat by the lure of higher expected

returns"31 but because of risk aversion this will not completely level term

premia. The idea that individuals have a single habitat must be described

as heuristic.32 The intertemporal capital asset pricing model typically

assumes maximization of an intertemporal utility function that involves the

entire future consumption stream, with exponentially declining weights, and

thus no single "habitat." However, the Modigliani-Sutch conclusion that

term premia might as well, on theoretical grounds, be positive as negative

seems now to be generally accepted.33

IV.5. Risk Preferences and Term uremia

If the representative agent maximizes the utility function (11), and

therefore satisfies the first-order condition (12) then it follows that:34

(20) er2nldl,r2) — E S(r1,r2)

31Modigliani and Sutch (1966), p. 184.

32Cox, Ingersoll and Ross (1981) consider an economy in which allinvestors desire to consume at one fixed date. They find that the riskpremium, defined as the expected instantaneous return minus the instant-aneous interest rate, may not be lowest for bonds maturing at this date.Still, they argue that a preferred habitat theory holds f the habitat isdefined in terms of a "stronger or weaker tendency to h. against changesin the interest rate." (p. 786).

LeRoy (1982) has argued that the risk premia are likely to bepositive on theoretical grounds in a model without production, but had noresults on the sign of the risk premium when production is introduced.

34To show this, use v(t') — v(r2) — 1 and v(t) — v(r1) — pd(rl,rO) —

exp((r2rl)rd(rl,r2)) in (12) where l < so that rd(rl,r2) as welt asv(r1) is known at time Tj.

32

where S(r1,r2) is the marginal rate of substitution between time and

r2.For equation (20), the precise definition of S(r1,r2) will depend on

whether we are dealing with index bonds or bonds whose principal is defined

in nominal terms, that is, on whether rd(rl,r2) is a real or nominal rate.

With index bonds, S(r1,r2) is defined as u,(C(r2))/(u?(C(r1))(l+)(72Tl)).

With bonds whose principal is defined in nominal terms S(r1,r2) is defined

as u'(C(r2))/u'(C(r1)) x ((T1)/(T2))/(l+)(T2nl). Here, (r) is a

commodity price index at time r, that is, the price of the consumption good

in terms of the unit of currency. Thus, S(r1,r2) is the marginal rate of

substitution between consumption at time and consumption at time if

the bond is an index bond and between a nominal dollar at time and a

nominal dollar at time if the bond is a conventional nominal bond.36

It follows from equation (20) for t < t' < T (setting (r1,r2) in (20)

as (t,t'), (t,T) and (t',T)) that

-(T-t)r t,T) — e -(t'-t)r t,t') -(T-t')r t',T)

(21)

+ cov(S(t,t'),S(t' ,T))

In order to put this in terms of the above definitions of term premia, we

use the linearization eX (l+x) for small x to derive from the above,

(22) f cov(S(t,t'),S(t',T))/(T.t')

351t follows that increasing the uncertainty at time about

consumption at will, if there is diminishing marginal utility, lowerrd(Tl,r2). This point was made by Fisher (1907) p. 214.

36Beninga and Protopopadakis (1983) describe the relation of risk premiaon nominal bonds to risk premia on index bonds.

33

The term premium d(t,t,T) depends on the covariance between the marginal

rate of substitution between t and t' and the marginal rate of substitution

between t' and T.37 If this covariance is negative, then forward rates tend

to be above expected spot rates, as Hicks originally hypothesized, and risk

the premium is positive.38 In the case of index bonds, a negative covariance

means that if real consumption should increase faster than usual between t

and t' , it tends to increase less fast than usual between t' and T. In the

case of nominal bonds, the interpretation of the sign of the term premium is

less straightforward. But consider the utility function u(C)log(C).

Then for nominal bonds S(t,t') equals nominal consumption at time t divided

by nominal consumption at time t÷l. Then, the nominal term premium

f would tend to be positive if it happens that when nominal

consumption increases faster than usual between t and t it tends to

increase less fast than usual between t' and T.

One can also derive (taking unconditional expectations of (20)) an

expression like (21) for unconditional expectations:

37 .

LeRoy (1984) gives an expression for the term premium defined as theexpected real j-period remrn on an i-period nominal bond minus the returnto maturity of a j-period real bond.

38Woodward [1983] discusses term premia in terms of the serialcorrelation of marginal utility of consumption rather than the serialcorrelation of marginal rates of substitution. Her principal result is thatin a case where the correlation conditional on information at t betweenu'(c,) and u'(cT) is negative then the sign of the term premium may besensitive to the definition of the premium. She defines as an alternativedefinition of the term premiui, the "solidity premium" based on forward andactual discounts. The negative correlation she defines is an unlikelyspecial case. Actual aggregate consumption in the United States roughlyresembles a random walk (Hall, 1978), for which the correlation she definesis positive.

34

Ee(Tt)rd(tT) — Ee •(t't)rd(t,t') Ee (Tt')rd(t',T)

(23)

+ cov(S(t,t'),S(t',T))

From which, by a linearization, we have

(24) E(tIfd(t,t,T)) -cov(S(t,t'),S(t',T))/(T-t')

Thus, the mean tern premium d(t,t,T) is positive if the unconditional

covariance between S(t,t') and S(t',T) is negative. Such a negative covari-

ance might be interpreted as saying that marginal utility is 'unsmooth'

between r—t and r—T. This means that when detrended marginal utility

increases between r—t and i--t' it tends to decrease between t' and T. A

positive covariance, and hence a negative term premium, would tend to occur

if marginal utility is 'smooth' between r—t and r—T.

If the values of bonds of all maturities are assumed to be deeermiriis-

tic functions of a small number of state variables that are continuous

diffusion processes, then theoretical restrictions on risk premia beyond

those defined here can also be derived e.g. Brennan and Schwartz (1983),

Cox, Ingersoll and Ross (1981), Dothan (1978), Langetieg (1980), Marsh

(1980), Richard (1978) and Vasicek (1978). When bond values are such

deterministic functions of diffusion processes, if the restrictions did not

hold there would be riskless arbitrage opportunities. The assumption of

such a state variable representation has been convenient for theoretical

models. It has even led to a complete general equilibrium model of the term

35

structure in a macroeconomy, Cox, Ingersoll and Ross (1985a), (l985b), a

model subjected to empirical testing by Brown and Dybvig (1986).

V. Empirical Studies of the Terni Structure

V.1 Empirical Expectations Hypotheses for the Term Structure

One need not assume rational expectations to proceed with studying an

expectations theory of the term structure if one has data on expectations or

can infer expectations from other data. The first study of the term struc-

tures using an expectations model was performed by Meiselman (1962).

Meiselman proposed the "error learning hypothesis" that economic agents

revise their expectations in proportion to the error just discovered in

their last period expectation for today's one-period rate. This hypothesis

then implies that f.(t,t+n,t+n+l)-f.(t-l,t+n,t+n+l) — a + b(r(t,t+l) -

f.(t-l,t,t+l)). He estimated a and b by regression analysis using U. S.

Durand's annual data l9Ol-51 for n — 1, 2, . . . , 8. I-fe took as

encouraging for the model that the signs of the estimated b were all

positive and declined with n. However, Buse (1967) criticized his

conclusion, saying that ". .. .such results are implied by any set of smoothed

yield curves in which the short-term interest rates have shown a greater

variability than long-term interest rates."

It was pointed out later by Diller (1969) and Nelson (1970a) that the

error learning principle is a property of optimal linear forecasts. They

found that the coefficients b that Meiselman estimated compared rathern

favorably with the coefficients implied by an estimated linear forecasting

equation. However, the univariate form of the error learning principle

proposed by Meiselman applies only to univariate optimal linear forecasts

36

(Shiller (1978)), and thus the Meiselman theory is unfortunately restric-

tive.

Other authors have used survey expectations data for market expecta-

tions of future interest rates. Survey methods seem particularly attractive

since surveys can be focussed on the institutional investors who hold most

government and corporate bonds, and who are probably not well-described in

terms of the expected utility of consumption models described in the

preceding section.

B. Friedman (1979) used data 1969-78 from a quarterly survey of

financial market participants by the Goldsniith-Nagan Bond and Money Market

Letter. He found that the term premium on U. S. treasury bills d(t,t+l,t+2)

and d(t,t+2,t+3) (where time is measured in months) was positive on average

and depended positively on the level of interest rates. He showed (1980c)

that his model differed substantially from a rational expectations model, in

that the survey expectations could be improved upon easily. Kane and Malkiel

(1967) conducted their own survey of banks, life insurance companies and

nonfinancial corporations to learn about the relation of expectations to the

term structure of interest rates. They learned that many investors seemed

not to formulate specific interest rate expectations (especially for the

distant future) and those that did did not have uniform expectations. Kane

(1983) found using additional Kane-Malkiel survey data 1969-72 that term

premia appear positively related to the level of interest rates.

V.2. The Rational Expectations Hypothesis in Enmirical Work

Although the rational expectations hypothesis regarding the term

39See Board of Governors [1985], pp. 20 and 54. Of course, individualsultimately have claims on the assets of these institutions; still there isan institutional layer between them and the bonds held on their behalf.

37

structure has had many forms, it has its simplest form used in empirical

work in terms of the continuously compounded yields discussed here. Often,

the other forms of the hypothesis do not differ importantly from that

discussed here (see Shiller, Campbell and Schoenholtz (1983), Campbell

(1986)). In the definition to be used here, the rational expectations

hypothesis is that all term premia: h(t,t+n,t±m+n) , 0 < m , 0 < n

'f .(t,t+n,t+m-4-n) , 0 < m, 0 < n r .(t,t+n,m) , 0 < m < n , do not

depend on time t.4° This means that all term premia depend only on

maturity and not time, and the changing slope of the term structure can onl7

be interpreted in terms of the changing expectations for future interest

41rates.

The literature testing forms of the rational expectations hypothesis

like that defined here is enormous.42 It is difficult to summarize what we

know about the expectations hypothesis from this literature. We are studyir

a two-dimensional array of term prernia; term premia depend on m (the

40When dealing with par bonds, the expectations model defined hererelates to the linearized model. The assuiption here is that the point oflinearization R does not depend on the level of interest rates, otherwisethe model will not be linear in interest rates.

41Note that in this expectations hypothesis, stated in terms ofcontinuously compounded yields, it is possible for all risk premia to bezero. We do not encounter the Jensen's inequality problem alluded to abovein connection with risk neutrality. The problem alluded to by vonWeizsaecker and Stiglitz was essentially one of compounding, and iseliminated when we couch the model in terms of continuously compoundedinterest rates.

42The literature has to do almost entirely with nominal interest rates.as an observed term structure of index bonds is observed only for briefperiods in certain countries. Campbell and Shiller (l986b) in effect lookedat the real term structure in the postwar United States corporate stockprice data by correcting the dividend price ratio for predictable changes ireal dividends, leaving a long-term real consol component of the dividendprice ratio. The expectations hypothesis was not supported by the evidence.

38

maturity of the forward instrument and n (the time into the future that the

forward instrument begins). Term premia may be approximately constant for

some m and n and not for others; certain functions of term premia may be

approximately constant and not others. Term premia may be approximately

constant for some time periods and not others, or in some countries and not

others.

Testing for the constancy of term premia ultimately means trying to

predict the right hand side of the equations defining terni premia (equations

(14), (15) or (16) above) from which the conditional expectations operator

is deleted, in terms of information at time t. This means predicting

either excess holding period returns or the difference between forward rates

and corresponding spot rates in terms of information at time t. Because of

the relations between the definitions of term premia (equations (17), (18),

or (19) above) it does not matter whether the regression has excess holding

yields or the difference between forward rates and corresponding spot rates

as the variable explained; the difference has to do only with a multipli-

cative constant for the dependent variable. Of course, most studies do not

use the exact definitions of term prenlia defined here, in terms of continu-

ously compounded rates or, in the case of par bonds, linearized holding

yields, but the differences in definition are generally not important.

Some studies may report some tests of the rational expectations

hypothesis that have the appearance of something very different; for

example, Roll (1970) tested (and rejected using 1-13 week U. S. treasury

bill data 1949-64) the martingale property of forward rates by testing

whether changes in forward rates fd(t,t'T) - fd(tl,t',T) are serially

correlated through time t. But in fact testing the hypothesis that there is

39

no such serial correlation is no different from testing the hypothesis that

changes in the difference between forward rates and corresponding spot rates

cannot be predicted based on information consisting of past changes in

forward rates. For another example, some researchers have noted that for

large m and small n the holding return h.(t,t+n,t+m) is approximately equ.a.

to r.(t,t+m)-r.(t+n,t+n+m), the change in the long rate, divided by D.(n).

If n is very small, l/Di(n) is a very large number, and the excess holding

return is heavily influenced by the change in the long rate. The rational

expectations hypothesis thus suggests that r.(tt+m)-r(t+nt+n+m) is

approximately unforecastable, and hence that long rates are in this sense

approximately random walks. The random walk property for long-term interest

rates was tested by Phillips and Pippenger (1976), (1979), Pesando (1981),

43(1983), and Mishkin (1978).

Of all the studies of the rational expectations hypothesis for the teri

structure, of greatest interest are the results in which the explanatory

variable is approximately (or approximately proportional to) the spread

between a forward rate f(tm.n) and the spot rate of the same maturity as

the forward rate, r.(t,t-i-m). This spread forecasts the change in r.(t,t+ni)

over the next n periods. Regressions in the literature that can be inter-

preted at least approximately as regressions of the actual change in spot

rates r.(t+n,t+m+n) - r.(t,t+m) on the predicted change f.(t,m,n)-r.(t,t+n

43The random walk property is an approximation useful only under

certain assumptions (see Mishkin (1980), Begg (1983)). Phillips andPippenger (1978) (1980) used the random walk approximation to assert thatthe Modigliani-Sutch (1966) and Modigliani and Shiller (1973) distributedlag regressions explaining the long rate must be spurious. Looking at theout-of sample fit of the equation does not suggest that term-structureequations like that in Modigliani-Shiller are completely spurious: see And:and Kennickell (1983).

40

and a constant are shown in Table 2.

What is clear front Table 2 is that the slope coefficient is quite far

below one - and often negative - for low forecast horizon n, regardless of

the maturity in of the forward interest rate, but rises closer to one for

higher n. This result may at first seem counterintuitive. One might have

thought that forecasts into the near future would be more accurate than

forecasts into the more distant future; the reverse seems to be true.

When both n are in are small, both less than a year or so, the slope

coefficients are positive (the right sign) but substantially lower than one.

Thus, for example when two month interest rates exceed one month rates by

more than the average term premium ri(t2l) the one month rate does tend

to increase as predicted, but by substantially less than the predicted

44amount.

The results in Table 2 look especially bad for the rational expecta-

tions hypothesis when the forecast horizon n is small (a year or less in the

Table) and the maturity of the forward rate m is large (20 or more years in

the Table). Here, the spread between the forward rate and spot rate

predicts the wrong direction of change of interest rates. One might

consider it the "essence" of the rational expectations hypothesis that art

unusually high spread between the forward rate and current spot rate

portends increases in interest rates, not the decrease as observed.

It is helpful in interpreting this result to consider a caricature, the

44 .

Regressions for large n and small m are not in Table 2. Since such

forward rates are very sensitive to rounding error or small noise in thelong term interest rates, we cannot accurately measure such forward rates.

Some more favorable results for the expectations theory with small nwere reported in Shiller (198la); however these results were later found tobe related to a couple anomalous observations (Shiller Campbell andSchoenholtz (1983)).

41

case of a perpetuity (for which m — ) paying coupon c once per period, and

where, for simplicity, the term premium is zero. Then the price of the

perpetuity p(t,) equals coupon over yield c/r(t,,l), and the spread

between the one-period-ahead forward consol yield and the one-period spot

rate is proportional to the spread between the consol yield and the one-

period rate. When the consol yield is above the one-period interest rate

r(t,t+l,l) then its current yield c/P(t) is greater than the one-period

rate r(t,t+l,l). This would suggest that consols are then a better invest-

ment for the short run than is short debt. Since the rational expectations

hypothesis with zero term premium would deny this, it follows that the

consol yield r(t,c,l) should be expected to increase over the next period,

producing a decline in price, a capital loss that offsets the high current

yield. But, in fact when the consol yield is high relative to the short

rate the consol yield tends to fall subsequently and not rise.45 The

capital gain tends to augment rather than offset the high current yield.

The naive rule that long bonds are a better investment (in an expected value

sense) whenever long rates are above short rates is thus confirmed.

Froot (1987) attempted a decomposition of the departure from 1.00 of

the coefficient in Table 2 here into two parts: a part due to expectation

error and a part due to time-varying term premium. He used used survey data

published in the investor newsletter Reporting on Governments (continuing

the Goldsmith-Nagan data series) to represent expectations. He found that

for three-month ahead forecasts of three-month rates, the departure from

45That long rates tend to move opposite the direction indicated by theexpectations theory was first noted by Macaulay (1938): "the yields of bondsof the highest grade should fall during a period when short rates arehigher than the yields of bonds and rise during a period in which shortrates are the lower. Now experience is more nearly the opposite." p. 33.

42

1.00 is due primarily to time-varying term premium. But for forecasts of

changes in 30-year mortgage rates, expectations error bears most of the

blame for the departure of the coefficient from i.oo.46

V.3 The Volatility of Long-Term Interest Rates

According to the rational expectations theory of the term structure, n-

period interest rates are a weighted moving average of one-period interest

rates plus a constant term premium; that is, from (16) and Table

1

(25) r.(t,t+m) — D (in) E (D.(k+l)-D (k))E r.(t+k,t-i-k+1) +1 k—01 i ti in

i — p,d.

Where r,i (t,t+m,1) is constant through time. Since long moving

averages tend to smooth the series averaged, one might expect to see that

long rates are a very smooth series. Are long-term rates too "choppy"

through time to accord with the expectations theory? It is natural to

inquire whether this is so, and, if so whether it is possibly related to the

poor results for the expectations hypothesis that were obtained in the Table

2 regressions.

Because of the choppiness of long-term interest rates, short-term

holding returns on long-term bonds, which are related to the short-term

change in long-term interest rates, are quite variable. Culbertson (1957),

in his well-known critique of expectations models of interest rates, thought

the volatility of holding yields was evidence against the model. He showed

46See Froot (1987) Table 3. Note that his regressions are run in aslightly different form than in Table 2 here, but that our Table 2coefficients can be inferred from his.

47 .For par bonds, it is necessary to evaluate D (k) with (5) using afixed point of linearization r, so that (25) will b linear in interest rates.

43

a time-series plot of holding yields on long bonds and, noting their great

variability, remarked "what sort of expectations, one might ask, could

48possibly have produced this result?"

It is possible, using the expectations hypothesis, to put limits on the

variability of both long-term interest rates themselves and on short-term

holding returns on long-term debt. The expectations hypothesis implies that

r.(t,t-i-m) — Etr.*(t,tIm) ÷ where r1*(t,t+m) is the "perfect foresight" or

"ex post rational" long-term interest rate defined as:

* 1ml(26) r. (t,t-fm) — D.(m) E(D.(k+l)D.(k))Er.(t+k,t+k+l) i — p,d.1 1 k—01

It follows that r.*(t,t+ni) — r.(t,t+m) + + u where u is a forecast1 1 m m nit

error made at time t and observed at time t+m. Since u is a forecastnit

error, if forecasts are rational u cannot be correlated with anything

known at time t; otherwise the forecast could be improved. Hence u mustnit

be uncorrelated with r.(t,t-4-ni). Since the variance of the sum of two

independent variables is the sum of their variances it follows that

Var(r.*(t,t+m)) — Var(r.(t,t+m)) + Var(u) and since Var(u) cannot be

negative, the rational expectations model implies (Shiller (l979):

*(27) Var(r.(t,t+rn)) � Var(r. (t,t+rn)) 1 — p,d

so there is an upper bound to the variance of rn-period rates given by the

variance of r.*(t,t.+m). One can also put an upper bound to the variance of

48Culbertson (1957), p. 508.

49LeRoy and Porter (1981) also noted this inequality in a differentcontext.

44

the holding period return in terms of the one-period rate (Shiller (1981a)):

(28) Var(h.(tt+l,t+m)) � (D(m)/D(l))Var(r(tt+l)) i — p,d

where in the case i—p of par bonds D(m) and D(l) are computed from equation

(5) above with interest rate 2r where r is the point of linearization.

Both of the above inequalities were found to be violated using U.S.

data and m of 2 or more years (Shiller (1979), (1981a), (1986),

Singleton (1980b)). Their rejection could have either of two interpreta-

tions: The rational expectations hypothesis could be wrong, in such a way

as to make long rates much more volatile than they should be. Or, the

measures of the upper bound in the inequalities could be faulty: the

measures of Var(r(t,t+m)) or Var(r.(t,t+l)) could understate the true

variance.

The latter view of the violation of the inequalities was argued by

Flavin (1983) who showed with Monte-Carlo experiments that if the one-period

interest rate r..(t,t+l) is a first order autoregressive process with

autoregressive parameter close to one (see the next section) the inequali-

ties are likely to be violated in small samples even if the rational

expectations model is true. Such a process shows a great deal of

persistence, and r.(t,t+l) may thus stay on one side of the true mean

throughout the sample thus, the sample variance around the sample mean of

r.(t,t+l) or of r(t,t+m) may be strikingly downward biassed measure of

their true variance.

Flavin's is apparently a viable interpretation of the excess volatility

results. The volatility tests do not allow us to tell whether there is too

45

much variability in long rates or just nonstationarity in short rates. They

allow us to reject the idea that movements in long rates can be inter-

preted in terms of rational expectations of movements in short rates within

the range historically observed.

V.4 Encouraging Results for the Rational Expectations Hvtothesis

It does not follow from the Table 2 results with small n that the

spread between very long-term interest rates and short-term interest rates

is totally wrong from the standpoint of the expectations hypothesis. One

way of summarizing the relatively good results (Shiller [1986]) for this

spread for larger n is to compute both actual and perfect foresight spreads

between very long-term interest rates and short-term interest rates.

Defining the spread S.(m) — r.(t,t+m) - r(tt+l) (in integer > 1) then the

rational expectations hypothesis implies:

(29) S.(In) — ES(m) + i — p,d

(30) St(m) ri*(t,t±m) - r.(t,t+l)* *

From the definition (26) of r. (t,t+m), it can be shown that S.(m) is the

duration weighted average of expected changes in the n-period rate.

Equation (29) thus asserts that when long rates are high relative to short

rates the weighted average of increases in short rates should tend to be

high. The values of Sd(m) and Sd(m) are plotted for m—l0 in figure 4 for

those years for which data are available in the appendix. The correspon-

dence between Sd(m) and Sd(m) is apparent. This might be viewed as a

striking confirmation of some element of truth in the expectations hypothe-

sis. Moreover, a variance inequality analogous to (28) above is that

var(S(m)) Var(S(m)). This variance inequality is satisfied by the data.

This result does not by itself establish whether or not Flavin's view of t-.e

46

variance inequality violation described in the preceding section is

50correct.

One must consider, though, whether this apparent confirmation of the

expectations theory for large m cou.d also be described in a less inspiring

way: as reflecting largely just that the long rate is much smoother than

the short rate. In fact, the correspondence in postwar U. S. data between

S(l0,l) and Std(lO,l) would still be apparent if the long rate r(t,t-4-l0)

had been a simple trend through the path of short rates. Short-term

interest rates have shown an apparent tendency to revert to trend, thus a

duration weighted average of future changes in the short rate is approxi-

mately minus the detrended short rate.

It was shown by Modigliani and Shiller [1973] and Shiller [1972]

(following Sutch [1967]) that a regression of a long rate on a distributed

lag of short rates produces distrib..ted lag coefficients that crudely

resembled the "optimal" distributed lag coefficients implied by an autore-

gression in first differences for the short rate. Similarly a regression of

the long rate on a distributed lag of short rates and a distributed lag of

inflation rates is consistent with a vector autoregression in first differ-

ences using the short rate and the inflation rate.51 The basic principle

50 *Indeed, even if S(m) and S.(m) look good by this criterion, there

could be some small noise contai.inafing S .(in) which, if the noise is not

highly serially uncorrelated, could cause*olding period yields to be muchmore volatile than would b implied by the expectations model. Moreover, theappearance of S.(m) and S.(m) may also be relatively little affected by agross overstatement of the variability of r.(t,t+m), so long as it issubstantially less variable than the short ate. See Shiller (1986).

51Distributed lag regressions explaining the term structure have haddifferent functional forms: see for example Bierwag and Grove (1967),Cargill and Meyer (1972) or Malkiel (1966). A comparison of eight differentdistributed models of the term stru:ture is in Dobson, Sutch and Vanderford

47

of these analyses can be illustrated by assuming for simplicity here (as in

Flavin) that the short rate r (t,t+l) follows a first-order autoregressive

process around a mean i: r(t+l.t+2)- p —

A(r(t1t+l)- p) + 0 < A 1,

where is a realization of a random variable with zero mean independent of

tk k"O. The optimal forecast at time t of r(t+k)t+k-i-l) is:

(31) Er(t+k,t-4k+l) — p + Ak(r(t,t+l)p).

From (25) and (5) for m — and i — p the consol yield is given by:

(32) r(t,t+cc) — (l-7)Z kEr (t,t+l) + •

where 'y e and r is the point of linearization. Thus, the consol yield

is a sort of present value of expected future one period rates. Together,

(31) and (32) imply:

(33) r (t,t+o) — (1-7)p r (t,t+1) + 'i'.

(l-7A)

One can therefore evaluate the rational expectations model by first

regressing r (t+l,t+2) on r(tt+l) and a constant (i.e. estimating A in

31), and computing the theoretical coefficient of r(t) using (33). This

theoretical coefficient can be compared with the slope coefficient in a

regression of r(t, t+co) onto r (t,t-i-l) and a constant. Now, in fact, our

assumption that r (t,t+l) was forecast by the market according to (31) would

imply that (33) should hold without error. However, it can be shown that

whether or not r(t,t+l) is an AR-l process if is the

(1976).

48

optimal forecast of r(t+kt+k+l) conditional on an information set that

includes r(t,t+l), then a theoretical regression of r (t,t+a) on r(t,t+1)

and a constant should produce the coefficient (l--y)/(1---1A), where A is the

slope coefficient in a theoretical regression of r (t+l,t+2) on r(t,t+l),

(Shiller [19721). With this assumption, there is an error term in (33)

reflecting information held by market participants beyond r(t,t+l).

Comparing such estimated coefficients using more complicated autoregression

models was the method used in the aforementioned papers.

Note that if y and A are both near one, then (l-y)/l--1A) may be very

sensitive to A. When data are limited, we cannot tell with much accuracy

what A is, and hence cannot pin down what is the value of (l-7)/(l-7A).

Thus, we cannot say with much assurance whether the consol yield in fact is

or is not too volatile.

Such simple comparisons of estimated coefficients are not formal tests

of the rational expectations model. Rather, they are indications of the

"fit" of the model. If we are given data on a consol yield r(t,cx) and the

one period rate r(t,l), then a likelihood ratio test of all restrictions of

the model (except for a restriction implied by the stationarity of

amounts to nothing more than a regression of the excess return h(t,t+1,) -

r(t,t+l) — (r(t,') - yr (t+l,))/(l-y) - r (t,t+l) on information at time

t. (Shiller (l98la)), Campbell and Shiller (1986a)).

Note that such tests may not have much power to determine whether long

rates are too volatile to accord with market efficiency. Suppose, for

example, that the short rate r(t,t+l) is a first-order autoregressive

process as above, and suppose that the long rate overreacts to the short

rate, r(t,c)-. (p44) + b(r (t,l)-p) where b > (l-y)/(l-A). Then the excess

49

holding return h (t,t+l,u)-r (t,t+l), defined as (r(t) - -y r(t+l,cc))/l-

(c-l)r (t,t+l) —1 - r(t,t+l), is equal (up to a constant) to p -

r(tt+2) where c b/(l-y). If -y is close to one and c large, then this

excess return is approximately proportional to b, and changing b would do

little more than scale it up or down. If the excess return is not very

forecastable for one b, it is likely also to be not very forecastable for

another b. Then, a regression of excess holding returns on the short rate

may have little power to detect even major departures of b from (1-)/(1-

Tests of the rational expectations model are not so straightforward

when using data on a single long rate that is not a consol yield and a short

rate. Sargent (1979) showed how, using a companion-form vector autore-

gression, it is readily possible to test the restrictions implied by the

rational expectations model even with such data. He was unable to reject

these restrictions on the vector autoregression of long and short rates

using a likelihood ratio test. However, it was later discovered that

Sargent's paper did not test all restrictions, and when the additional

restrictions were incorporated into the analysis, the hypothesis was

rejected (Hansen and Sargent (1981), Shiller (l98la)). These rejections,

however, do not deny the similarity between actual and optimal distributed

lag coefficients. Campbell and Shiller (l986a) used a cointegrated vector

autoregressive framework, where the vector contained two elements, the long

rate and the short rate, and confirmed both that the rational expectations

model is rejected with a Wald test and that the model is of some value in

describing how long rates respond to short rates and their own lagged

values.

There is also some evidence that the relation of long rates to lagged

50

interest rates changes approximately appropriately when the stochastic

properties of interest rates change. It was shown by Shiller (1985) that

such a correspondence between the distribution lag coefficients holds up

crudely speaking even when one uses 19th century U.S. data, or 19th or 20th

century British data. In the 19th century in Britain, for example, short

rates appeared to be sharply mean reverting, so that long rates should have

been nearly constant: indeed the distributed lag regressions of the British

consol yield on the short rates showed sharply reduced coefficients relative

to the 20th century coefficients in a distributed lag regression of long

rates on short rates. Mankiw, Miron, and Weil (1986) found an abrupt, and

they interpreted appropriate, given the rational expectations model, change

in the distributed lag coefficients, at the time of the founding of the

Federal Reserve.

How is it then that the forward-spot spread f(t,t+m, t+m+n) - r(t,t+m)

seems to predict well only for large n and not small n? Faina and Bliss

(1986) interpreted this finding as reflecting the fact that interest rates

are not very forecastable into the near future, but better forecastable into

the more distant future. He gave as an example the story of AR-l Model

described in connection with equations 31 - 33 above. The expectation as of

time t or the change r(t+n,t+n+l) - r(t,t+l) is (ynl)(r(t,t+l)i). For -y

close to, but below, one, the variance of the expected change is quite small

for small n, and grows with n. Thus, for small n any noise in the term

premium might swamp out the component in the forward spot spread

F(t,t+n,t+m+n) - r(t,t+m) that is due to predictable change in interest

rates.

V.5. Interpreting Departures from the Expectations Theory

51

Of course, as a matter of tautology, the fact that the coefficients in

Table 2 do not all equal one has something to do with time-varying term

premia. But the nature of the time varying term premia has not been given

an ample description for all n and in.

One story for the negative coefficients in Table 2 for large in (� 20

years) and small n (� one year) is that there might be noise in term premia

on long-term interest rates unrelated to short-term interest rates. The

noise might be due to exogenous shift to investor demand, or even to

changing fashions and fads in investing. Suppose, for example, that this

"noise" is serially uncorrelated, as though it were due to an error in

measuring long-term interest rates.52 Consider for simplicity consols, m,

for which f(t,t+n,) - r(t,) — (D(n)/(D() - D(n))(r(t,) - r(t,n)).If one regresses r(t+n,') - r(t,) on this then one has r(t,) on both

sides of the equation with opposite signs. Thus, any "noise" in r(tco)

might give a negative slope coefficient in the regression.

This simple story about extraneous noise like measurement error in long

rates, while suggestive, is not completely adequate in explaining the wrong

signs in the Table 2 regressions for large in the and small n. If the

problem were just exogenous noise in long rates then an instrumental

variables approach to the estimation of the above regressions with economic

variables as instruments would correct the wrong sign; yet it does not

52Just as well, the wrong signs in some regressions could be due tomeasurement error in interest rates, a point considered and rejected as themain explanation for the wrong signs by Shiller (1979) and Mankiw (1986).However, measurement error is taken more seriously by Stambaugh (1986) andBrown and Dybvig (1986). The latter needed measurement error to study theone-factor version of the Cox-Ingersoll-Ross model because without it therewould be a perfect dependence among the interest rates of different

maturity.

52

(Mankiw (1986).

A different story for the wrong sign in the regression is that long

rates do not react properly to short rates. The distributed lag regressions

noted above of long rates on short rates, while similar to the distributed

lag implied by an autoregressive forecasting regression for short rates, are

not quite the same. In fact, the distributed lag coefficient of long rates

on short rates tend to show too simple a pattern, like a simple exponential

decay pattern instead of a relatively choppy pattern seen in the optimal

responses of long rates to short rates implied by the forecasting equation

(Shiller (1985)). This result might come about because people who price

long bonds tend to blur the past somewhat in their memories, or because

people use a simple "conventional" pricing rule for long bonds.53

V.6 Seasonality and Interest Rates

The above discussion suggests that the expectations hypothesis works

best when interest rate movements are well forecastable. With many economic

variables seasonal movements are forecastable far into the future. If there

is any seasonality in interest rates, one would expect to see a seasonal

pattern to the term structure. We would not expect that long rates and short

rates should show the same seasonal pattern, that is, reach their highest

point in the same month. Instead, the expectations theory would predict a

phase shift between long and short rates. Macaulay (1938) investigated

whether this occurred using data on call money and time rates 1890 to 1913,

Keynes (1936) said that the long rate is highly conventional . .

its actual value is largely governed by the prevailing view as to what itsvalue is expected to be." The idea here is apparently that a simple rule ofthumb used to price long-term bonds may become validated when market pricesappear to follow the rule.

53

and concluded that there was "evidence of definite and relatively successful

forecasting,"54 for seasonal movements, though not for movements other than

seasonals.

Sargent (1971) noted that the maturity on the call rates was not well-

defined, and in fact the actual maturities of the call loans are likely to

have had a seasonality themselves. He thus sought to reproduce Macaulay's

work using more recent data for which maturity can be defined more

precisely. Sargent showed that in a perfect foresight model the simple

expectations theory for discount bonds implies that the rn-period rate

rd(t,t+m) should lead the 1-period rate rd(t,t+l) by (m-l)/2 periods across

all frequencies. He used U. S. Treasury Bill rates on one to thirteen week

bills for 1953 to 1960. He found that long rates did tend to lead short

rates at the seasonal frequencies, but by much less than the theoretical (m-

l)/2.

The post World War II data set that Sargent used, however, contained a

much milder seasonal than was evident in the prewar data that Macaulay had

used. The Federal Reserve was founded in 1913 to "provide an elastic

currency" and this clearly meant that one of their missions was to eliminate

seasonals, which they then largely did (See Shiller (1980) and Miron (1984),

(l986)). Mankiw and Miron (1986) found a time series on pre-1913 U. S.

interest rates for which maturity was better defined, and found more

encouraging results for the expectations theory.

54F. Macaulay, (1938), p. 36.

55Clark (1986) questioned whether the decline in seasonality was due tothe founding of the Fed. He noted that seasonals disappeared in the UnitedStates and other countries at about the same time, and that seasonalsdisappeared approximately three years before the seasonals in currency andhigh powered money changed.

54

V.7 The Sign of Term Premia

Kim (1986) investigated whether the observed terni premium between

nominal 3 and 6-month treasury bills in the United States 1959-1986 could be

reconciled with the covariance between S(t,t+3) and S(t+3,t+6) as described

by equation 22 or 24 as the theory prescribes. He used a co-integrated

vector autoregressive model for the two log interest rates, log consumption

and a log price index, and a lognormality assumption for the error term. He

transformed the vector with the cointegrating vector so that the transformed

vector has as elements the spread between the two log interest rates, the

change in one of the interest rates, the change in log real consumption and

the change in the log price index. For the model, the covariance in

equation 22 is constant through time. He tested the restrictions across the

mean vector, coefficient vector and variance matrix of residuals using a

Wald test. The test rejected the restrictions; on the other hand, the sign

of the term premium is as predicted by the sign of the covariance.

Other studies of consumption and the term structure of interest rates

looked at short-term real returns on long and short bonds and their correla-

tion with real consumption changes, to see if the difference in mean real

returns between long and short bonds could be reconciled with the covariance

of real returns with real marginal rates of substitution. Grossman, Melino

and Shiller (1985) found that the excess real one-period returns between

long-term debt and short-term debt had negligible correlation with real per

capita consumption changes with annual U. S. data 1890-1981 and with U. S.

quarterly data 1953-83. They rejected at high significance levels the

covariance restrictions using a vector autoregression model including real

55

returns on long-term debt, short-term debt, and corporate stocks.56

V.8. Modelling Time-Varying Term Premip

Since the rational expectations hypothesis can be rejected, as dis-

cussed above, it follows that the term premium is time-varying. Although the

term premium is not observed itself without error, we can study its projec-

tion onto any information set by regressing the variables represented on the

right-hand-sides of the expressions defining term premia, (14), (15), and

(16) above, from which the expectations operators have been deleted, onto

information available at time t. The above discussion of the projection onto

the forward-spot spread concerns only one possible such regression.

There is no theory of the term structure well-developed enough to allow us

to predict what variables to use, so the empirical literature here often

looks like a "fishing expedition."

Kessel (1965) regressed the forward-spot spread fd(t,t+1,t+2)

rd(t+l,t+2) Ofl rd(t,t+l) where the time unit is four weeks to test whether

term premia are related to the level of interest rates. He found using

monthly U. S. treasury bill data 1949-61 that there was a positive coeffi-

cient on rd(t,t+l). However, Nelson (l972b) using analogous methodology

found the opposite sign for the coefficient of the interest rate. Both

Kessel and Nelson gave theories why risk considerations should imply the

sign they got. Shiller (1979) in effect regressed f(t,t+1,t+in+1) -

r(t+1,t+m-4-l) on r(t,t+m) for m very large with quarterly, monthly and

annual time periods for U. S. and U. K. history and found a consistently

56 . . . . . .Mankiw (1986) inquired whether the time variation in the covariancecould be reconciled with time variation in the spread between long arid shortrates in the United States, Canada, United Kingdom and Germany 1961-84. Heconcluded that it could not.

56

positive coefficient, which was interpreted as a sign of possible excess

volatility of long-rates. Campbell and Shiller (1984) in effect found a

negative slope coefficient in a regression (in effect) of fd(t,t+l,t+24l) -

rd(t+l,t+24l) on rd(t,t+l) where time is measured in months, and interpreted

this result as reflecting a possible underreaction of long rates to short

rates. It is difficult to produce a useful summary of these conflicting

results.

Other researchers have used some measure of the variability of interest

rates in such regressions. Modigliarii and Shiller (1973) and Shiller

Campbell and Schoenholtz (1983) used a moving standard deviation of interest

rates. Fama (1976), Mishkin (1982) and Jones and Roley (1983) and Bodie,

Kane and MacDonald (1984) used other measures of the variability in interest

rates. Such measures were often statistically significant. Engle, Lilien and

Robins (1987) used an ARCH model to model time-varying variance of interest

rates, and concluded that the risk premium so modelled helps to explain the

failures of the expectations theory.

Still other variables have been used to explain time-varying term

premia. Nelson (l972b) used an index of business confidence. Shiller

Campbell and Schoenholtz (1983) used a measure of the volume of trade in

bonds. Keini and Stambaugh (1986) used a low-grade yield spread variable (the

difference between yields on long-term under-BAA-rated corporate bonds and

short-term treasury bills), and a small-firm variable (the log of the share

price, averaged equally across the quintile of smallest market value on the

New York Stock Exchange). Campbell (1987) used a latent variable model of

the returns on bills, bonds and common stocks to infer time-varying risk

premia in all three markets.

57

V.9. Flow of Funds Models

Clearly term premia do vary and are correlated with observable

economic variables. But what kind of structural model might clarify why they

vary? One might expect that when the federal government issues a large

amount of long term debt, the supply of long-term debt should rise and other

things equal term premia should rise. One might also expect that in time

when funds flow into life insurance companies, major purchasers of long-term

bonds, then the demand for long term debt should rise and, other things

equal, term premia should decline. Thus, the term structure might be related

to such flows of funds.57

There was a flurry of research on the effects of government debt policy

on the term structure following the policy, brought in by the Kennedy

Administration in the U. S. in 1961, known as "Operation Twist". Operation

Twist consisted of Federal Reserve open market operations and Treasury debt

management operations directed toward shortening the average term to

maturity of outstanding public debt, with the intention of 'twisting' the

term structure.58 Okun (1963) and Scott (1965) correlated the term structure

with federal debt measures without accounting for expectations. Modigliani

and Sutch (1966), (1967) added dummy or debt composition variables to their

distributed lag regressions of long rates on short rates, but found evidence

D7Co:iversely when the government attempts to peg the term structure,there should be consequential flows of demand across maturities. Walker(1954) noted that when the Federal Reserve attempted to peg an upwardsloping term structure there was a great shift out of short-term securitiesinto long-term securities by the holders of government debt. Such a shift is

implied by the expectations hypothesis.

58Operation twist also involved relaxing some interest rate ceilings.The federal debt structure during the early 1960's in fact went in exactlythe opposite direction to what was implied by operation twist, as theTreasury's debt policy was contradicting the Fed's. See B. Friedman, (1981).

58

of only a "weak" effect of national debt on the term structure. Indeed the

simple distributed lag on short rates explained long rates so well, that

there was little room for much improvement of fit using debt policy varia-

59 . . . .bles. The Modigliani-Sutch conclusions were criticized by Wallace (1965)

for the assumption that government debt policy is exogenous over the sample

period.

There is a substantial literature on models that relate interest rates

to such flows of funds, see for example Ando and Modigliani (1975), Brainard

and Tobin (1968), DeLeeuw (1965), Friedman (1977), (1980a), Hendershott

(1971), or Backus, Brainard, Smith and Tobin (1980). But much of this

literature makes no explicit use of expectations of future interest rates

that ought to play a pivotal role in the term structure of interest rates.

Many of the models are not complete, e. g. providing estimates of some

demands for funds, and not providing a general equilibrium that might give a

theory of the term structure.

Friedman and Roley (1979) and Roley (1982) estimated a flow of funds

model (along lines of Friedman (1977), (l980b), and Roley (1977)) but

incorporating as determinants of the demand functions not yields to maturity

but rational expectations of short-run returns.

Flow of funds modelling has offered the promise of estimating

consistently general equilibrium models of the determination of interest

rates, but such modelling has to date been hampered by the same problems

that have prevented any consensus on other inacroeconometric models. A lot of

subjective judgment goes into specifying the identifying restrictions,

59B. Friedman (1977), (1981) did find a significant coefficient in aterm structure equation for a variable which was the ratio of outstandingfederal long-term securities to outstanding federal short-term securities.

59

exogeneity specifications and other assumptions that lie behind a compli-

cated simultaneous equation model. Hence, there is a lot of uncertainty

about the validity of particular models.

VI. Some Concluding Observations

There has been a lot of progress in our understanding of the term

structure in the last twenty years. We now have formal heuristic theoretical

models of the term structure in terms of the ultimate objectives of economic

agents and the stochastic properties of forcing variables. These models are

beginnings that have changed our way of thinking about the term structure.

We have now an extensive empirical literature describing in great detail how

the term structure is correlated with other economic variables. But we could

hope for still more progress.

It is of course very difficult to say where the actual opportunities

for productive research lie, but it is possible to say where there are

problems to be solved.

Theoretical work on the term structure, while it has offered many

insights, still does not allow us to say much about the term structure we

observe. Most theorists are currently using a representative individual

utility of consumption model, while most corporate and government bonds in

the United States are held by institutions. Even if institutions were

somehow behaving as if they were representative consumers, we must face the

fact that the expected present value of utility of consumption model has not

held up well in tests of the returns on assets other than bonds. Probably,

the theoretical model is just not a good descriptor of human behavior.

Most of the theoretical work on the expectations hypothesis has worked

on the term structure of index bonds, but freely tradable true index bonds

60

of varying maturity are virtually nonexistent. The theoretical literature

has tried to find justifications for a zero term premia model, while the

assumption of zero term premia has never been an issue for empirical

researchers. That term premia are not zero and change through time has not

suggested any well-posed problems for theoretical researchers working in the

current paradigm that would produce any idea as to how to expect them to

change.

Empirical work on the term structure has produced consensus on little

more than that the rational expectations model, while perhaps containing an

element of truth, can be rejected. There is no consensus on why terni premia

vary. There does not seem even to be agreement on how to describe the

correlation of the term premia with other variables. A lot more research

could be done leading to consensus on, for example, the senses in which long

rates may be influenced by government fiscal policy, term premia are related

to some measures of risk, interest rates overreact or underreact to short

rates, or be influenced by or depend on rules of thumb or "satisficing"

behavior. Flow of funds models have some interest, but seem to have been

largely dropped by researchers in the wake of the rational expectations

revolution, just when they should have been integrated with it.

61

Figure 2. The term structure of interest rates r (t,t+n) (solid line) andthe instantaneous forward rate f(t,t+n,t+n) (dased line) for the end ofAugust, 1978.Source of data: Tables A-i and A-2 in Appendix by J. Huston McCulloch

Figure 3. Instantaneous Forward Rates. Data plotted are f(t-n,t,t) againsttime t and horizon n. The annual data series f(t,t,t) — rt seen at the farright of the surface is the instantaneous interest rate for the end of Juneof each year, 1957 to 1985. Curves on surface parallel to n axis show paththrough time of the 'forecast' implicit in the term structure of theinstantaneous forward rate applying to the date shown on the t axis. Ifthere were perfect foresight, we would expect these curves to be horizontalstraigbt lines. In the expectations theory of the term structure with zerorisk premiuni, they should be random walks. Curves on the surface parallel tthe t axis show the path through time of a forward rate of fixed forecasthorizon. Data plotted here are from Table A-2 of the Appendix by J. FiustonMcCul loch

Fig. 4. The long-short spread StA() rdtt+n) - rd(tt+l), solid line.and the perfet-foresight spread n) — rA(t,t+n) - rd(t,t+n), dashedline, where rA(t,t+n) is the perfect foreiht n-period rate defined byexpression (2), n — 40 quarters. Thus, rd(t,t+n) — (E(r=t,t+39)r(r,i-+l))/40. Data plotted are quarterly series for end of the firs:month of each quarter using McCulloch's three-month and ten-year discountbond yield series, Table A-i, Appendix. S,(n) is plotted for 1947 firstquarter to 1975 third quarter at annual rates.

62

1

63

yeOt

9 te

It5p0 0te

7

mct'2

0 p.

0 0 40

:3- 0 N 0 :3

:3

0 :3

40

0•

'0

'0 0

0

perc

ent

P..)

P

b.)

(4)

(4)

• '

0 0

—

0 0

0 0

-o

0 1

I I

I I

I

__)

____

_,

4.

C

—

'0

4,

- -

P..,

- .4...

1.

.1

0'

Lfl

4ç

r

0 P

...)

3'

—

4:

t U

, )_

- - -

-—:

:a-

- - -

-

- -

—-

'0

P.,)

'0

'.4

'.4

Table 1.

Formulas for Computation of Forward Rates and Holding Rates

I. Time t forward rate applying to interval from t' to Tt � t' � T

D.(Tt)r(tT) - D.(t'-t)r.(t,t')(1) f.(t,t',T) —

1D.(T-t) - D.(t'-t)

1

II. Holding period rate or return from t to t' on bond maturing timeT, t � t' � T

D.(Tt)r(tT) - [D.(T-t) - D.(t'-t)Jr.(t',T)(2) h.(t,t',T) —

1

1(t'-t)

1 1

III. Holding period rate of return from t to t' rolling over bonds ofterm m — T-t , t � T � t'

s-i(3) h.(t,t',T) — ( Z (D.(km+m) - D.(km))r.(t+km,t+km-+-m)1 k—O 1 1 1

+ [D.(t"-t) - D.(sm)]h.(t+sm,t',t+sm+m))/D.(t'-t)

where s = largest integer � (t'-t)/m

Note: In above formulas, substitute i — d for discount bonds, i — p forpar bonds. Par bond formulas give linear approximation to true rates.Duration (from which the second argument, t , has been dropped here) is

-Rm

given by Dd(m) — , D(m) = (1 - e )/R , where R is the point of

linearization, which might be taken as r(t,T) . These formulas may be

applied to data in Tables A-i and A-3.66

Table 2.

Regressions of Changes in rn-period Interest Rates onChanges Predicted by the Term Structure

Study

r.(t+n,t+m+n)-r.(t,t±m) onf.(t,t+n,t+m+n)-r.(t,+m) and constant

Country Sample n slope Std.

(years) (years) Coef. Err

Shiller U. S. 1966-77 >20.0 0.25 -5.56 1.67 .201(1979)a

U. S. 1919-58 >20.0 1.00 -0.44 0.75 0.01

U. K. 1956-77 0.25 -5.88 2.09 0.09

Shiller, U. S. 1959-74 0.25 0.25 0.27 0.18 0.03CampbellSchoenholtz U. S. 1959-73 30.0 0.50 -1.46 (1.79) 0.02(1983)b

Mankiw Canada 1961-84 0.25 0.25

(1986)cW.Geruiany 1961-84 0.25 0.25

Fama U. S. 1959-82 1/12 1/12(1984)d

1/12 2/12

1/12 3/12

1/12 4/12

1/12 5/12

Fania and Bliss U. S. 1964-84 1.00 1.00

(l986)e1.00 2.00

1.00 3.00

1.00 4.00

Shiller U. S. 1953-86 0.25 rollover*

(1986)

Note: Expectations theory of term structure asserts slope coefficient should be1.00. Not all regressions summarized here were in exactly the form shown here:some cases a linearization was assumed to transform results to the form ShOwn

in

0.10

0.14

0.46

0.25

0.26

0.17

0.11

0.09

0.69

1.30

(0.07)

(0.07)

(0.07)

(0.10)

(0.12)

(0.10)

(0.10)

(0.28)

(0.26)

(0.10)

(0.34)

(0.17)

0.02

0.03

0.13

0.02

0.02

0.01

0.00

0.00

0.08

0.24

0.48

0.090

1.61

0.61

here. Significance level refers to tes of hypothesis that coefficient is 1.00.*Dependent variable is approximately S (m,n) and independent variable Is S(ni,n) asdefined in expression 26 in text.

a. Page 1210, Table 3, rows 1, 4 and 5,. Column 2 coefficient was convertedusing duration implicit in y given in Table 1 rows 1, 4 and 5 column 1.

b. Page. 192, Table 3, rows 4 and 10, columns 5 and 6.

c. Page 81, Table 9, rows 2 and 4 columns 3 and 4.

d. Page 517, Table 4, rows 6-10, columns 1-2.

e. Page 23, Table 3, rows 1-4, columns 3-4.

f. Page 103.

68

APPENDIX I

Mathematical Symbols

D(m,t) - Duration of an rn-period bond at time t . Second argument will

sometimes be omitted.

fd(t,t' ,T) - The forward discount interest rate at time t applying to the

interval from t' to T , t t' � T . The term of the forward instrument

is niT-t'

F(t,t',T) - The forward par interest rate at time t applying to the interval

from t to T , t � t' s T . The term of the forward instrument is m = T-t'

f(t,t' ,T) - Linear approximation to F(t,t' T)

hd(t,t',T) - The discount holding period return. If t � t' � T it is the return

from buying a discount bond at time t that matures at time T and selling it at

time t '. If t � T � t' , it is the rate of return from rolling over discount

bonds of maturity in T-t , until time t'

H(ttT) - The par holding period return.If t s t' T , it is the return from

buying a par bond at time t that matures at time T , receiving coupons between

t and t' and selling it at time t' . If t � T t' , it is the rate of return

from rolling over par bonds of maturity m — T-t . Until time t'h(t,t',T) - Linear approximation to H(t,t',T)

In - the term of a bond, equal to the time to maturity T-t

p(t,T) - The price at time t of a bond that matures at time T , whose princi-

pal is 1.

Pd(t,T) - The price at time t of a discount bond that matures at time T , whose

principal is 1.

69

r(t,T) - The interest rate (yield to maturity) at date t on a bond that natures

at date T

r(t,T,h) - The interest rate (yield to maturity), compounded every h periods, at

date t on a bond that matures at date T

rd(t,T) - The interest rate (yield to maturity) at date t on a discount bond

that matures at date T

r(t,T) - The interest rate (yield to maturity) at date t on a par bond that

matures at date T

Si - The amount of the th payment made on a bond, made at date t. . On a

coupon bond s. — c , i < T ,ST

— l+c

t - The date of the 1tF payment on a bond.

T - The date on which a bond matures.

f,i(t,t,T) - forward term premium, equal to f(t,t',T) - Er.(t',T) , t < t'

< T , i — p, d.

h,i(t,t.T) - Holding period term premium, equal to Eh(t,t',T) - r.(t,t') ,t

<t <T , i—p, d

- Rollover term premium, equal to r.(t,t') - Eth.(t,t',t±m) , t <

t' , t<t+m<t' , i—p,d.w - The number of payments promised by a bond when it was issued.

70

References

Amsler, Christine (1984), "A 'Pure' Long-Term Interest Rate and the Demand forMoney," Journal of Economics and Business, 36: 359-70.

Ando, Albert, and Arthur Kennickell, "A Reappraisal of the Phillips Curve and theTerm Structure of Interest Rates," xeroxed, University of Pennsylvania, 1983.

Ando, Albert, and Franco Modigliani, (1975) "Some Reflections on DescribingStructures in Financial Sectors," in Gary Fromin and Lawrence Klein, editors, TheBrookings Model: Perspectives and Recent Developments, North Holland, Amsterdam.

Backus, David, William C. Brainard, Gary Smith, and James Tobin (1980), "AModel of U. S. Financial and Nonfinancial Economic Behavior," JournalMoney. Credit and Banking, 12:259-93.

Begg, David K. H. (1984), "Rational Expectations and Bond Pricing: Modelling theTerm Structure with and without Certainty Equivalence," Economic Journal, 94:45-58.

Benninga, Simon and Aris Protopapadakis (1983), "Real and Nominal Interest Ratesunder Uncertainty: A The Fisher Theorem and the Term Structure," Journal ofPolitical Economy 91:856-67.

Bierwag, G. 0., and M. A. Grove (1967), "A Model of the Term Structure of InterestRates," Review of Economics and Statistics, 49:50-62.

Bohm-Bawerk Eugen, V., The Positive Theory of Capital, G.E. Stechert & Co., 1891.

Board of Governors of the Federal Reserve System, Flow of Funds Accounts FinancialAssets and Liabilities Year-End, 1961-84, Washington, 1985.

Bodie, Zvi, Alex Kane and Robert McDonald (1984), "Why Haven't Nominal RatesDeclined?" Financial Analysts Journal, 40: 16-27.

Brainard, William C., and James Tobin (1968), "Pitfalls in Financial ModelBuilding," American Economic Review Papers and Proceedings, 58:99-122, May.

Brealey, Richard, and Stephen Schaefer (1977), "Term Structure with UncertainInflation," Journal of Finance, 32: 277-89.

Brennan, Michael J., and Eduardo S. Schwartz (1980), "Conditional Predictions ofBond Prices and Returns," Journal of Finance, 35: 405-17.

Brown, Stephen, and Philip Dybvig (1986), "The Empirical Implications of the Cox,Ingersoll, Ross Theory of the Term Structure of Interest Rates." Journal ofFinance, 41:616-30.

Buse, Adolf, (1967), "Interest Rates, the Meiselman Model and Random Numbers,"

71

Journal of Political Economy, 75: 49-62.

Campbell, John Y. (1984a), "Asset Duration and Time-Varying Risk Premia,' unpub-lished Ph.D. dissertation, Yale University.

_____________ (1986) "Bond and Stock Returns in a Simple Exchange Model,"Quarterly Journal of Economics, 786-803.

_____________ (1986), "A Defense of Traditional Hypotheses about the TermStructure of Interest Rates," Journal of Finance, 41:183-93.

_____________ (1987), "Stock Returns and the Term Structure," Journal ofFinancial Economics, in press.

Campbell, John Y. and Robert J. Shiller (1984), "A Simple Account of the Behaviorof Long-Term Interest Rates," Anerican Economic Review Papers and Proceedings, 74:44-8.

______________ (1986) "Cointegration and Tests of Present Value Models,' NBER

Working Paper No. 1885. (a)

_______________ (1986) "The Dividend Price Ratio and Expectations of FutureDividends and Discount Factors," NBER Working Paper. (b)

Cargill, Thomas F., and Robert A. Meyer (1972), "A Spectral Approach to Estimatingthe Distributed Lag Relationship Between Long-Term and Short-Term Interest Rates,'International Economic Review 13:223-38 June.

Chambers, Donald R., Willard T. Carlton, and Donald W. Waldman (198-.), "A NewApproach to the Estimation of the Term Structure of Interest Rates,' Journal ofFinancial and Quantitative Analysis, 19: 233-52.

Chen, Eva T. (1986), "Estimation of the Term Structure of Interest Rates via CubicExponential Spline Functions," unpublished doctoral dissertation draft, The OhioState University.

Clark, 3. B., (1895) "The Gold Standard of Currency in Light of Recent Theory."Political Science Quarterly, 10:389-403.

Conard, Joseph W. An Introduction to the Theory of Interest, University ofCalifornia Press, Berkeley and Los Angeles, 1959.

Cox, John C., Jonathan E. Ingersoll Jr. and Stephen A. Ross (1985a). "An Inter-temporal General Equilibrium Model of Asset Prices," Econometrica, 53: 33-84.

______________ (1985b), "A Theory of the Term Structure of Interest Rates,"Econometrica, 53: 385-408.

________________ (1981), "A Reexamination of Traditional Hypotheses about the TerrLStructure of Interest Rates," Journal of Finance, 36: 769-99.

Culbertson, J. M. (1957), "The Term Structure of Interest Rates," 0-_arte:y

72

Journal of Economics, 71:485-517.

De Leeuw, Frank (1965), "A Model of Financial Behavior," in J. Duesenberry et al.,Editors, The Brookings Quarterly Economic Model of the United States, Rand

McNally, Chicago, pp. 465-530.

Diller, S. (1969) "Expectations and the Term Structure of Interest Rates," in J.Mincer, editor, Economic Forecasts and Expectations, National Bureau of EconomicResearch, New York.

Dobson, Steven W. (1978), "Estimating Term Structure Equations with IndividualBond Data," Journal of Finance, 33: 75-92.

Dobson, Steven W., Richard C. Sutch and David E. Vanderford (1976), "An Eva1uaionof Alternative Empirical Models of the Term Structure of Interest Rates," Journalof Finance, 31: 1035-65.

Dothan, L. Un (1978), "On the Term Structure of Interest Rates," Journal ofFinancial Economics, 6: 59-69.

Durand, David, Basic Yields of Corporate Bonds. 1900-1942, Technical PaperNo. 3, NBER, 1942.

Dybvig, Philip H., Jonathan E. Ingersoll Jr., and Stephen A. Ross, "Long Forward

Rates Can Never Fall," xeroxed, 1986.

Dunn, Kenneth B., and Kenneth J. Singleton (1984), "Modelling the TermStructure of Interest Rates under Nonseparable Utility and Durability of

Goods," NBER Working Paper No. 1415.

Echols, Michael E., and Jan Walter Elliott (1976), "A Quantitative Yield CurveModel for Estimating the Term Structure of Interest Rates," Journal of Financialand Quantitative Analysis, 11: 87-114.

Engle, Robert F., David M. Lilien, and Russell P. Robins (1987), "Estimating TifleVarying Risk Premia in the Term Structure: The ARCH-M Model," Econometrica,55:391-407.

Fania, Eugene F. (1984a), "The Information in the Term Structure," Journal ofFinancial Economics, 13: 509-28.

Fama, Eugene F., (l984b), "Term Premiums in Bond Returns," Journal ofFinancial Economics, 13: 529-46.

Fama, Eugene F., (1985) "Term Premiums and Default Premiums in Money Markets,Journal of Financial Economics.

Fania, Eugene F. and Robert R. Bliss, "The Information in Long-Maturity ForwardRates," Graduate School of Business, University of Chicago, October, 1986.

Financial Publishing Company (1970), Expanded Bond Values Tables, London:Routledge and Kegan Paul, Ltd.

73

Fisher, Irving (1896) "Appreciation and Interest," Publications of theAmerican Economic Association, pp. 23-9 and 88-92.

_____________ (1907), The Rate of Interest, Its Nature, Determination andRelation to Economic Phenomena, Macmillan, New York.

Fisher, Irving (1930) Theory of Interest, New York, Macmillan.

Fisher, Lawrence (1966), "An Algorithm for Finding Exact Rates of Return," Journalof Business, 39:111-18, January.

Flavin, Marjorie (1983), "Excess Volatility in the Financial Markets: A Reassess-ment of the Empirical Evidence," Journal of Political Economy, 91: 929-56.

______________ (l984a), "Excess Sensitivity of Consumption to Current Income:

Liquidity Constraints or Myopia?" NBER Working Paper #1341.

________________ (l984b) "Time Series Evidence on the Expectations Hypothesis ofthe Term Structure," Carnegie-Rochester Conference Series on Public Policy, 20:211-38.

Friedman, Benjamin M. (1981), "Debt Management Policy, Interest Rates and EconomicActivity," National Bureau of Economic Research Working Paper.

_____________ (l980a), "The Determination of Long-Term Interest Rates: Implica-tions for Fiscal and Monetary Policies," Journal of Money. Credit and

Banking, l2(Part 2):331-52.

_____________ (l980b), "The Effect of Shifting Wealth Ownership on the TermStructure of Interest Rates: the Case of Pensions," Quarterly Journal ofmics, 94:567-90.

_____________ (l977a), "Financial Flow Variables and the Short-Run Determination of Long-Run Interest Rates," Journal of Political Economy, 85: 661-89.

_____________ (l977b), "The Inefficiency of Short-Run Monetary Targets for

Monetary Policy," Brookings Papers on Economic Activity, 2: 293-335.

_____________ (1979), "Interest Rate Expectations Versus Forward Rates: Evidencefrom an Expectations Survey," Journal of Finance, 34: 965-73.

________________ (l980c) "Survey Evidence on the Rationality of Interest RateExpectations," Journal of Monetary Economics, 6:453-65.

Friedman, Benjamin M. , and V. Vance Roley (1979), "Investors PortfolioBehavior Under Alternative Models of Long-Term Interest Rate Expectations:Unitary, Rational or Autoregressive," Econometrica 47:1475-97.

Froot, Kenneth A. (1987), "New Hope for the Expectations Hypothesis of the TermStructure of Interest Rates," reproduced, Sloan School of Management, M. I. T.

74

Grossman, Sanford J., Angelo Melino, and Robert J. Shiller (1985), "Estimating theContinuous Time Consumption Based Asset Pricing Model," NBER Working Paper No.1643.

Hall, Robert E., ""Stochastic Implications of the Life Cycle-Permanent IncomeHypothesis," Journal of Political Economy, 6:971-88, December 1978.

Hamburger, Michael J., and Elliott N. Platt, (1975) "The ExpectationsHypothesis and the Efficiency of the Treasury Bill Market," Review ofEconomics and Statistics, 57:190-99.

Hansen, Lars Peter, and Thomas J. Sargent (1981), "Exact Linear Rational Expecta-tions Models: Specification and Estimation," Staff Report, Federal Reserve Bankof Minneapolis.

Hansen, Lars Peter, and Kenneth J. Singleton (1983), "Stochastic Consumption,Risk Aversion, and the Temporal Behavior of Asset Returns," Journal ofPolitical Economy, 91:249-65.

Hendershott, Partric H. (1971), "A Flow of Funds Model Estimated for the Non-BankFinance Sector," Journal of Money. Credit and Banking, 3:815-32, November.

Hickman, W. Braddock (1942), "The Term Structure of Interest Rates: an Exploratory

Analysis, NBER, mimeographed. Results shown in Kessel (1965), Appendix A. pp. 103-5.

Hicks, John R. (1946), Value and Capital, 2nd. edition, Oxford University Press.

Homer, Sidney (1963), A History of Interest Rates, Rutgers University Press, NewBrunswick, New Jersey.

Fiopewell, Michael, and George Kaufman (1973), "Bond Price Volatility and Term toMaturity: A Generalized Respecification," American Economic Review, 63: 749-53,

September.

Huizinga, John, and Frederic S. Mishkin (1984), "The Measurement of Ex-Ante RealInterest Rates on Assets with Different Risk Characteristics," unpublished paper,Graduate School of Business, University of Chicago.

Ingersoll, Jonathan E. Jr., Jeffrey Skelton and Roman L. Well (1978), "DurationForty Years Later," Journal of Financial and Quantitative Analysis, 13: 627-50.

Jarrow, Robert A. (1981), "Liquidity Premiums and the Expectations Hypothesis,"Journal of Banking and Finance, 5: 539-46.

Jones, David S. and V. Vance Roley (1983), "Rational Expectations and the Expecta-tions Model of the Term Structure: A Test Using Weekly Data," Journal of MonetaryEconomics, 12: 453-65.

Jordan, James V. (1984), "Tax Effects in Term Structure Estimation," Journal ofFinance, 39: 393-406.

75

Kaldor, Nicholas (1939), "Speculation and Instability," Review of EconomicStudies, 7: 1-27.

Kane, Edward J. (1980), "Market Incompleteness and Divergences Between Forward andFutures Interest Rates," Journal of Finance, 35: 221-34.

______________ (1983), "Nested Tests of Alternative Term Structure Theories,"Review of Economics and Statistics, 65: 115-23.

______________ (1970), "The Term Structure of Interest Rates: An Attempt toReconcile Teaching with Practice," The Journal of Finance, 25:361-74.

Kane, Edward J., and Burton C. Malkiel (1967), "The Terni Structure ofInterest Rates: An Analysis of a Survey of Interest Rate Expectations," Review ofEconomics and Statistics, 49:343-55.

Keini, Donald B., and Robert F. Stambaugh (1986), "Predicting Returns in the Stockand Bond Markets," Journal of Financial Economics, 17:357-90.

Kessel, Reuben A. (1965), The Cyclical Behavior of the Term Structure ofInterest Rates, NEER, New York.

Keynes, John M. (1936), The General Theory of Employment. Interest and Money.

______________ (1930), Treatise on Money.

Kim, Se-Jin (1986), "Explaining the Risk Premium: Nominal Interest Rates, Infla-tion and Consumption," reproduced, Yale University.

Langetieg, Terence C., "A Multivariate Model of the Term Structure," Journal ofFinance, 35: 71-97, 1980.

LeRoy, Stephen F. (1982a), "Expectations Models of Asset Prices: A Survey of

Theory," Journal of Finance, 37: 185-217. (a)

______________ (1984), "Nominal Prices and Interest Rates in General Equilibrium:Endowment Shocks," Journal of Business, 57:197-213.

______________ (1982b), "Risk Aversion and the Term Structure of RealInterest Rates," Economics Letters, 10:355-61. (b)

______________ (1983), "Risk Aversion and the Term Structure of Real InterestRates: A Correction," Economics Letters, 12:339-40.

LeRoy, Stephen F., and Richard D. Porter (1981), "The Present Value Relation:Tests Based on Implied Variance Bounds," Econometrica, 49: 555-74.

Lindahi, Erik, Studies in the Theory of Money and Capital, Rinehart and Company,New York, 1939.

Long, John B. (1974), "Stock Prices, Inflation and the Term Structure ofInterest Rates," Journal of Financial Economics, 1:131-70.

76

Lutz, Frederick A. (1940), "The Structure of Interest Rates," Quarterly Journal of

Economics, 55:36-63.

Macaulay, Frederick R., (1938) Some Theoretical Problems Suggested by thements of Interest Rates. Bond Yields, and Stock Prices in the United States Since1856, New York, National Bureau of Economic Research.

Malkiel, Burton C. (1966), "The Terni Structure of Interest Rates: Expectationsand Behavior Patterns," Princeton University Press.

Mankiw, N. Gregory (1986), "The Term Structure of Interest Rates Revisited,"Brookings Papers on Economic Activity, 61-96.

Mankiw, N. Gregory, and Jeffrey A. Miron (1986), "The Changing Behavior ofthe Term Structure of Interest Rates," QuarterlyJournal of Economics,101:211-28.

Mankiw, N. Gregory, Jeffrey A. Miron, and David N. Weil, "The Adjustment ofExpectations of a Change in Regime: A Study of the Founding of the Federal

American Economic Review, June 1987.

Mankiw, N. Gregory, and Lawrence H. Summers (1984), "Do Long-Term Interest RatesOverreact to Short-Term Interest Rates?," Brookings Papers on Economic Activity,

pp. 223-42.

Marsh, Terry A. (1980), "Equilibrium Term Structure Models: Test Methodology,"Journal of Finance, 35: 421-35.

Marsh, Terry A. , and Eric R. Rosenfeld (1983), "Stochastic Processes for InterestRates and Equilibrium Bond Prices," Journal of Finance, 38: 635-46.

McCallum, John S. (1975), "The Expected Holding Period Return, Uncertaintyand the Term Structure of Interest Rates," Journal of Finance, 30:307- 23.

McCulloch, J. Huston (1971) "Measuring the Term Structure of Interest Rates,"Journal of Business, 44: 19-31.

_____________ (l975a), "An Estimate of the Liquidity Premium," Journal ofPolitical Economy, 83: 95-119.

_____________ (1981), "Interest Rate Risk and Capital Adequacy for TraditionalBanks and Financial Intermediaries," in Sherman J. Maisel, ed., Risk and CapitalAdeQuacy in Commercial Banks, University of Chicago Press and NBER, Chicago, pp.223-48.

_____________ (1975b), "The Tax Adjusted Yield Curve," Journal of Finance, 30:811-30.

______________ (1977), "Cumulative Unanticipated Changes in Interest Rates," NBER

Working Paper No. 222, Cambridge.

77

_____________ (1984), "Term Structure Modeling Using Constrained ExponentialSplines," Reproduced, Ohio State University.

Meiselman, David (1962), The Term Structure of Interest Rates, Prentice Hall,

Englewood Cliffs, New Jersey.

Melino, Angelo (1983)), "Estimation of a Rational Expectations Model of theTerm Structure," in Essays on Estimation and Inference in Linear Rational

Expectations Models, unpublished Ph.D. Dissertation, Harvard University.

______________ (1986), "The Term Structure of Interest Rates: Evidence and

Theory," NBER Working Paper No. 1828.

Michaelsen, Jacob 8. (1965), "The Term Structure of Interest Rates and HoldingPeriod Yields on Government Securities," Journal of Finance, 20:444-63.

Miron, Jeffrey A. (1984), The Economics of Seasonal Time Series, Ph.D. disserta-tion, M. 1. T.

_____________ (1986), "Financial Panics, the Seasonality of the Nominal InterestRate, and the Founding of the Fed," American Economic Review, 76:125-40.

Mishkin, Frederic S. (1978), "Efficient Markets Theory: Implications for

Monetary Policy," Brookings Papers on Economic Activity, pp. 707-52.

_____________ (1980), "Is the Preferred Habitat Model of the Term StructureInconsistent with Financial Market Efficiency?" Journal of Political Economy,88:406-11.

______________ (1982), "Monetary Policy and Short-Term Interest Rates: AnEfficient Markets - Rational Expectations Approach," Journal of Monetarymics, 37:63-72.

Modigliani, Franco, and Robert J. Shiller (1973), "Inflation, Rational Expecta-tions and the Term Structure of Interest Rates," Economica, 40:12-43, February.

Modigliani, Franco and Richard Sutch (1967), "Debt Management and the TermStructure of Interest Rates: An Analysis of Recent Experience," Journal ofPolitical Economy, 75:569-89.

______________ (1966), "Innovations in Interest Rate Policy," AmericanEconomic Review, 56: 178-97.

Nelson, Charles R. (l970a), "A Critique of some Recent Empirical Research inthe Explanation of the Term Structure of Interest Rates," Journal of PoliticalEconomy, 78:764-7.

_____________ (l972a), "Estimation of Term Structure Premiums from Average YieldDifferentials in the Term Structure of Interest Rates," Econometrica, 40: 277-882.

________________ (1970), "Testing a Model of the Term Structure of InterestRates by Simulation of Market Forecasts," Journal of the American Statistical

78

Association. 65:1163-90.

______________ (1972b), The Term Structure of Interest Rates, Basic Rooks,New York.

Nelson, Charles R., and Andrew F. Siegel (1985), "Parsimonious Modelling ofYield Curves for U. S. Treasury Bills, NBER working paper No. 1594.

Okun, Arthur M. (1963), "Monetary Policy, Debt Management, and Interest Rates:

A Quantitative Appraisal," in Commission on Money and Credit, Stabilization

Policies, Prentice Hall, New Jersey, pp. 331-80.

Pesando, James E. (1975), "Determinants of Term Premiums in the Market forUnited States Treasury Bills," Journal of Finance, 30:1317-27.

______________ (1983), "On Expectations, Term Premiums and the Volatility of

Long-Term Interest Rates," Journal of Monetary Economics, 12: 467-74.

______________ (1981), "On Forecasting Interest Rates: An Efficient MarketsPerspective," Journal of Monetary Economics, 8:305-18.

_____________ (1978), On the Efficiency of the Bond Market: Some CanadianEvidence," Journal of Political Economy, 86:1057-76.

Phillips, Liad and John Pippenger (1976), "Preferred Habitat vs. Efficient Market:A Test of Alternative Hypotheses," Federal Reserve Bank of St. Louis Review, vol58 No. 5 pp. 151-64.

______________ (1979), "The Term Structure of Interest Rates in the MPS Model:

Reality of Illusion?" Journal of Money. Credit and Banking, 11: 151-64.

Richard, Scott F. (1978), "An Arbitrage Model of the Term Structure of InterestRates," Journal of Financial Econoaics, 6: 33-57.

Roley, V. Vance (1977), "A Structural Model of the U. S. Government Securities

Market," unpublished Ph.D. dissertation, Harvard University.

______________ (1982), "The Effect of Federal Debt Management Policy on CorporateBond and Equity Yields," Quarterly ..'ournal of Economics, 97:645-68.

______________ (1981), "The Determinants of the Treasury Security Yield Curve,"Journal of Finance, 36:1103-26.

Roll, Richard (1970), The Behavior of Interest Rates, New York, Basic Books.

______________ (1971), "Investaent Diversification and Bond Maturity," Journal

of Finance, 26:51-66.

Salomon Brothers, Inc. (1983), An Aralytical Record of Yields and Yield Spreads:From 1945, New York.

Samuelson, Paul A. (1945), "The Effect of Interest Rate Increases on the Banking

79

System," American Economic Review, 35:16-27, March.

Sargent, Thomas J. (1971), "Expectations at the Short End of the Yield Curve: AnApplication of Macaulay's Test," in Jack M. Guttentag, ed,, Essays on InterestRates, Volume II, NBER New York. pp. 391-412.

______________ (1979), "A Note on the Estimation of the Rational ExpectationsModel of the Term Structure," Journal of Monetary Economics, 5: 133-43.

Say, J. B., (1853) A Treatise on Political Economy, Philadlephia, LippincottCrambo & Co.

Schaefer, Stephen M. (1981), "Measuring a Tax-Specific Term Structure of InterestRates in the Market for British Government Securities," Economic Journal, 91:415-38.

Scott, Robert H. (1965), "Liquidity and the Term Structure of Interest Rates,"Quarterly Journal of Economics, 79:135-45.

Shea, Gary S. (1984), "Pitfalls in Smoothing Interest Rate Term Structure Data:Equilibrium Models and Spline Approximations," Journal of Financial andtive Analysis, 19: 253-69.

______________ (1985), "Interest Rate Term Structure Equations with ExponentialSplines: A Note," Journal of Finance, 40: 319-25.

Shiller, Robert J. (l981a), "Alternative Tests of Rational Expectations Models,The Case of the Term Structure," Journal of Econo xnetrics,16:l7-87.

______________ (1980), "Can the Federal Reserve Control Real Interest Rates?" inStanley Fischer, ed., Rational Expectations and Economic Policy, NBER and Univer-sity of Chicago Press.

______________ (1986), "Comments and Discussion," Brookings Papers on Economic

Activity, 1:100-7.

________________ (l98lb) "Do Stock Prices Move too Much to be Justified by Subse-quent Changes in Dividends?" American Economic Review, 71: 421-36.

________________ (1972) "Rational Expectations and the Structure of Interest

Rates", unpublished Ph.D. dissertation, M.I.T., 1972.

________________ (1979) "The Volatility of Long-Term Interest Rates and Expecta-tions Models of the Term Structure," Journal of Political Economy, Vol. 87, No. 6,

pp. 1190-19.

________________ (1985) "Conventional Valuation and the Term Structure ofInterest Rates," NBER Working Paper No. 1610.

Shiller, Robert J., John Y. Campbell and Kermit L. Schoenholtz (1983), "ForwardRates and Future Policy: Interpreting the Term Structure of Interest Rates,"

Brookings Papers on Economic Activity, pp. 173-217.

80

Sidgwick, Henry, (1887) The principles of Political Economy, London, Macmillan.

Singleton, Kenneth J. (l980a), "A Latent Time Series Model of the CyclicalBehavior of Irterest Rates," International Economic Review, 21: 559-75.

_____________ (l980b), "Expectations Models of the Term Structure and ImpliedVariance Bounds," Journal of Political Economy, 88: 1159-76.

_____________ (1980c), "Maturity-Specific Disturbances and the Term Structure of

Interest Rates," Journal of Money. Credit and Banking, 12 (Part I): 603-14.

Skinner, Earnest B., (1913), The Mathematical Theory of Investment, Ginn and Co.,Boston.

Startz, Richard (1982), "Do Forecast Errors or Term Premia Really Make theDifference Between Long and Short Rates," Journal of Financial Economics, 10:323-9.

Stiglitz, Joseph (1970), "A Consumption-Oriented Theory of the Demand for Finan-cial Assets and the Term Structure of Interest Rates," Review of Economic Studies,

pp. 321-51.

Stiguin, Marcia (1978), The Money Market: Myth. Reality and Practice, Homewood,Illinois, Dow Jones-Irwin.

______________ (1981), Money Market Calculations: Yields. Break-Evens and

Arbitrage, Homewood, Illinois, Dow Jones-Irwin.

Summers, Lawrence H. (1982), "Do We Really Know that Markets are Efficient?" NEER

Working Paper.

Sutch, Richard (1968)," Expectations, Risk and the Term Structure of Interest

Rates," unpublished Ph.D. dissertation, M.I.T.

Telser, Lester C. (1967), "A Critique of Some Recent Empirical Research onthe Explanation of the Term Structure of Interest Rates," Journal of PoliticalEconomy, 75:546-61.

Vasicek, Oldrich A. (1978), "An Equilibrium Characterization of the Term Struc-ture," Journal of Financial Economics, 6: 33-57.

Vasicek, Oldrich A., and H. Gifford Fong (1982), "Term Structure Modelling UsingExponential Splines," Journal of Finance, 37: 339-48.

Volterra, Vito (1959), Theory of Functionals and of Integral andential Equations, Dover Publications, New York.

Walker, Charles E. (1954), "Federal Reserve Policy and the Structure ofInterest Rates in Government Securities," Quarterly Journal of Economics, 68:19-

42.

81

Wallace, Neil (1966), "Comment,," American Economic Review, 75:135-45.

Walsh, Carl E. (1985), "A Rational Expectations Model of Term Premia with SomeImplications for Empirical Asset Demand Functions," Journal of Finance, 40: 63-83.

Williams, J. B. (1938), The Theory of Investment Value, Cambridge, HarvardUniversity Press.

Wood, John H. (1963), "Expectations, Errors and the Term Structure ofInterest Rates," Journal of Political Economy, 71:160-71.

Woodward, Susan, "The Liquidity Premium and the Solidity Premium," AmericanEconomic Review, 73:348-61, 1983.

82

Appendix II.

U. S. Term Structure Data, 1946.8760

J. Huston McCulloch

The three tables which follow summarize the term structure of interest

rates on U.S. Treasury securities from December 1946 to February 1987.

Table 1 shows the zero-coupon yield curve on an annual percentage,

continuously compounded basis. This yield curve is inferred from the prices

of whole securities, rather than being based on the recently developed (but

much less liquid) market for stripped Treasury securities. In Shiller's

notation, this is 100 rd(t, t-+-m), as used in his (1).

Table 2 shows the instantaneous forward rate curve on the same annual

percentage, continuous compounding basis. This curve shows the marginal

return to lengthening an investment in rn-year zeroes by one instant. The

zero coupon yield for maturity in is the unweighted average of these forward

rates between 0 and in. In Shiller's notation, these forward rates are 100

f(t, t+m), as used in his (7).

Table 3 shows the par bond yield curve, again on an annual percentage,

continuously compounded basis. This is defined as the (unique) coupon rate

that would make a bond of maturity in be quoted at par, and gives a precise

meaning to the ambiguous conventional concept of a "yield curve" for coupon

bonds. The par bond yield for maturity in is a weighted average, with

60Written while the author was Visiting Professor at l'Ecole Superieure desSciences Economiques et Commerciales (ESSEC), Cergy, France, on Professional Leavefrom the Ohio State University Economics Department.

83

declining weights, of the forward rates between 0 and m. The tabulated

values are essentially 100 r(t t+m), as used in Shiller's (3).

These values were computed by fitting the discount function that gives

the present value of a future dollar to Treasury security prices. This

discount function (pd(t, T) in Shiller's nomenclature) was curve fit with a

cubic spline, as described in McCulloch (1975b), and as modified at NBER-

West during l97778.61

Briefly, the data sets include most of the marketable U.S. government

bills, notes and bonds. Closing bid and asked quotations for the last

working day of the month indicated, as reported in dealer quote sheets or

the next day's Wall Street Journal, were averaged. These observations were

given weights inversely proportional to the bid-asked spread. Callable

bonds were treated as if maturing on their call dates, if currently selling

above par, and as if running to final maturity, if currently selling below

par. "Flower bonds" (redeemable at par in payment of estate taxes if owned

by the decedent at the time of death) could not all be eliminated, as they

constituted the bulk of the observations for many maturities during the

earlier part of the period. Accordingly, they were selectively eliminated

during these years if the estate feature appeared to be active. For further

6l. NBER version fits the actual "flat" price to the sum of the values of

the individual payments. This is slightly more accurate than the version Ideveloped at the Treasury in 1973 and described in McCulloch (l975b), which fitthe "and interest" price to an idealized continuous coupon flow. McCulloch (1971)contains further background information on this procedure.

In only one instance prior to March 1986, namely the zero-maturity rates forMay 1958, did the cubic spline indicate a negative interest rate, of -0.11%. Thisvalue was not significantly negative, however (its estimated standard error was0.54%), and so it was replaced with a zero in the tables. The zero at m — 0 for

May 1947 is the actually estimated value. Since March 1986 forward rates in therange 27 to 29 years have often been negative, but these maturities are beyond therange of the tables.

84

details see McCulloch (1981, 229-30). Since August, 1985, callables are not

used.

During the early 1970's, a legislative ceiling on the interest rates

the Treasury could pay on long-term debt effectively prevented the issue of

new bonds. As existing bonds approached maturity, the longest available

maturity therefore fell to under 15 years, so that values over 10 years are

occasionally missing during this period. The longest available maturity

sometimes also fluctuates by five years from month to month if the longest

securities are callable and hovering near par. Since the cubic spline does

not lend itself to extrapolation, this methodology cannot be used to infer

longer term interest rates than those shown.62

The curve-fitting procedure was adjusted for tax effects, as described

in McCulloch (l975b). The capital gains advantage on deep discount bonds

could not be ignored during the earlier part of the period, when most of the

long-term bonds were heavily discounted. The importance of this adjustment

greatly diminished after 1969, however, when the tax laws were changed so

that commercial banks were required to treat capital gains and losses

symmetrically. After this date, the best fitting apparent marginal tax rate

generally was much lower than before, and was often less than lO%63

The par bond yields in Table 3 are based on hypothetical continuous-

62The exponential spline approach proposed by Vasicek and Fong (1982) has theconsiderable virtue of making such as extrapolation meaningful. Chen (1986) hasimplemented the VF approach along with a modification proposed by the presentauthor, with mixed preliminary results; the forward curves are better behaved atthe long end, but often the restrictions implicit in the VF model and in themodified model can be formally rejected with a likelihood ratio test.

should be noted that the identity (9') holds for the values in Table 3,but only using the discount function that applies to after-tax payments. Cf.Shiller's footnote 13.

85

InN

Nm >40 .N N NNNNNNC4NNNCi10 40 04- 04.4010 ,-I— >4 .4 4 .4.1 4 .4CM .4 N4 m

In .CM .4 NCMNNNCMNNNCMCMCM

N>4• 04 .4 -4.-4NN

-4 N0tfl00m>4 m m 0 In 105- In _4...4_4...4

om >4 NNmIn40.4.4.-I

In Or-b0%mO%NNmCM.4 >4

- mIn 'In0CMNONInN.00% >4 0% 000O CM 000N NU40,-4N >4 NNbN0%OO,-4CM.-IC ooooooo-4-..-.tn 0 S10NN%0In4010 NNN0O.-4.-l0C 0 0000000,-4,-4.-4,.-4,-4

0If1 In In .0 0 S 10 0% 0.4.40%C 40 00000000,-I.4.4OIn 0 .4NInN0%InIn 5040000%d UI 00 00-4..NIt) D%CS0%a%000%C

0 04N.'-4CMNSN10IAON.1E O%O00%c m 0CM 0 mNNLfl.-4In00 CM 000000000000CM 0 CMmCM00%mm'm.-lO

Z mm mm CM mm It) N N 0%.4 ccoccoooo0 1004 0%0%Dc'lInOm.-I.-4.-40000 -4 In 0 0%

C 0 cccc000c0cc0N .4NmoIn4oNo%o..4N.4 .4.4.4

.4_I m in o n mm mm ..o'0'0'0CM N CM CM CM CM CM N N CM CM CM

04 N N ID 0 004 NO.4 .-4 .-4 .40.4-1 CM .-4 N ('4.-I

('INN (MN (MN (MN N N N

10 '0(MO C N .4N '00'0SN Nb 40 SN S

40 '0 m 0 in .1 m SO m10101010 '0Ifl 40 N N NS10.4.4.4.4.4.1.4.4.4.4-1.4

N .-40 510 m 04 N In Cl CMIn In It) '0 m'0 it) In 4040 It) In- ,-4 - .-4 .1.4 - .4 4 .-4 .4.4

N C) m () It) N Sm -4100mm mm N mm '0 '0 In'0'0

.4 0 it) N it) 10.4 04O .4 .1.1.1.-I .4 (MN N (N ('4

.4.1-4.1.1.4-1.4-1.4.4

N In 00% .-I0 In CNN (''0000.-40.1.-l.1 N CM N N- .-I .4.4.4.4.4.4.4.4.1.4

100% c In In 10'0O% 40105.-I0% 04000009.4.4.1CMC 0.1.4.4.4-4.4.4.4.4.1

'0N -4 '0 0 In ('45 '0 m In 0%040% 000000.-4.-4.-4.4C 0.-4.-14.-4.4.-4,-4.4.-I.-4

N 100 N mm .-4 In N .-l '0050% 000000,-4.1,-4.-40.1.-4.1_I.-4.-4r4_I.4.1

.4 It) .4CM ('404 m 00% ('41004040% 000049.40.1.4

00 ,-4 -4-40 .-4 .1.4.4.4

CM '0NO% 00 m C. .4'004050% 05000% 900.-I_I

000.4.40.1-1.1 ,-4_4

101" 04 0410 0.1

0000000'0C'40 S It) N 04-4010000 040'0%0.-40.-4,-4O

.1-4000-4.4.4 ,4 .-4 -.

,4 Nm '0 In 105 0.0.4CM.4-4-1

m 0.-I ("I N 11% 010mm m 04'0'0 '0'0 o mm CM ('4 ('4 ('4-1

moo 050 .-4 N ('4.45.-4.I.4.-400040%0%040%('IN CM ('4 ..4 .4 .4 .4

0% '0'0 050 m 00CM NO1010101010 '0'0'0 '0 '00'0-1.1.4-1.1-1-1-1-1-1 .4

NC. m .o 0'0-1004-4.1-1In It) it) it) In mm m N mm m.4.I.4.-4.-lr4.-4.-4 .-4.4 .-4.-4

40 mm 0 '0040 0.4 ('4'0'0'0 0 '0CM .-4 CM .4CM ('4 ('4

It) 10 '0m .4N0 ('40 CM'0'0mm mm m ,.4.-I .4-1-1.1-1

.1.-4.-4.4.-4 .4.4.4.4.1.4

it) 4' 10 m '0 04001.-I In NCM N ('4(1 ('4.10.40-4.4.1

NO 00 m 04000.410 ('4N CM ('4CM ('4.4.4-1.4-1.4.1

.-4,-I ,4 ,-4.-I.-4.-4_I .4.4.4

010CM .1000.4.4 SCM('4.1CM .4 ('4 ('4.4 ,-4 .4.4.-I ,-4

.1.1 . . .-4 .-I,-4.-44_4

04 In .4N0000..-I.-4 ('4.1.-I N .-4 (N N .40.4.1.1.1,-4 .-4.-4 4 4.-I.-4.-4.-4.4.4.-i

'0010 0%C.OOs .4 .-4N CM('4 -4 .4.4.40.4-4.'-4.-

N 00% 1(4 0010.4.1.1-4.4.-I .-400.-4.-4.-4.1.4.1.4.1.1 .-4.-4.1 .-4.1.-4

It) 1'4 10 '0 i m ill In N ('401.—I —4 .4 ,-4.—I —40000 r'l C

m'0.1 ('40% N 10 .'-IOl m- .4.4-400040400400.-4.-4-4.1-4r-400.40.-I.-I

0410moo '-4050 N CI m0,40.40 0'-I.-4.-4.-40000000

.1 (Mm 0 In 10N 040,-I N'-4.4-4

InN

0('4

>4It)-4

0'-4

>4

It)

>4

'0

>4

m

>4

N

>4

-4

0'

40

It)

'0

m

('4

-4

0

>4InCM

x>40N

>4In'-4

>40-4

>4

Ill

x>4

>4

m

>4

CM

>4

-.4

0'

10

it)

•0

N

-4

0

'004-4

0N

0'-4

It)

'0

m

N

'-4

0%

10

In

'0

m

N

-4

0

05'0OS.4

1-4

Q

I-I 40>4 '0ZZCMr ,-.0

o0F-.—

(1

'0 '004

'-4 -4

87

O.4OOinNIA>4 IANNN401040IONNoN NNNCINNNC'IC4N

0%0.4IO('4ONf1ON0>4 000%ONNIOLnIQI-NNin.4 C'1NNCINNNNC4(I(Ic4

X IA.40O,-40IflN.4NO>4 .4NN'0IAIAONOUIIAIAo.4 CICINCINC.INNCMNNN

X N.4NIO.4NGI0NIOO>4 N0O,-4.-4O0s.-4.4,4NIA ,-44-4NN('4NCI,-4NC'4C1N

O.4N04ONOONOINN>4 NN00000IGI000.40 .-4.4NNNN.4,-4NN('IN

z NNIOCINNIn.4O,-01>4 40IOGI0%0404O00'ON -4.-I.4NCM.4N

NIANIA04'0NNCI.4CI> U,UINN0%0%ON .4.4.4,-5,-4.-4.-4.4.4.4.4(I

NIAi0N,-4,-40NNO>4 (nIANNNIDN0I.4 .4.4.4.4.-I.4,-4.4.-4,-4

o GINNU'I.4O0%O40(1E (AIflNNNN40NN0%Q%

o OGINO,-40%OIA40N0140UI4DSNI0NNNN10 .4.4.4,-l.-4.4.4.4.-4.4,-4.4

o 0%0'40ON0I0NONE UIUI%0NNNiOSNNNIn .44.4.4,-4.4.4.-4,-4.4.4.4

o N'00%NNO%.-I'-4O0UIUI10IO%0NIDIDNNN'0 .4-4.4.-,-4.4.4o OIA.4NLflNtfl0%UI40'0Z UIUIIO.010N101040IOION .-4.4.4,-4.-4.-4,-4,-4,-4,-l,-4.4

o OLfl40UI.0'0,-4UIIONDIO'0UIU1411404040.DInUIN

(I .-4.4.4.-4,-I.4.4.4.-4.4,-4.-$

o .4,-4ON'00%O040Q4OE (IC'IONNN'OLtIIOC'4NN_ .4.4.4.4.4.4.4,4.4.4.4.4

o OON40OO%('90%,-IUIUIE NCMO'0'.0%r4NOOtn0 O0,-4000,-4,-4.40.-I.4

N '0UI ION 010.4(1.4.4.4

.4U,0%.4

InN

>4 55N404040405p-50 0(I NNNNN(1(ICI(1c's(I Cl

00'O.4r4,-40%N04Oh40>4 N5N40In40404040ION >4In.4 NNC'ICIC4CINNNCINN .4

X 0.OMO01NNI000NIO>4 '0IOIAONOOInNInUIIn0 0.4 NN(INClN(4NNCI('4CM .4

0ON0N.4N'0iA>4 .4C'4.-I.-4.-INNOONNN >4II'. ('4NNC1NNCINNC'4('4C4 in

'0O(14004%014l.4'0>4 O.4.40O.4CINNCINN >4

0 NNNN(1N(INNN(I(I '0N.4O0NC'ION('s(-0>4 0100DI0.4NNC"4'sc.4N

N .4CMN.4CINCMNNNNC4 N

N0Nin11%N.4NN.4O>4 O%040%0.4N.4.-4NCIN '-4.4.4.4.4CICINNNCIN ('I

01InN.40.iA.404o,-INIO>4 NNGIOO00.4.4 >4.4 .-4.-4.4.4.-4NCMNCI('4N .4

o %0.4NObNU1ON.4'0E N5DI0%O00.-4.40'. ,4.-4.4,-4r4.-4.4NN.4(IN 04

o NI1INOOOOC'4N.4E NN40N0%00%0iO.440 .4.4.4.4.4.4-IN.4.-4c4N IC

o 0'NIO40NNN.40iUI0'E ION40N00%0OOUI 4,-I..4.-4NC', in

o %00NLflNU1,-4UIN40Nx 40N10N0'0%OO'0 ,4.4.-4.-S,-4.4.-4.-I,-4.-.CIN '0

o N1flO'.C'IO%NOI.4.4("lO%IOi04DLflNN0%040N .- -4.-4-4.4.4.-4.4CI No '04ON1AO40IA0U,N(Iz 1fl1iIL11IONNN0%OCM -4.-4.-lN No N'0N0U,0.440Q'ToE NNNIAIONNOIn1'—0..4 .44-4.4_4.4.4.4.4.4.4__ -4o 0000%NItIC'44010C'4IANZ 0I0I.4NOIONO0NIA0 00.4.4.4.4.4.4.4.4-4.4 0

.4 N NO UI ION 040.-SN——.4

NIA0%.4

-4-4.4inN NNN

ON 40.4 in 0.05.4(10-5-4040.05N N N N N ('4 (INN

00(4.00.40 in NO 54(4.4• 0.04.4'400 - 5N (IN ff4 N N N N N N (1(1

IC N 400.4.40.041(4.41(45'0550 N 001040550N (INN N N (IN ('4 ("4(1 N

005 10 04 400400400-40005055 0001CI (INN N (I N N (IN ('4(1

N N.40.4OS 40 N NOONN N 05040 40 N ON N .4CI (INN N N N N (IN (IN

40.4 N N .40.5000404.1N N N 4004 UI In ION ('4.-CO

N (MN N ('4(1 N ('4 ('4 CI (IN

5400 41% 0' 541% IA 5015.4 CI CI IAN 000.40404CI (IN N N N ('4 N N .4.4.-I

0. (IN ION N.4.-4IA 04010O CI .4040 N N ('.404 N N 40N CI CI N (IN N N .4.4.4,-S

4000 N 5 %OONOOIOO ('4.40 i(4 Nfl -I 5104040(I (1 (INN N ('4 ('4.4.4.-4.4

N GIN 0100.410 N 04 Nfl0.40(1 (ANN ON 0404(4(IN CI (IN(1(1(1.4.4.4.4

N S IONS NO' N NO 0019.40 NO.4.-b 4" OIAO(IN N N CI N ('4 CI .4.4.4.-b

010 ((ION ,-4 400.50410 400.-4O NO.4.40440 N 110NNNNNNN,4.4,-4.-4.-.

SIAN 0 in N CI UI 04.-I N (10 .404-1 NO.40% IflN 1110. (IN N f4 (I N 4.4.4.4,.i

N .4,-IN N 04.00' '00*0' 1(40'-4OC'IC1SO0.4'0-4NNNN..4N,-4.-4.4.4.-.

(IN 0.100.400% (150 III5004001100.5(1010(1CI .4 N .4.4.4-4.40.4—

N 10(115 N S (MN IOIA 4004050150.45105 ION 0.4.4.4-4.4.4.4-I 00.4*.-4CI NO III ION 5040.4(1

.4.4—

N01.4

0N

-0.4

In

0

N

N

.4

0"

.0

III

'0

N

N

.4

0

0U,0%.4

IA 5.4010400% GIN (IN III('INN N N N N N 00N N N N N (INN (INN

4001 N IAN 0.4 N IA GiN N0401000.4.4 - .4.4(1CM- .4 N N ('4 ("1 N N CIN (IN

N 5.4.01(4010 NO 00500 IA IA IA 40 IA ION S N.4,-. .4.4.4,4.4.4.4.4.4.4

404004 IAN 040.4.4(105N N N 000044140N N 40.-4.4.4.4.-44r-I.-4,-4.-4.4.-4

IA IA S ON DIN .-4 NO 00%N N CIN N N N '0 IA '040111.4.4.4.4.4.4.4.4.4.4.4.4

40504000 IA 0010(1(1.4.4 (MN N ('1 CI N '0 (A ((4 IA

IA N .-440N04 0140 N N 01441.-4.4N (INN (MN '0 IA '0(11—'-4.4-4--4,-4,-4.4-4--I.4—

111 SN tO S ON 010 (I 0% 5N N NO IA 0411

.4.4.4.4.4.4.4.4.4.4.4.4

400' 050.-S 0 N IA 00.4.4.4(1 (INN N '00 (A 44110.4 .4.4.4.4.4.4.4.4.4.4

100% '0N.4,-SS(I UI 0'. 0.4.4.-4N (INN N '000 UI.0

.4.4.4.4.4.4.4.44-4.4.4

400105.405 NO 5010.4.4 CI (MN ('1 N 000010—.4.4.4.4—.4—.4——.4

UI S N 114 0'. 0110.-I NO 55.4.4N (INN N 0000 UI,-4.4,-4.-4,-4.4.4.4.4,-4.4.4

N4004 NIA UI .4505 NO'..4.-4 .4 N (1(14 N NON '00- _4 .4.4,-s .4r-4.4.4.4.4.4

NO 5.40 N 40 UI 0% 5.44410.40.4.4.-I ,-4 ('4(1.4 N N

N0O% (IN 0.000 N NO'.0*0504010450.45050.400000.4.40.404 (IN 0 UI ION 50'. 0.4 N

—.4.4

88

>4 0%0%0%0%In . LAN 4NNN N

Ne4Lbm0>4 0%00%0% >40 . 0(4 NNNN N

0% 0 %O 00 —0 - LA N N>4 N0%0 0%O% >4U . LA.-l NCMNNNNmN(MNNN 4

LA 0 Lb m Lb 0% Lb 0% .-4 0% N>4 Lb N N NN 0% 0% 0% >40 • 0,-4 NNNNNNNNNNe4N -l

0%NNmm%omo0%oIn X>4 LALflLb0%NN >4LA JNNNNNN LA

OLOO LDIANLOON>4 LANNLb >4

1NNNN 0LAN0LANLbN0mLb>4 0LbNLbLb >4

m NNNNC'4NNNNNNN enc LA LA0%0% O-IO Z>4 NOLA >4N .4NNNNC'4NNNNNN NQ NNLAN -40%c>4 0N0% -4ONcnLbN >4.-4 44.-4NNNNNN ,.4o Lfl0uLAtDLAmcnLArnZ0% NNN 0%o NLbNenN,-ILben0%Lfl-4E NNLbNQ%C4.-%,-ILALbLb e-4.-4.-4.-INNNNN Lb

o N 0%NLbC'40%NLA,-40% 0Z 4LA0LbLAL00%r-%.-4LflU1 Z11% NNNc4r4 LA

o cntnenu,..-IONLALbenNLD

0 -l,-INNNNC4 0

o 0%0%0NOLA0-4enE 00LA-%,-,0LAen NNNNC4 C%o N0%LALANmLA N.-40% 0*: ONenLA NO-lOenO EN .I,-I,-44NNNNN No OOenOLANNLALALA.-IN 01 0%0 Z.-4 -1.-I-I,-4NNNN lo 0%-LDm00 0£ 0%OLt%-40%) E0 .-40 0.-4.-4NN.-4N 0

-I N en 0 LA Lb N 0% 0 N-I

0LA0%-4

Lb LA Lb N ( ON N .-4 N0% 00% 0% 0 4 ON N 0

N N N N N mm mmIA 0% IA mm Lb N 0 N — N

0% 00% 0% ON .-4 N en LA

ON 0000-40% 0 en O en000% 0% 4 N N en en Lb

ON D 0% N 0% 0 ON 0% .-400 0% ,-4 0 en 00 NN N fl en N N en men

( N Lb en 0 N N 0 N ,-4 f .-4N NO .-4 0% 0% N LAO LA Lb N

NC4Nmenm0% 0 LA Lb IA %D LA LAO Lb .-ILb NO - 0.0% N LA LA Lb N

N N A% N N en mm e) men

0 0 r4 Lb N N 00% ONLb NON 0% 0. N LA 00 N NN N mm N N en men mm en

Lb Lb en LA LAO -4 Lb 0 NLA Lb 0% N Q% NO LA ON N

N N N en N N mm en men en

LA LA en OLD 0% O 0000% 00 LA NO Lb 0% N en en LbN N N N N N mm mm m

0% NO en 0% o en o Lb00 Lb 0% Lb N -4 N N Lb NN N N N N N m mm men

,-i en o LAO en ,-4 .10 enen o 0 N 111 Lb 0% 4 .-l 0 LbN N N N N N N N men men

Lb en .-4 N en Lb Lb LA Lb 0men 0 N N LA LA 00 en LA

NN N N N mm men

LAO ON 0 0 N mm en en N Lb 0 0 N 00% N 0N N N N N N C4 N en N men

NON LA LA 0% en 0% LA 0%% N en N Lb 0 en Lb 0% — ('4

N N N N N ('4 N N ('4 ('4

— Lb Lb en 00 0 Lb 4 LAen .—4 N Lb IA 0 N LA NO r—I

N ('4 N ('4 N N N ('4 N N men

NO en N en 00 Lb —N 0% N%D mm o ON IA 0N 4 N N N N N N N N (N m

00 en — en 0 NON en 0N Lb 4 Lb .-4 N '-4 LA (N LA .-4

N .-l N N N N -4 N N N N en

— ('4 en o LA LbS 0% 0 N.-4 _4

Q>4LA("I mmenen..,enen

>4Oci

>4 .-4mNenoLb%n%nNU%CLA.-4 mmm'enmm.-

,-lLbNen0o,-4,-Ioc>4 NenmLALb0.o0.eno0— menmmenmenm'ommr

LA0%NenriNOLAN%.'>4 Nm0LALbO0%gLA

OOenLbONNONNU%...>4 NOOLALDNO0%00a: ON0LbOO00O.-4C>4

en

LDLALAOLbLAO0..-.LAX?->4

N mmenmenmmenooen

NNLDen0L,-IO,-40%,-4L0->4 NoenooN.-..-4Oo—.-4 enmcnmnmenoooen

o OOLr%OLbOOenE NONenOLDNO-40%o—0% mmrn'ooenmeo en(NLb,-%omoLAmNbZ'.0 enmenenenenmenenmen.

o -40%.-4NLb0%NC%I0%00E NmoLANenNIn menmmmenenenmenen

o LANNen-I,.-4CN00 enmenmenenmenenmeneL

o 00%.-4NO'.0NOLA0E .-INoOmeno.Ar--Lb—en menen(nenenmrL

o 00%InenLbLb0Oen.O..4,E 0%ONNenOLAIAOr—N Neneneenme'o Lb00%N,4.0r-mLbo.-E O0LQN0OONenLb—— enNNmenmm.-.enencif%

o LA0%,-4LA0,-4LbOOLbE omwo-4'.o.nN..,e.O mNNcnmNrc.Nmrei

.-4 N ("s 0 IA '.0 N 0' 0 Ct— — -

NIA0'-I

0N

0-4

IA

0

en

N

-4

0%

Lb

LA

0

en

N

-4

0

0LA0'-I

(N — 4') ON tO I') 0 LA 0% 0%'.0 Lb Lb Lb LA IA LA LA IA Lb Lb

NNNNNNC%INNNNN

0'. en o .- — .40 N en o'.0 LA LA IA Lb LAO IA LA in o Lb

N N N N N N N N CNN N N

0' N (NO .-4 en 0% Lb Lb 00% —0004') LAO enoo 00 LAN N N N N N N N CNN (N (N

0' m N en If) 0'. 4 NO-I00% 0% NO 0% 0' -4-I NN .-4 .-4 .-4 N .-4 .-4 .-I .-I ('3 N N

LA lAO 0 IA 0% ,-4 0%0' N N Lb Lb Lb N 0% 0% 0.-l .-4 .-4 — .-4 e4 .-4 ,-4 ,-I — .1 ('4

'.0 LA (N .4 00% 00% 0%N LA LAO LA 000 Lb N N

,..4 4 ,4 — ,4 ,4 r1 .4

00 en LA Lb Ill LA IA enLA en en N N .-4 — NO LA LA Lb

— .-4 — — 4 ——

0'. N 00% Lb LA 000 Lb LA 0.,'-l 4 .-40% 0% 0% ON N ('4 N

0 Lb 0% .-I N N 00% N 0.4 000% N 0% — .4 ('4

0O0O4.-4,-4,-0.-40 en 0.0% 0% Lb en 00000 N Lb N 0% 00.4 '-4- -4.400000-1 ,-4 -4-I

N 0% N .4 Lb Lb N LAO Lb N0.0% 0 N Lb N 0% 9000

en en LA LA N N LAO0' 0. ON N Lb N CL 0% 00000.-s 000000 .-4 ,-4 4

moo LA 0% c en Lb 0 (N ,-40% 0% 0 N Lb LbS 0' 00% 0000 -4 0000000 .-4 '-IN N 0. ,-4 Lb 0.-I Oc's en0% 0.0% 5 Lb Lb N 00% 0% 0% 0%oc-oOccLb 0% 0% NON .-I0 LA en LA0% 0 Lb Lb Lb Lb .-l0% 0% 0%

0000000-40000LA 00% N .-I N 000 ('4 Lb ('90 0% Lb LA Lb Lb en 0% N 0%ooo-oocc.4 N en 0 LI) Lb N 0% 0 '-I N

-4 -4 -4

LAIA0%-4

89

InONOØN,..IN>4 mN,-4(0NNNNNInN

>4 mNN.4.-4NInNNDDNo("4 me',(

NUC"1.-l0'N0%iflIn>4 N(.4r40,-4NNN4DDIn.-4

X>4o,-4 NC'4N'r)m>4In NNNNm>4

>4 Nmr'00N.0InDInInO(' NNC'4NNNC'(%)4.)

In00N0NDOU>4 In.4NO%0mLn.0'0N N44Nme'.i(.'

N0ND%O0%U0%>4

'-I N-4,-I.-4-4.-4Nmmmo In0%N0,-4,-4N(.OE0% .-4.4-NNo N,-INNNi%&fl0c(.NE N'0(.)C40.O,-I%ONN0%'.0 .400'4NNo '0.4InInN0'.EIn

o %O'0I00.00.'0NN'0E ifl('NNN0'0O'.OO0

o 0%N-0%e'-E '0N,-4.-40NO%e'1n0.N(% -4-4000C'4(.4c"4Nc"a

o '00In'.00%NCONOE '0N0,-I'0t--NInmInN 000 c"clco .0'00%InInInONInNONz 0.,INNN0.0'.-I O,40004,-4,-I.4No (')00'00E 0InmOD-Io O.-4OO0.-4.j,-4ON

.I('4 '0 In '.0 N 0.0 .-I ('1'-4-I-I

>4 OO04040004000%0Nin .('4

NNNInOInN0%'.O>4 000%0O..ieQO0oN '0'0'0'0'0'0'0'0OONNOIn'.Om,.I,.I>4 O040%.-I,-4.'INNNNmInin .

>4 040%N.-4N'0'0'0'InNo.-4 i'm'0'0'0'0'0'0'0'00O%.-N0'.04.,,.i

>4 0N04.-IN'0In'.D'.Q'.Ci-in '0'0'0'0'0'0'0'0O>4 0%N04'0In'.0N'.ON'0

0NN'00('ONO'0,.4>4 0%N0%04'0InN'0l'>4N ''o'o'o

00%b,4'.oe'.inlnr'4>4

,-l c'm(''0'0'0'0'0Ino NO%Q%0%,-4'.Oif%0r').X NN'0'0N0'.N'..flO%n0.O0%

o 0%In0.N0In0.0.ON.0.0'.0

o '4N0%NN'.0r-NNNZ 0ON'0'.0InNNmNin

o 0N'.D0NN,-4mmoNz 0%0'.0,-IN'.4U1C'4'.DN'0

o '.')'.0'0InN'.DN.0N0.0.-I00'9DLflUI4) NNC'4N'0'0'0'0o '0N0N.n'0oQ%Z '.0'.DN0%NN0%NNNN

o NN'0InNO0N'0rnInNNN.0..4In0-4

o C'4'0'0O0N'0e',-4NE CNNQ%'0,-I'.".O0 NNNe'4NN,-IcmNN.o

N '0In %0 N 0.0.-IN.4-4-I

0'.In04-4

'.omo..4'.0in>4In("4 '0'0rme.>4 4-40NN0SNØ'.Ø'.oN '0'0'0'0'0e.mmmX>4 4N0N0%OIn.4

X 0N'0In0NN.-4OoIn>4 %O'00('.4N0%eN0

>4

In m'mm>4

'0

NNIn0%N0'.N>4

>4 N('4.-4b,-I,-4O,-4mN

>4

.4

o 0In0.In'.''.O,-,.IZ0.

o .-4(')InInONO'.,-4N.0Z inin 0N0'.0

o N'0'0'0NOIn'.N04.-EIn

o In'0NInW0.'.tz NNN(')In'0

o .-4U10N'.QO,-4,-4(.0%'0Nz(. C'4C'4NN('4C"4C'4

o NIn,-4NNNN0.'.z 00.000.0)04,.-4,-4N N('.-INNN..4NNo N0In(.N'040%r')z In04'.0'.DO'.'.0N0'-4

o 00.NInInIn0'.N.0Nz0 N-4,.'4(.4,-4.-4,.4

("4 '.'0 In '.0 N 0% 0.IN-I—

0'.00%'-4

90

m'0NN'00,.-4N'.000'0>4 04NN0%0%000O0inN

>4oN.40NNtn'00%.-4e.4'4N>4 0.0400%0%00In.4 mm'.')m'0(.,(.'0'0

>4 NN0.00o'.ooo.-4

>4 InIn'0InNNNin

IninQ'0'.b.>4

'0

>4 '0('NNe.)InIn%0In'0InIn

In0%In'0O'.0SN'0(.>4

("4

%0NNOInO'.Ør.>4

.-4 NNNNNNNNmmo '00e'0'.InIn'..E '.00%NNr-Nr-0'.o0% NNNNNNNNNNNo N0'00'0O.4Ob0.OE In'0'.InInNN'.J%'.D N(.4NNN(.4(.4Nc4NNc.4

o bInNe.,In.I0NInOE '0NInf')In'0'0%InN0.In NNNNNNNNNNN('4o O0%00'.0br0r-a'0x'0

o 0NNbO,-.'.'0'.'0N0in:)o '0NO'0'00'0InInNmZ .-I In ()N ) N .4 ('4.-I -40 '.0N NNNNNNNNNNNNo.4

o NN('4OOc"i'00%'0oNE0 .4NN.-4N-I,-4,-,-I,-4NN

N In '.0 N-4-4—

'-4'.00%-I

enNm0%.u,Ln4 O0%0%Q0%0%0%0%In .N en mm en en en en en N

%00000%No%LnenoD4 O,-40000010%0%0%o .o(N ' en en en en en en N

>1 $0%0%C%OO0O0%0%0%In • in,-i enenmenmen ,-i

ensr-sci>o • 0i enenenen'enmenmen .-iNN%D00%en0%NNOIflLn

in enmenenenenmenmenmm in

(NnenOU%IflIflenmenenenenenmenenenm

NNLflNN0iONNNinD.i senenoenenmenenen m enenenmenenen en

p: NOin%O.-iNU10%QN(N

N mmenmenencnenenenenm N

0%oen4en.-iNOO0%ON,-I0%OO- enenr4enCNencnencNc.lenen -4

o eNLncno0%0%(Nulz .OOOO0%0% mmN 4rnenenN(N(NCN 0%

o InenNNO%0%c'lenue-os0%ooo%o%o%D N N (N (N (N (N en en N N N en

o OenencE 0%O0%O00((I C4NNN(NNCnC4NN(Nen In

o NONWN(N-4u1OE 0%0%o\0%o%oNNNNNNCN(N(NNNm

o entnou,mNZ NNN0%r-0%0%en N (N N N (N (N N N (N N (N (N en

o %0IflO(0CNbUD0E sbsr-oN NNNNNNN(NNN(N(N No eno%Lneno%o%enr-0%tti.00.-4 NN(N(N(NNN(N(N(NN(N .-l

oE e4Nu%Denen0%o NCN(N(NNNNCNe4(N(N(N 0

N en 0 in %D S 0% 0 '4 (N..-4 .-4 .-I

N0%-l

0 en o ..o in .-i -i f• Q% en9999090000,.I.-4•0 pen 0 (N N %0 - en oi0% 00000000 '.4 -4

In % %0 en 0 %D 00% 0000000 4 ,.l .-I Nen

(N 0% N 00 N r4 00% en0% 0% 000009 .-l- .-l -en en

en oi 0 ,-4 %0 .-4 (NO .-l N'.0 NN N 0% 0% 00'. 0

0 Ifl '.00 IA 0% .-4 en N enLA In LA '.0 '.0 '.0 0% 0% 0

menen C', mm en

en '.0 4 en 0 ..4 it) In LA Sen en en LA LA S N N 0'.

en en en men en en en en en en en

it, '.0 u en en in in o'. ,-i.-4 .-4 en en LA '.0 '.0 N N 0%

en en mmmm en en men enm

0'. 0% ,-4 en '.000% S0% 0% 0'. 0 v- '.0 In '.o NCNN N en en en en en en en en en

N en II 0 -4 0% '.0 N0% 0% 00 .-4 .- it'. U .055(N C men en men C". mm en en

In (N en N (N (N ens u ON00% .-4 0 '-4 0 In it) '.055en (N men 4". men en men en en

(N SN (N (N (N 00 '.0 O 50%00% .-$ 0 l .-4 l.A In '.0 '.0 '.0

en N en en en en en en en en en en

(N N 0% 00 '.0 NO00% 00% en '.0 '.0

m (N en m m en en en enen en

In '.0 LA 0'. S NO'.0'. 0'. 0% 00 (N 0 0 l.A In Inr (NCNN en men en en en en en

Il'. 0'. 000% (N '.0 '.0 en0% LA 00 en ' In(N (N (N (N m enm menen

en 0 InIn In '.0 .05 .-40% ('4 0% 0% ON 4". (N In(N (N ('4 (N (NNenenmenen en

it) cN en en -40% en in ,-4 (N0% 0'. 0% 0 In (NO LAN (N - (N (N (N N en en en en en

.-4 (N en in '.05 0% 0 .-4 (N,-I l

0%0enm'.00%-I N-4-44-4.4,-l,-l-4In(N

N('N.-4ifl'00%.-4O5O(N>4 (N.-4.-l,-4Ne.4.-4(N(NO(N

X .-4NIn.-40%4en,-40%Oen>1 N(N('4N44(NNCN,-4NNIt)-X .-4'0'.0,.40'.0,-IenO'.Na'.N>4 N(NN(NCN.-4C'4N,-4.-4.-4(NO

z ONN'.D'.0%0InN>4 O.-lN.4OOO0OO,-In .00'O0.0

0smen0ONN>4 09,-4-4OO00O0.-iO•0

enocNcnInNen>4 0l0O0%0i0%0'.0%,-40en menmenenenmoox r-r-NenLno0%Nen0%>4 0%00%0%N0'.,-40'.(N menoenenenmenmenoen

Z O-4-4NLA.oo'.sino>4 N0%0%NNN,-4O.-4 menenenenenenenenen'o'o

o en.oLnu'.%o(Nenr-InLn'ONZ NNSNSr'-.-40a'. menenenmenenenenenq.o

o N0S'.00enIn0%0'.oz s'.o'.o'.o'.o'.osseo'.o menmmmmenmenm.oo

O senNeo%ol.ntnenwz '.0SN%0%OLfl'.0%.OSSOQ%In menmmmmenmmm.oe',

oE ..0N'.DInit)it)In'.0'.0S0%O'.0 mmmenenenmenenencnen

o 1nLAO,INLA,-,'.Oen(NOO'.z InLAInIntt)'.0'.bQ%en enenenenenmenenenenmen

o 0%CNen'.0O'.0,-4sen,-4oLA LA 'U it'. In in

(N menmenenmenmenenenen

o Nen.-4cN0%'.o(NoenInen'.oE 'ULALAenCNqenou,'U'.DIn- enenenenenenenenenenenen

o Uh-4NO(N.-l00'.SenE mLnLn.-io(NocNLnenen.-,O mmmenmmenenenmenen

.-4 N en In 'ON 0% 0 .-4 N.-4 .-4 _l

'U'.00%'-4

>4 -4-4-.4-4Nic'.ien'Uin(N 'U'U'U'UU'U

0000.4oNIn.-4en'U>4 (NNCN(NCN(NNc'.lenen'Uu'.0('4 'U'U'U'U'U'U'U'U'U'U'U.

.-4en,-4,-4enNensNoo,o.>4 CNN(NCNNc1('4Nen'U'UInin-4

0%ene0enoc',-4'.o'Uo>4 -4(N(N(NNNCNN'U'UInbo

0 oen'U'Usen'.o.ooN>4 -4-4('4en'Uit)0%I',

Z %DO('4(N,lNino'UenM'U>4 o-4o,-I('4en'U'.no0

CN0'O(N,-4Nen0'Uo>4 000.-400C"OIn,-4en 'U'U'U'U'U'U'U'U'UO'UIn

>4 0%00000%o,-4enc',LAo(N m'U'Um'U'U'U'UIn

en'UC'.Nbenoo'.nIn'U'.>4 0-40O00%0,-,me-'.'U0%,-4

O '.0tn0'.0'U0%NenencN(NE 0,-4,-4000%0%,--44".en'U0%0'. 'U'Uenen'U'U'UOUo i0ensNeneno%o,sNX 0.-l0000.0%0(Nc'irn'.0 'U'U'U'U'Umen'Uooo (N-4InLflO('9'.0'U'U'.D('4InE 0,-I0000%0%ON('4e".NIn O'U'Umenoooo NN.-4('40%enO'.(N'.n'.oE O%O000%0%O.-4C'N.0

en'U'U'Uenenen'U'Uo'U

o en'Uo%'.nsa%m,-4enr-E 0%0O%0%0%.-Inen m'omenenmeneno.oo 0%0'U'Uen'U'U,-4(Na'.(N

(N en'Ummenenenen'U'U'U'U

o in5('40%'.Ø0%,-4,-41fl(.4en'UE N0%0%0%'U.-4 mmmenmmenmmr'.en'o

o O(N-4,-40eninN.-Z NNSO%'.0eno enenenenene-'.mc',enenen.o

.4 (N en 'U LA '.OS 0% 0 .-4 N-4 ,-4 —

Ii,00%-4

en'.00''-4

91

OIDNONIAIDO%4 Inr.N0%r-I0NInIn • InN VVV N

4 0NI.DO • 001 VVVV NVO.-INN.-400%DUN4 ON0%O%O• In

010 V.-4IONr-NNN-4NInQ0Nr-o • 0.-4 flLflIfltflV ,-I

4 001010 VOVmcIll f%ItflIDLflLflIflV In

zM 00101000%V U%IflU V

4 NO0LflU0%LflIflInV

V000%rN0.-40ONIDID0N LSLflLI% N

X OIDr4.-404Nmo%oON0%a%r-V0— InU -l

o V(%NInN0%NVLfl 0Z e0%01o010%r-Vo E0% flInIflLfl Q%

o NInNDCO,-z10 UInInflII ID

o N00.-I%CVNNNZ 0%NO%C'OIn L$UILnInLfl In

o 01010 O-4ONN 0E NNNN 0%NIDIflOV tLflIn Vo N O0%INIDV 0E eo

V V V In In In Lfl V

o — In 0% Lfl 0 10 Lfl Olin N 0x In10IDN0-4,-4% EN OV0OVVVInInLnIn0 N

o 0100 0%0%O%OInNOZ InNNCN0VVVVVVInVVV 1-4

o O-4N%'-40V'0O 0z I1%N,- ONON%CID Eo OVV VVVLnVCIV 0

.-4 N C) V In ID N 010 ,-4 N-4 -4 .-l

NID

I-I

0O,-INV InU;

.4 Inm 004 ID 0 N 0 NIn N 1001010 .4 01 In N NV 0 V V V In In In In Ia) In In

V10 N ) In QI 0% ON 4U) 10 0 0% 0404 NO) ID N

N NOV10 V U) N 04 .-4 01 N10 100% Oi V 0% V In N N

V U) U) U) IAU) In U)

0% 001 N4NO V .-4 01ID V N N V N V In ION N

VV V in in in In In Il) U)

O OlIn 0% 040% - 0 In V 0% VN N mID ID 01.4 V V DNInInInlflInInU)OlD ON '0 U) 00% NN 0% In ID NO 0% V ID ID

V V V Ia) in U) In In U) In

.4 V V 0% 04 V 0101.4 NIDOl N V 11% 04.40% 0)1010

V V V U) U) Ill U) In In U)

0% V ID V ID .-4 00 In0ID ID 000 ,-l 0% V 11% N 0%

In I1% In In UI U)

0% V V 0% U) 0% 10 4 ID V ,-4 010 ID ,-I 010% 010040% V 0%

rn V in U) In In U) UI

V ID 100% V 0 In ID NONID ID4 01 ill N 0% 4 N ID NiI)U)U)In0% V N 0% InON N V 0% 0%D ID ,-4 N 0% ID 00 V In

U) In In In

.4.4 N .4 N N 0% N 010ID 10.4 ID '-40110 N V

V 00101 '0 '0 '0 V V 11% In

.4010 N OlIn 010% '010ID ID N V 0.4 V In 100VVInIn.40% 000)10 '010 In In ID U)ID In '-I N VOle N 4% V N

VN .4 VU) In 0'. 0% U) QI N 01In 0)0 U) 01 N N 0104N '0

NN 4 .4 ON 0'. 0% V ON 00) N 0% ON 0) N V 0) 1)0%

0 01 0)

.4 N 0) '0 In ID N 010.-I 04,-I 4 ,-4

.-4 ID0%VIn0%in.4VV In O)0%V 010In • •04 in In In In In In If) In U) 04

IO0%VNInN0%C4-44 InID0II0NInVVIDOVO •0('4 InU)InInInInInIflII)InIOIO N

•4 0ID0In.oOVIn • In.4 InInIOInInU)U)IflInInIOIO .-4

O)O%N0%Inmi0InO1In04M NIO010%0%NiflInION0%VO •0'-4 InInLninU)IniniflIflInInID I-I

>4 IONO%O0%InInInIONVIn InInU%IDU)InInInIflU)InIO In

>4 ION0100InVVIDNVV InInInIDinInIninIfliflIflIO o

>4 InNO%00LflVVIOIDIn0) iflInIn%DIDIninInIflIf%InID 0)

InIONNVONIO0)Ln0%>4 InIDO.-40%i1%m'0IOIOinN U1U)LflIOIDIflInIflInInInID €4

In,-4OIVNVID,-4,-401001>4 VInID0%,40%VVVNIO.4 IDInISIInInUIInID '-I

o 0)InOOInNID0%VONO)Z0% U)inU)in10If)InInLflInInlD 01

O NVNO0%V,-liflNIDZ I-40)VOICVVVIDNLflID InInIninIDInInInU)LflInID 10

o .4In.-l0IDOIO0)O%E 4NVNOID0%mVIONVIl) IDL1%InU)InInI0 U)

o V.4V04.-4OO%0)VNO% 0E ON0%ID0%InmN0%IDIDeV) EV )If)InIninInU)U')InID VO InNIVN0)%DW0%O.4E 0%4NIDVNNNInIDe')0% '0i1%InInLfli$)Inif)U)LflInID 0)

o V.4-4IDV0%VVIDNVz O'-4InN0),-INNV0NN VU%LflI1)InInInInInU)InID €4

O VO'.VVNOIOInOVNVx NNOImU)("4.-40C'40)O,-4I-i VVVU)InInInInLnU)InIO .-4

o VNNIDONN001InNZ IDminoNNO.ID.4Nm0'.O V'0Lninl1)'0'0inIn'0u) 0

'-4 N 0) V In ION 0'. 0.4 N1-4 .-4 .-4

ID0''—4

0%ID0%'-I

NO.-4 .4

U) N ID 0% 0% 0) '00 '00100)U) V '001100)01 l NV 1010

ID 101010 10101010

0 .—4 ON 0)101001 N -4 UIIt) V U) 0)10 V 0)0110 In 0101

10101010 ID 10 '0 ID ID 10

VN NV N01 '000101010)0% V 0% In InV If) 04 NO 0%

1010101010101010 N lOP- N

VU) 010% 4) N V In 10 V NNV ("I 0)1001 01100 Cl NID O 10 ID ID ID ID ID N N N C-

N N .-4 0% In '00101001 N€4 In ("IV N .-4 00 N ("IV 0%%OI01010IOC-NNC-NNN

("I N 0)01 U) ,-*IO0.-4 In ID0)1001V 0% 0) N 0) U) —1010101010 N NN N N N

01011010 V 0% In 010001100% U) N 0% N V UI'00101 U) 0ID '0 ID '.0 ID N NO-N SN

.-4 — VON U) 0% ,-4 0 V VV In 0)01101010 If) V 011010101010 N N N N N N N

0110 V 0 ION V 010) ('40In U) 0% 04 U) 1010 U) N In 0 .-1010101010 NN N N N

— Ia) N In 0501 V 0110 NU) U) ("4,-I V NV 0% U) In 0 ("NNN

0' 0' 0400 N 11% In ,-4O ("4V V -4.4'00% 0) NV V 0%ID D ID ID ID ID NN N N N

,-I 0) N N 0% 01 In IllV 0) '-400% 0 ('4 —40)01

ID ID ID '0 N N N NN

0' 001,-400% 0' N 0. U) N V('40) 4 ON V 404 OlD—IDIDIDlD.DIDNNNNN

0 N 0. V U) ON 0) .0.-I,.-400%0'0O - 00 V 0101010 In 1010 N N N ION

In 01 V 4 N 10 14) U) 44) 0.0.0% D.-4 0.01 U) 00 ID 04

U) U) In 1011% ID 10 N N 1010 N

N V ID '-40)0% '0 V 001000)10 In In 40100'

U) Il) In IC N IDID it

.4 N 4") '0 in 10 N 010,-IN'-4

92

If)N

mN if) NO'.>4 ON mc',0 . .N '.010 10 '.010

N0OIn>4 NOWC'S InUIIn ..-I lD'.0'.DN 'DID

O.-40,-4000N000'.o>4 0NNInIDInC'1NOC'Sa— N'.DIDNNNtSNNNIOI0

>4 0l.-I,-411IID0me0%OoIn NNNNNNNNNID'.0'0

>4

O NNNNNN('-NIDI0UIUI

>4 01,-IONNIO0CINIDNC'S NNNNNNNN'.DIOInIn

NNIn,-4mr.40'UINo'.N>4 O'.ONNInN,-4'.CInmoN NNIbNNNNN1010UIIA

'00%InC'5UIN0%O0ID0I>4 01IDLAONN'.Dm,-l0'.,-4 N10%ONNNIOID'.DIDUI'0

o 00.-4ON10'.0a'0'.O0o'.Z 0.01'.DOmONInm..0% NIDIONNNICIO.OICUSO

o 100.m..-lO10NomUI.-4O0%UIC'SNIDIOLflN001

10 '.DIDNNlDlDI010IDIno

o N,-40'OsO10(-IDc',z OOLAN,-IIOUILI%0,-400'.UI NIONNID10IO10'.OUIO

o InNNO10UIN0'.NZ OOUI'4,4UIc.somooo'.O N10NNI0IO%D'.DIDUIO

o -4-moNmmIONoZ Ooooousomoa'ç,o'.('1 NIDNNID'.O'.D10It1UIO

o moo'.NInom,.-4ICNoZ O%0%mNOulmNN0'.N NIOIDID%01010ID10In0'0

O NmN10NNo10In14sNE 10100InUI.-4O%NUIONUI— NlOI0.OIDIDInIOUIInoo

o lI10O0%IO('4N,-4In0co0..-IIDNc.lOmQ..0000'.o IOIOIDIDI.DUIOIDLIIUIOC',

NC'S 0 UI ID N 0' 0.-IN— — ,-4

0N0%'-4

>4 O'.mo".0N (AIDIn

ONLA N>4 om —141— '.010(F) ID

NNNIAI'SmUIONIOUIN>4 N-IIn0.m0'a'Q'.o— 1010In'.01010'.0%DInInInit,

>4 0-4O,-40%0%0%NOIA IAIAIn'.0101010InInInInIn

C'IOICONOIONN0.00',>4

O U'%LAOIAInIDIOUIIAUlLAUI

('sNm'.oNooa'.No>4 NNIn10N'..0NNN,-4N0C'S In00InInI0IOInI(5InUIL(I

mC'IIn00%O%OLA.-IIDNO>4 NN,-4NO0InLflUIOON 000UIIAIOID1(suIUILIso

OIn.40N10..ONNOUIO'.>4 NNIDO%00NNIONN— mmOO%D.0UIIn0'0O

o 010'0.-IInmNNN0'm,.-4Z -4l0NUIN0.-4,-40lD.-40'. mm0uI0'.J1UIo'0oo In,-I0%m,-4NNN0%OmN10I0mUIoO'.oInO'.0 omooInusooo.0o UIIO01000'.'0'00%NIn0'NUIIDNONI0I0O00IUI OLflLflO'0Ou.5o NO0'.0OLI1NNO0',NE NUIIO,-4O,-IO.0NOm0 c'smooLnU1ooo'0mo Nm0oN.—IoInNIn0C's ommOoInIn0000mo InUIONIISmmNNO'.cO,-4E .-ImUI01m,-4mOUINNUIN mu.1LnLnoo'0'0mo .-IIn104OuI'-i10NIO'.omE .-4NONO\NO.oO9m'-4 ommc'Ioou1ooomo 0'm'.00'NNWO'.UII0NOE O'.,-4000'.LANOInUIIDNO rnmmmm000ornmm

.-i Nm o UI ID N O%0,-4C'4.-4 — _-I

-4N0'-I

o '.0 N -4000.4 ('.4 In '0,-ION N ('1 .-4 ('40141100 m 0

'.D ID 10 '.0 10 '.D 10 '.0 '.0 '.0

'.0 omm ON ('50.00' ('4N 100 N 010' .-4 0.-40'.-4US In ID In UI In UI 101010 In ID

100.0 N 10 10 ('40 u.s NIn 00'. N UI 0000.0In USIA US LII UI In '.0 '.0 .0 '4110

0'. m In'0 ('40'. N 0N N N US m 10 N 010' 0IA In USIA In USIA In UI UI 11110

0(45 N In 0 '0 N '0.-I NOIn .400(4510 100'.

O In In 111111 In In UI 111 In In

000'0 N N m '.00'.— NO In In — ('5100 ('1 In0OIn 0 110 UI UIIn In In

0' In 0 C'S ('40 '.010100' 0 N ('500 N ('4 In mm UI('S000'OLIIOUIUSInIAUI

N N ('4,-40% m N .40.10N 0,-4C9 1000 m .-4 NO('S m 0 00 OOUS In ill iA in

('4(45 .-40%,-I IS In N 010 ('SION moo IAN .-4.-40mm o 00000111 IA In In

u.S 01010 LA -4 0 UI N 00UI lD,'-4 0'. m 0 ID 01010 mmm o mm o oooo In In

(45 N ,-4,-410. ON ('4 Nm 00'. '.0 ,-4 UI N 0. Nmm C'S mm o m 0000 UI

0'. 00 N — ON ('4 In mN N 100100% N 014110 0mm mm mm m 0000141

N In ('4 0 ID In '00% mON ON UI ID 141 ('1 UI NO'.(45(4 ('5 mm ("cc', 00000

,-I10 (40 ,-40'.,-4 N In 0' N N0mm mm m ON '00 N('4 mm mm mm 00000

— N 0 UI 10 N 0%0,-4tN— — —

00. OIOmNOONO'.0 'DNNNNNNNLA 00,-4 ID — In ('S ,-4-I'.D 0''0101010 0'. N '-40. (40 -I'010101010 ION N '.0 N N N

UI In N 0.00100.010%0 In LII LAN N ON N N'.01010101010 N N '.0101010

ON In -'.10100% 10 NON C'SC'S '010 '.0 '.0 N In 010 1010'.0 '.0 10 '.0 D '.0 N N '.OIDIDIO

00 .-I ('50.0 m c' — ONC'S'01010'0 N N .-'. '.00% N '.0

'.0 '.0 10 10 '.0 IDNN '.0 '.0 '0 '.0

In '.0,-4 NON 00'. (45 NOC'S ION 1010 '.00% N N

'.0 10 10 '.0 ID '.0 N N '.0 '.D '.0 10

10.-I 0'. 14500 N .-4 m ("4(45 NNION ID 0,-IN 0'. N NID '.0 ID '.0 '.D NN ID '.0 '.0 10

'.00110 ('4 In 0110 N u.s —O (450'. NO 111100 In 010,-I'0 ID '.0 '.0 N N N '.0 N N

N 010'. m 010'. N N N Ill 00ON 0', N ON N ION C'S C',

'0 '.0 '.0 10 NNNNN N

N 10,-4 0 UI 0 0% Na'01,-I 0'.10.-4 ION N '.00 UI11110 IDNNNNN() — N 0010 NON ('S N m0% — 10.-IN In N 1010010111101010 N N N N N

Nm o In N 010% N 0'. InoON In 010(41000 '0In '.0 10 ION NNNN N

NO (4)01-4 ('410 UI ,-4 ('4 NNO'. it'. me'0 m '.D N OlD IllIflUI10I0NNNNNNNm ON 10100'.,-'. 010 C'S InION m ,.-i In m N 0(4500In UI '.0'.0 ID N N N N N

00' ,-4 m rS 010 m In 010'01(5001110 N 10.-4,-40NInLAI010IDNNNNNN 0'. N UI 0010,-I 0% 0 In ('S('ION ale m 01 In N 0III In UI In 10 N N N ION N

.-4 NC'S 0 In '.0 N 0'. 0,-I N-4 -4

0('4

x>4In

>40'-I

>4

In

>4

0

>4

'.5

xN

x>4

'-4

0%

ID

UI

'0

m

N

'-4

0

InN

0N

0-4

'IS

0

u.s

N

'-I

0%

'.0

In

0

('4

"-4

0

mN0%-4

93

InN

0N

0-4

In

N

-4

0

ID

In

N

-4

0

m c.i N ID N ID In NOIn Q NO% 0 0% '-4 00% '-4N N N N N N

N .-4 ID ,-40 '-4 .-4 .-I In 0% 0In 0% 0% ,-4 -4

N N N N N N S N S

0010% 01 m ...e N01 In 0 m00 0 N N N 'NN N N N N SN N NN In In m 01-4 000 N NN In 0 N 0% U% 1

1010 N N NNr-NID -401 In N 01 N In 0%N ID 001 N 1 0% N

SrIn 00% ID N In .-4 ,-4 In ,.4 44N 10-40% N In 9 NID ID N N NNN( ID N N 0% N 10 In N IDN N NO 4 0 N 0% N N -4ID ID N NNNNN 0% 0% In 01 ID 400% N 0% N ID ID N N In

OlIn ID In N 4 N 00% 10N .-4 ON ID 100% N 10-4N N 0' 555N ID 0% '-4 '-4 ID ') ON l 4)

In 0% In N N N NO NNN0%NNNIn N N 0% -40% ,-4 10In In In 0' 000 In m 0 NNN0%NNN

ID ,-4 10 10 0' N ) IDID In 'C 0 '0 N .-4 N NNN0%NN0%IDNNID 0' '0-40% 10 '0 ON 0% ')ID Lfl ID ON SN m e' 010 -NN0%NN0%%0NNN In In 0 In 0% '0%') NON 01010100% 0% 105 N ON In 0SN 5550% 10 N SN

N .4 'DIn In 010 N In SIn Ui Ui N 1010 Ui NO ,-4 NN N SN 50% 10 N S ID

10 -4 N Ui Ui '0SIDS '0,40 In In 'DO .4 N () '0Ifi

NN NNNO' 10105 '0(N 'i 0 Ui ID S 0' 0.4 N

-4

m '0'0OI N>4 .4 0 N'0InIll .IN

>4 NIDmN,-4O,In,.O0N0"-4IDInIDN')OION

>4In-4 NNNNX 0%'iNNeOINn,NO>4 0%.-40%O0('0%CWO.4 NNNNNNr_>4

In NNNNNNN5NN()UiN0%.4(',-40ID.4IDN>4 .4N'0IDO%O(N'0'CN

'0 NIDNNNNNNNN

>4 0%InONN'0-4N.0Om %DI0NNNNNNNNN

.4ID00%InOe'iID>4 NNNUiNID5O0%NION %DIDIDNIDNNNIDNID

InNIDIOSIflIDIDNQ%NO>4

,.4 IDLflIOID%D%DNNNIIDI

o InIn,-4UiN(.i'0'DN0%Z OOUi'C'0O0% IDLflIOIOII4%DNNNIDIDIn

o NNI0OONOi'DUieZ ONNIDe0%ON,-IDID LflIDIIiIDIONNInIDUI

o OCir4.4O10UimO%,-4Nz ONOUiNSOI0OInIn LnIOUi10IO'DNLn10Ln

o '0NIOOOID'0,-1'iIØNZ 0%IDN'0,-4LnNOIN'0

o '0Lt%IDO'0'0O0%Pi0%Z Lfl'D10(NOmIn10'D'C(N4') InInInInL$)'D'D10'DInInIn

o Nmc.i'0N0'CIOz 'D'0-ic''0.,-(N UiInInInInInIDIO'DUiUiUi

o 4'Nui'05O,-410m10Z UiO%N,-40%IDO0%m,4O"-4

o N10N100%5c'iuiNO In'0'0InLflLIi'DLn

.4 N 4') '011) lOS 010-4 (N.4-4-4

'00% Ui 4') N ('4 ('I ('40CM010 0 '-400 4')

NSNNNNNNN ID N N '0 '0 N ('4'0 CM'0 In010% 0010% N S IDS SN N N('-SS N SN

0100% 0' 0% .'4 In %0 ID 100% C')1010 ID In '00

N N SN NN NNSN N ID

U) 4') 4') 010% .4-44').4 .-4'00 4') N ID'0 4') -40N0 4NNNNNNNNSIDID4')0% -401000000% 0N CM-4 .4 ID 4') N 01 U) 50%

rNN N SN N 101010 In U)

001 ID N SN '0 N ID ID0100. '0-40% ID In ('4101010510 ID N N 101010 ID In In

0110 1(1 In .4014') N CM 0' In4') 510 '4) ('40110 '0NO 4') 4')

ID ID 101051010101010 In IS)

10.-I '0-41010 4') ID '0It4.-400 ID 4') 0 N '0 W10 Ii) In 11) '0 '0

4') 00 '00 IS) U) In 0105 N4') 0% N '0.4 ID In I') NNIn U) In U) ID ID 11)44)4)11) '0'0

.4-404') '-4 e') U) 050 4')

.410 Ui '0.40110 .1) '0 ('41010

U)UiInInInIS)U)In'0'0

('45N,-4100 4) ,-4 U) 4') 0%0 '0 C') 4') 0' U) '04') '-410 IS)

N 0% ('4 N 4') . ID 4')(N ('4-4510 0 ('4 NO In U)

In U) U) I1)I1)In'0 '0

010 t) 0' .-4 N C'-N .400110'0 ('4 .4.-4Q% U)'0

4') 010% 0 10100101 In 10'0ID '04') ,-l 09 C')

'00'0 IS) IS) U) U) It) '0 '0'0

N N 0 '000 N '0-410It) ION N CM 4') .40% ON '0 N

Lfl U) U) 0 U) '0

N %DNID0ID0% 04') 0' SCM11) '0 N (NO '-4104') ('4

CM 4') '0 U) IDS 010-4CM.4-4 _-4

0 In 0110 0' 4') ON 4')>4 0NS ID N0%OIS)N 5554'- N N5N

mIDNNIDIDC'),.l10ID,,ul>4 NNNInID'C'C0%0('4 NNNNNNNNN555N0'N'0InInNI')N,.4ID>4 10IDIDIDID.0In"0InSNIn'-4 NNNNNNNNNNNN

>4o.-4 NNNNNNNNNNNr-.

>4

In 'DIDIDIDIDIOIDIDNNNN

CMID0%O'0C')'0N.0N>4 NNIDNIDInN01(..(N'0 1OIDDIDIDIOID%DIDNNN

10I0O0%C')U)In01U)>4

4') 10%DID%DIDIDIDIDIDNNN

N,-4NNI')%D.--4NInONO>4 OO'-40'0'ONNO,..IN IDIDInIDID%DIDIØIDNNN

Z>4 InIn'0IS)NIDOCNIn00'.4 InInInIflInL1)IDIDIDNID%O

o '00N0InNNNIDZ m-lC')InU)o%,-.4'0N0% InU)U)InInU)UDIDID

o InN,-4InN0%,Z ,4OO0O4')NuID InIn'0InUiUiUi'CID'Co ID0%4')LnONN'0InuiE O0%0C')C')N0%NID'0u)In In'0'0'0InInInLS)IDIDID

o InOInIn.-4i"iO(N0%InID'CE O'O1NINNID-4Inm'0'0 '0'0'0'0IS)LflU)UiIDIDID%C

o 'ILflN,-4NIS)'0N0%NE NI0S4'D9(.,,..(.)4')

o IDInLfl0,-4'C0'In'CNE 'CIDInInO0C')'0,-I0'1.4(N '0'0'0U)U)InUiInIOInI0

o NN4'0O'CIn'0,.4NNE InInIn'00'0'NmN.4'D4 '0'0'0'0'0'0InU)InIOInU)

o .4 '04') 0 10 It) U) 5 ,-4E In'0Inm0%,-')5(Nr0 '0'0'0'0'0'0InUiUiIDUiIn

.4 N (")'0 U) ION 0%0,-4 N_4_4 _4

NN0%-4

>4InCM

>40CM

>4I')-I

>40'-4

>4

U)

'0

>4

>4

N

>'-4

0'

I0

It)

'0

C')

N

'-4

0z0

'.04'-.0%'-4

'0N0%'-I

InN0%'-4

914

N0%-4

o N 0% N 0010-4 N -4 mN N N mm 4 mm '.0 in

10 N 0% ID em 100% 0% _i in.-4 N N m in in N 10 In .0

010% N 10 N m .-4 in 0 in00 _i N In N In 10 ION

_i N NON In 0% em 10 N 10ale _i N m N 1010

_i 10 in mm _i 0100% mm ,'.DN 0% .-4 m N N N ID 0N N N N 0%

N 0% m _i 0% mOb ID mIn ID 00 m N m 00 mN N Nb 01010101010% 010%

01 0101100% 1010 in in In 0%N In N 0% N N N N N NN N N Nb 010101010% 010%

N ION N 0100010 '.010 m-4 _i f in N N N m '.0 ION mNb N Nb 01010101010% 0

_iN 0101 N N 0001 0% N00_i m ID 00 N ID Nb ('INNNbr-01001Ol0.O

-4ID N N 0110 In N in 10001100101010mb '00110 In In1010 0 N Nb N N 01010% 0%

N _i Nmi m 0% _i in NNO 010% N ID 01 in N10 10 ID ID Nb Nh O% 010%

b 0110 N 01 m 010% _i 01010b NO Nb 0mbIC ID ID ID N N N N0% 0'

m '.o N _i _i in_i0 N 0 binin In 10 in 01 NON m a. ,-I in1010101010 N N N 01010101

In_i_i In 0 In 1010 in 10 Inm in N 100%Nb 01 ON10101010101010 b 01010% 0%

bm ID m .-i in010 In 0% %'ON in 0 10 010100% 10

0% 01 ' ID m 0 In Olin in100% 0101 C', N '00 01 m 0110

ID D '.0 ON 0101 N

'0Lfl ION 01010_i N-4_i_i

ininomo'a'mommo_i>4 In%0ID0101InIOI0IDInInInN 01001010101010101010101

m 010% In 0 N 10001 '00101ION N 010110 b 01 N 10 N

N NON N ('1 0110_i_i NN 010%_i 011001 0._i 4% -400101 0% 01 0101000-4_i_iN 010% 01100 ON b_i m%00 01 Or'- in N _i Cl N NO01 010101001 0101000

— _i_i01 N m 01 NO'0IOOIO '0 ('1'.00010 bin N .-. m 01 N_i

0101000-4 -4 -4

NON '0_iNN 01 In m '00%01 Cl_i N Olin 01 Cl'0 Cl mm0101 0101001 0.0%_iO0

-4 -4 -4

'001 01mb_i C'S 100101010m in '0 ION NO UI bin 0101Ol00%0%0'q'_ioo

_4 -4_i

ON b.-l 4 _i 01-i 01 '0 %1 _i0% N GIlD_i '0 N '00 '0

0101010101010100 N_i_i-4 _i -4 -4_i

1001001 Nm o ON 01 C'S CII010010% N N .0 4'I in ION 01010 010041_i_i

-4 -4_i_i_i_im 0mm 01 in m 0% N 4'% meNO' 0101 N Cl '0 _i in NO m010101010101010041 Cl N

-4 -4 -4 _4_i

01 N 000% '00100141 In InID 010101 N N In_i o N 01 m

OlOl0%0%0lO0h_iCII-4 _i _4_i_i

mm 10100mm Cl ON I'ION N 0101 N In 001001 C'

0041.-IN-4_i -4 _4_i

o IDN-401001DN0.01N01

4I OlOl0%Ol010%0'OON.-4C1-4_i_i—_i

o OlIninN01C',NO'0Ln0%a: 0101010N.-lCl_im.-l,001N GOl0'OlOl0%O'OOCII_i_i-4_i -l -4_io 4m0101'0.-..N-N

in ID In in N 0% .-l Nb N N,-4 Ol0'O%01O%010%OO.-.-iO

-4 _i -4 -4 _i

o 00000C'%N010%0-NID In in in 01 N (101014141-4

O Ol01O'010'0%01Ob'-4 _i

0%N0%-4

_i C'i C'I'0 in 10 N 01010-I N—-4 -4

N 4', NON in 0010 ON IAIn 0% 01 0-410 '00 0' Cl 010000000100 _i.40-4_i_i_i_i _i,-I_i_i.-4_4ION 0 m o 0110 In N o -4NO Nb '00 01 0 In 01010c _i _i_i .- - _i_i _i _i_i_i _i.-l _i _i_i_i _i01mb IA 0110001 C'% 0' N0101-4010 NON N _i_i 0%0 $ N 000 .-l _i _i N 04_i-4_i_i_i_i_i_i_i_i_i -4 -4

Z O01N_i_iC'.iOOInoi'.o'0>40_4 ONN0000_i_i_iN,-i

-4_i _4 -4_i _4 -4 _i -4_i_i_iX N_i0100N,.IO01q0100>4 NOlm'0N'0_i'0InmIO,in Or4NOQ'.GIC_i,.INNN_i_i_i_i -4_i_i_i__i

'0Ol_iOOlIOInmN,ONIO>4 N_iN'00C10%'0i0001_i'0 O('%C'400%0%01.-4_iClCl-4_i_i_i -4_i_i_i_i

oinIno'0Olbro1010r-.>4 OlIONm_iONN.oin_iNm or',moo0%0%_i_ir.m_i_i_i_i -4_i_i_i_ix _iIOinOC-4IntA10001'0N>4 C'lO_iInOOlION011010In('4 -4_i_i_i -4_i_i_i_i

>4 N0%N01100_i_i01_i'0(1_i _iOIn001010%_i_im'0m_i_i_4_i -4_i_i_i_io InNInNC'4'0_iOIONC-40a: O01010iflmOal0%N01O01 Nin,-4O01O_im.m-4_i_i_i -4_i_i_i_io mmN'0'.Dom_iInNmo'0NOlOmNOlb0.ClOo10 NOIn01o_imoo-4_i_i_i -4_i_i_i_iO m'0minInm000%_iIn01

'0in In 0'. N -4011001 N N IAIA NIn0O01O.-4'%'0'0-4_i_i_i -4I_i-4_io 0'NOlN0.-lbO0%'0binE mm,401..l,-401InN_iNN'0

-4_i_i_i -4_i_i__iO 10N'001_i10010mOlIDIDa:m NInON0101O_iN'0'0-4_i_i_i -4_i_i_i_io 01100CMIO'0NIDNOOOa: ('4N'0LflN0%inO'0In0(1('4 N'0LI%ONNO01.-IN'0'0-4_i_i_i -4_i_i_io NNN0%In100001001oE O01OmO'.DOr'iNIDmOl_i _i',In0N%DN0%,4_i'0r'4-4_i_i_i -4_i_i_io m01OlONOlr'-N01mmoa: 0N419'0N0'CIIIOo .4mmO'0%D01oorno-4_i_i_i -4_i__i

.-4 Nm '0 in ID b 01 0' 0 -4 CM-4—_i

0010'-4

>4 GI'0_iID0InID10'0NInN 0_i_i.-4_i.-4_iNc.iCl_i_i-4_i_i -4 -4_i_i_i_i_i_i -4ONICO_iOmID_i.-401N>4 NN410mIO0101'0IDInNO

N _ir'41'4NCIINNm'0mNm-4-4-4-4-4_i_i_i_i_i_i_iN01Ifl01O_iN01'0_i01O>4 _iNIOmO_iIDm_iNNNIn_i NNNm('4mm'0In'0NC',-4_i_i_i -4 -4_i -4 -4 _4 -4 -4Ol_i0100%0%0%,-410_iInN

>4o_i NNNmC'4mm'0in'0Nm-4_i_i -4 _4_i -4 -4 -4_i_i_i

>4 mc400lmN'.D'0NmNNin NmmmmmuInin'0Clm

_i ..4 _i ,..4 _i 4_i ,..4 _i _i_i001000NOOininalN>4 '0mOOIAOoIIDOmInN

'0 NmInin'0NC',-4_i_i_i_i_i_i_i_i_i_i -4

>4 O'0Ol_i01O%01001NInInC'I NC'I41'0mc-IInInIn'0Nm-4_i_i_i_i_i_i_i_i_i_i_i

ON_iOIn0%min_i'.om>4 OInOm_i_im_i_i_iNinN 4100 InIDIDOCMm

-4_i_i_i_i -4 -4 -4 -4 -4_i_iInOION0INIAIn_iOIDm>4 1000110ID'0NmOI_iInm

_i rnc4'0'0'0'0n'0_i',-4 -4 -4 -4 -4 -4 _i -4 -4_i_i_i

o 0_i'0NIDIOOmNN_iina: _imNOOIn01100_imo01

-4_i -4 -4 -4_i_i_i_i -4-4-4o 0%DIAIOInIOO.-IinOIIOIna: InIn%O0IO%IOOIAIDIDOinID O'0N000InIDinrn_ic.,I

-4_i_i_i -4 -4_i_i_i -4 -4 -4o O%N100%DInNC'IInN'0OE IDID100010IOmm'001r',In 00NInInOIflIOIn(',

-4_i_i_i -4 -4 -4 -4 -4_i_i -4o IDONO.41flNIn'0_i01ba: OION_imIOin,.4ONN0%'0

-4 _i_i -4 -4 -4 -4 -4 -4 -4 _i -4

o NON0NIn_iO,Df',Na: 0N01NID'.000N,-.tflinm inN'0U10'0t%4..4

-4 _i_i -4 -4-4_i -4 -4 -4_i -4o IO10'001_i_iN'001OlNa.a: OIn0101OIDNOmomr-N in'0N%D'0tAIn'0r',O0-4-4-4-4_i_i_i_i_i_i_i_i0 NOm_i_i_i01001N'0_iE NNO_iNq'0%IDIOOlb_i '0'0m'01000InmNoG

-4 -4 _4 -4 -4_i_i__i_i_io bOONN4',Oo'0Ina: 0%0_iinN0in'0IDr-_iNO 'mNI0'0'0inNNO

-4 -4 -4 -4 -4 -4 -4 -4_i_i_i_i CM ('10 In ION 01010_i N

-4-4—

-4010.-4

>4inCII

x>4 >4O 0CII N

N01010_iOc'i01_iC',1010Di IDOl0100IbOO_i00al >4U, in.-4 0100101010101010100101 '-4

-4

0-4

In

0.01010 NO_ibm 010.U ID N oo m N ('4 ('401100

'0

0('4

0-4

In

'0

m

N

-4

0.

ID

In

'0

m

N

-4

0

N

-4

0'.

ID

In

'0

95

OinoNa%_I.-rr--4t.,I Nr'I.4,-4ooo

-4_I .4.1 .-4 .4 .-4 .4 ,-4 .4.4.4x L(%LAOLn0%0%>4 _INIOr40',-4N0 .N

.4.4.4.4.4.4.4 .-I .4.4 ..4 .4

>4In.-I N.-IOO0.4.4.4.4.4.4.4.4 ,-I .-4 .4.4

>4 C4NO%Iflo.4

.-4 ,-4 _l I — -44 .4.4.4.4 -4or-Inr-InIn,-4m

>4

In.4 _4 .4.4.4.4— ,-l .4 ,-I -4O.-4ONr-

>4 01D'DuI.-I4OOO

_I ,-4 4 — .-I -4 .-4 .4.4 -4 4 .4

>4 01O0mnQ%m CCO%.4.4.4.4.4— .4 4 .4 _4 4D>4

N 4C004 4 p.4.4 4 4 .4.4x _INNNIflNO%>4-

-4 .-4 ,-4 .4.4— _4 -4 .-4

o 0C4NU1In

0%.-4 - .4.1 -4 .4.4.4

o _INMMN0E0 0a

.-4 .4 .l .4 .-I .-I.4o 0'N00%0OOM.4.iEIn

,-4 _l .14 4 .-4.4o -40%N00%,-4InLfl

.-I.-4 ,-I .4 - .4_Io LflN0c,r-.-4Q%N£ 0NN-_Ie-.-C.,

_4_4 _4 .4.4_I_Io _INNNrC..-4,-4£ InNNC40-N N.-4CNO'r-NN

.1.-. - ,-4 .4.4o UI,-40NNNz UNO.4 C-4_I0%N0NN

,-1 - .4 .-4 .4

o N_I .-4NulE N,c0'0oo _ION0bN_4.4__I 4 .-I.4 N C.) In 0 N 0' 0.4 N

.4-4 .-4

z>4IAN .-4000000..4..4.4.4 _I — 4.4.-4,-4..4_I..4_I,.4O4NUl>4 InOOOinoN

.1.4.4.4.4.4.4.4.4Z>4In-4 0OO.-44.-4.1.-4.-4,-i,.-.4.4.1.4-4.4-4.4.4-4.4.4

>40.-4 .40000.4_I.-4,-4..4,-4,-4.1-4-4-4.4-4.4.4.4.4.4.4

>4 Me'U, OG00lOQ.-4..4..4.-44.4 _4 .4.1.4.4.4.4.4.4z NIn'0NO0In>4

00lOOlO0.-4.-4O,.l,.4.I-4 -4 -4.1-4-4.4.4-4.4

>4 ON0'0.1.-C.-O%M OO'O0100.-4,-40000

—-4-4.4.4.4-4-4

>4

N 010%010'0%00.-40000.4-4.4.4-4.4.4

>4

-4 000101010%

o£ 0MO.-4'mnNa' ooo N_IIfl_I.4C'IN0z O.0ON0N.1mni0 0%00%0%

ID M .4 0 0 (NO In ,-4N 0 M O,-4 N .-4O N0' 010' 0% 010' 0% 0% 0%

N In UI C.) 0% 0 In N 0% ('10' 0.0 0 100 .-4 C.500000%

IDID ID C.') C.') '0 (N N NONO (N 0% 0 0% NO N0'0'00%

In '0 (N 0' .4.4 N In ( N.4Q' N .4 N ('4 (N U')

N 0' 0'

N .4 D 010 N U') .4 N 0% C.')00' LI') - UI I4') 0' ON UI'0

o .4010NIDN£O

.4 C.) '0 U') ID N 0% 0.4 N.1-4_I

0' 0%-4 -4

U,N .-4,-4NNMMNN,.4.-4,.4_I.4.4.1.4-4.4-4-4-4.4.4.4

>4 '0Oc'DNf)0N p-IC'4NNMMC'4N..4.-4..4..4.4.4-4.4.4.4-4-4-4-4.4-4

>4U,.4 .-4.-4N(NM('I(NNN,-I..I.4.-44..4-4_I444_I-4-4

MON NU')IflM'0>4 Ifl%D4Ifl..00.4 r4.-4NNMMNe'4N.-4_I..4

z>4

U,

Z>4 O0%.-IMCfl',-4N0%N

-4.-4.-4NC.)MC%aC.4N.-sOO.-I-4I.4.4.44_I_I-4_I-4

>4 '00It)C'4OMN0%,-4'0.0C.)

.4.4.4.4.4-4.1.4.4-4-4-4X>4

N 00.-4.-INNNC'4.-I000%.4.1-4.1-4.1.4-4-4-4-4

>4 ID0%I00%01%0I0.-I,4LI')_I.-4 0%0O0.-4,-4_I.-4,-4O0%0%-4-4-4.1-4_I_I_Io 0%In0%ON_I.0M0%_I

Ci O0.'4_I-4O0%0.u,——-4-4-4——

o NN'4mM10iflQsN£ NMO0%0_IND0%e4'0 010%OO.-4O_I_IO0%-4-4_I_I -4 I.4o UIONN0%M,-N0%NU,£ N%0NNNN0'O'0iflNLI') O00O0.-4O0%

————.1—_Io 4NLfl%DNI0(NO'0N0'E IIn .4.4 '0 N 0% 10 N40 010%OOOOO0O0%—.1-4.4.4-4_Io ONN10'0N(NOUINME -4M0%01O.-4UIU')N'0..IC.') 01010%01OOOO000__4_4-4 -4o NN0%IID.0ID.0E O.-1NN10MN'00%N O%0%0'0'0%0%O0ON

-4——o O0%'00%m'0'0NOMz 0%Lfl'0UIMON'0UIOC.4-4 00%0%0%00ON—-I-I

.4 (NM UI ID N 0' 0-4 N-4-4—

MC.- Cl ID N In N N N 0% U)ID C.) .00 C.) In NM

0.4000O0.Oi0._4_4_I _4 -4 -4_I_I_IW .40.4,-I N N .4 N.401 N 01.4 0 C.).-4.4.-4.-400.-40 00001

.4MM .-4 ('4 U) 0% ON MN 0MeNlO ID 010% ON ,4 ID.-4N.-4.-I0000.4000%.4-4-4-4-4-4.4-4.4_I_I('4100.0 C.) 01 In M ID 0% N.4 In M 0 In 0.4.4.4.40000000% 0%-4.4.4-4.4.4.1.4.4-4

0 W 0 '0 In NO U) M lON U) ('401 .-I In N U)

0.4.40010100101010%r4_4,-4,-4 ,-4

C') MO iD_I ('4 N 04In NO '0 Il) 11)0 In In ('401 C.)

0O0%0i0%010i0i.-4__ -4NM .1.0.4 C.) 100 In ('4('401 ION M.-4 If) C.) NO ION

000001010101010%———-4101000% 0 NN MN .00% N U) C.)01000% NI.4

In N .4 In Qi MN ,-4 InCi '0 010 N 0010 N In010% N N N N N

MID '0N N U) %DO U') ('INM IlO D'0 N N '0010% N N N N N N N N

MN 0% '0 U') .-4 '00.-I C.) ('4 '011') I!) '0 C')

01NNNNNNNNIn 0 NO MOO'0 '0.-lIDMON C.) '0 N '0'0 C.) N

0% N N N N N N N N

0 ND0%M,-4'0.-l0%OME'0 NNNNNNNN

0% .4 .-4 N MON '0111 N.4 N U') 0% MO'0 N N MM_INNNNNNNNNC') N N'000 LI) (N 010 '00O 4') C.) N N 0% N (N .-4 1') .40'

C) N '0NO N ID'0.1 '001'0N '00 In ON ON .-44 It') NN N N N O N N N NIDID

.4 NC.') 0 U) ID N 0% 0.-I N.4-4-4

Ii)-4

0-4

In

'0

M

('4

-I

0%

ID

U')

.4

C.)

N

N0%-4

o NInNIDIJ')C.5NNNNO 0 0%NN'00%01NIn(NtnID0%£ NND'0NUIC'0M Z NNIDN'0'0e")O0%ID,4O 0%C'OONID 0 NIONNIOIDIDNNIDIr)U')

-4 -4

LI)

0%-4

96

ONO>4 01N01 en-4en >4 OOiIn . InN N N

-4mN01enenN>4 O-4t-enu0 >4 (PIN0 • 0N N

N .-I en 010 N 01 en ,-4 en>4 NUI0N01NNCnN00 >4 0%If, . If,1-4 .-l NN

Nen U10en InN>4 N .-4 N 10 en ION en >4 en N0 .001 N N N N N N N N N .-4 N N

IONr1NNNUen>4 loeenenoenN10NolN >4 NNIn NNNN10r-IOID10 In 1010

100.Or400,-4r-100o>4 01 N N 04 en .-4 In ON 10 N >4 1010

NNNNNN%0N%0%010 4010

011010NIOIn0n01.-4OO>4 -4N00N,-4NNIflON >4

en N N N N N 101010101010 en 1010

NInN,-40140O%'ee-IN IflOl>4 01400404eno01oN.-4en >4N NNIO10N.OIOII11010IO10 N 1010

en N 04 N -4 N 0.0 In 04>4 enN IONO% >4.-4 NNIOI0101010InLflLflLflUI .-4 InIn

o 0O%100 4 en en 00 N -4 N 0 0 NN If, en N NO N In 10 N

0% N N 1010101010 In lt* In In Ifl 01 In In

o NNNOON.404fl..-4IDN 0N N en 10-401 C4 In In N N 10

ID NN flU1In 10 IflIn

o NON N it. N en en in 10 0 - 4Z N N N In .-I 01 N en In N N 0In NNID flIt.LflInIn In I$Iifl

o 010Nm0InNenc4,-4.-l 0E .-lNOr4In.-401Nmenu, Nit.NN%OIDIDIOInLnInInLnIn

o 10010 en .-4 N 0100 0 0 InZ .-4N C'4enen E Nit.en N N 101010 I In LI In in In In en in in

o ID N 040 ION It. 10 '0 N N 0 benE 0,-4'0.-4en0N.-,NNenIn lOinN r-N101010%DInInIninLnIn N U4In

o .l01en,-4010Oen'0Q 0E in01NOiflNOe4O Ino104010 In 1010 in '0in in in In ,-4 In It.

o enN%or-'0In10r- 0-lO%Orn.-l1000O %OlOlOinIOinIn0In0u1O 0 mm

1-4 ,-4 4

10 N0% 0%1-4

97

'.0W U N .-4 rn it. U 10-40WW 0' 0% 00% W W W 0% W

N (N N N C') (N N (N N N N N

ON 10,-I_I U) LII 4.-4W WOW 0'. 0% 0% W W NW 0%

(N N N N (N (N (N (N (N CNN N

NO'. N U) U 50.0W .-4 Win4') N N N_INN C') (N 4-) N N(N N N ('4 (N (N N (N N N N N

W U.-IN U N C') U) m W N N_I.-4_I00% 0.-4.4,-I.-I.40N N N N_I N N N (N N N N

NW m W (N UI U 0% W 4') UN00% 0% W NW 0' 0.0'. 00% W

(N-I_I .l_I.-I_I_I_IN .'.4.4

.04') 0'. '.00% (N 4') 0.4.0.0 inSN '.010 U '.0 NW W W N '.0

U U U'. It. in C') 0'. 10.-I U W.-lml') 4-'. 4') N 4') 4') U U) Ill U

.44') U '.00% U W U 04') 0% it.(N (N N N .-4 (N N m U U 4')

.4.4-4.4.4.4-4-4-4-4.4_I

NO U LAN 10 '.0.4W 0% W WN N (N

.-_I 1-4.4_I .—4r4.-4.-4

N '.00 NO ') (N '.0 U U It. '.0OO.-4'-I.-.-I_I.-IN N N (N,-4.-4.-4.-4 44I-4.4.4.41-4.4

W N '.00% WON .—4 00. .4 U0'. 000 0.-40,-4 (N .4 ('4 (NC .4_I.4,4.-4.-4.-4-4

4') SN U) '.0 N 4') '.014) U SN0% 0% 000000.4 .-4 .-4 N

C') W .-4 4') U W -4-40% 1') 0.W 0.0% 0000.O.-40.-4,-00.40_I4,-,-_I

04". 100._ION NW NO LA0% 040% 0% 000%000.4.40000.-I.40.-i.-4-4.-4.-4

U N OWN U) Nm 0. .40100000% 0'. 0% 0.40.4.40I.4'—4000.4_I.-I.-4.-4.-

_I(N 4-) U It. '.05W 0.0.4 NI-I 1-4.4

0iU U0. 0'1-4 -4

0% NON UN .4 U 0055W W 0'.0.0% W W SN(%4('.4 (N N N N N N CNN ('.4 N

N '.0 N C') U) LAO'. I') 0'N1000' W 0% 0% 0% 0'. W W SNNN

(N N N N ('4 N (N 4-IN CNN 4-4

.-40% U U U U W N%00(N _I_I.4.4000% 0' 0% 0.0.(N N N N N N (N .-I_I_I .4.-4

.-4.-4 U UI N 0.0%_I U U 0'.000% 0. 0'.NNNNSr-.o(N N .4-4.4.4.4-4.4_I_I -.

ON 4') 11)0' 4-4 r-4NW C WW SN '.014)14) U U in U U

4_I_I_I .4.4.4.4.4-4.4-4

N_I.-44') Nm .4 NO ('1 N Uit'. '.014114) U (N N N (N N 4-I N

1-4.4.4.4.4.4_I_I_I .4.4.-I

11)54') me U NO.00% ONm m m C.'. m .10.400.4.41-I_I 4.4rI4_I,4_I.4

000W SlOW 0% NO'. ('-1.-I4') 4') C') CNN -4000 O_I.-4.•4-4_I.-41-4_I_I-4_I,4.4_4

It. 4') N C') It. 0000-410 N('4 N N (N 4-IN .4.4_I.4.4.4-4-4-4-4.4-4-4-4 .4-4-4

U010_I LA_I .-4.-4 . C'-) W I')(N (N (N N N (N .4-4.4.4 r .4

NW LAOU mmN C.) U 0 UN ,4 N ('I (N N .4.4_I,-4 ('4.4

-4-4-4-4.4_I_I -4.4.4.4

.-4 U) U W U LA Urn U LAN 14)N_I (N .4 N (N .-4 .-4 .4_IN .4

4_I.4I.4.I_I_I.4_I,.0. mm N 4-'. ('-10 U '.ON U) in.4_IN .4 N (N _I.4_I_IN.4-4-41-1-4.4-4-4.4_I .41-4.-I

'.0 NW U NW 0.5.4.-lID N.4 _I.4-4.-I_I00.-.-4,-4.4

0110 rnOe.40IDW (N NI')0.40.40W W W W W 0'-4_I.-4.-4.-40000000.44-44') U U) '.05W 0.0.-Ii

—4.4—

>4U)N

>4U)('.4

>4141(4

>40N.0N >4• 0(N (N

0%0%00.UIU)O0.mO.0U U '.0 U U .4 W NON 0•.-.....•..N N (N ('44-4 (N.-4_I('4 .4 N

>40N

>4it.U)0 >4• it.4') .4

.4OmWONNO%_IWo'.U)000'.0'.OO.-I0000%W••••••.....mmCNc'ImmmrnmmNN

>4It..4

>40.4

4')14) >4

• 0('4 .4

N(NLt.'.4mO%0Wo-.m.4OInInInLflUIin.0%O10NWWN(N(NNN(NN(NCN(NNN

0.4

>4

UI

U0% >4

.4 LI'.

rnm,-4,-IWUS.-4.0SO_I0'0'0'0'W0'WWWo.-4rn >4

It.

>4

U>4

.4 U0'_I%OWUIOC')O'.r>Om.0NWNNNWN10'.OWO%.

.4_I 4.-4'-4(N>4

U

>4

4')

.0'.0 >4

.4 4')

.—I LA N U W U 0.014)4') U 010%0In%DUI10LAUU'.ONO-4I-4.-4_I4.-4_I.-.-4N

>4

C')

>4

(N

W C1') >4

.4 ('4

UNO0'In0.QLA0..0U.0'-)C.'.C'1(')4-lmUC'.4e9UihN

-4_I_I4,.-4.-4.4.-4Z>4

('4

>4

.40% >4

0 .4

10W'.OmUIIflC'-CU.0Om.00'.0%0%000.-4.-4_Irnmrn000 l_I,-44_I_I,-I.-4 .4

00'.

5 0E

0 0"

'.0.05 (N '.010141.44-'. .0W UWWWO%0'Q%O_I_IN(N(.I000000.-4 .4.4_I_I_I

0Z0.

0 U 0N UU'.0.-4NiflNO_I(Nm.NNNWWW0'_I.4N(N_I'.0 0 '.0 0000000,-i .-4 .-I.4.-4 '.0

0ZIll

0 0N0 II)

00'mNUNU0%0.4_I.0N'.ONNODW0'0,-4N(NO0000000_I_I-I .-4 1-4 LA

0EU

11'. 0'.0 Z0 U

10LA0'C.'._IW.-40%C%0%0..-i%0%0%ON W N 0.00.4.400000000.-_I.-4.- U

0 .4 0'0 (40.00. W U) W W 0' W 5.0

10%0'.0'.ONNWO0.4.-l0%m 0 4') 0000000.4.4.4.40 4-)

0Z 5 0U) Z W'.0NLfl0.-4Lt.NWNLt..414114) '.0.0 N N WOO_I .40%

N 0 N 0000000.4.4.4.40 N

0E U) 0U 11) U NO'. (N U N (N '.00.0.

UUUUU)tfl.0W0.0%0.W.4 0 .4 000000000000 .4

0E W 0.- E '.00% m W 0 W '.O%D(N U) 0 m.4 -4.40000.-lit.U 100%0 0 0 000000000000 0

N .-INmUIII%DNW0%e.-4(N

0Cl

>40.4

>4

I')

U

>4

m

N

.4

0.

'0

U)

U

m

N

-4

0zuxI-I E4N U)ZI X-40

I- tuiti'0U0'.

NU0..4

98

LA('4

>40N NNNNNNNNNN

>4U) .,-4 NC'4CNC')t'INC.)NC.')NC.)C.)' N.-IOU)0%CAUQ%C.)U),->4 OONO%,-4.-lO0%o.-ioO ..l NNNC')C.)C.)

'.0CAUNNUC.)NN,-4O%>4

LA NNNNNNNNNNNN

>4

U ,-INNNNNNNNNNN

>4

C.) .-4'-4C'INNNNNCINNN

U)'.O N 0 U) C.) 0'. ON N>4

N .-I.-4NNN-I.-4.-4NNNN

>4 LI)U0%N'.00%0%O.-4 -4.--I..4,-4_I_I,-4No U)0%UC'4N_I.-l'.00'.0')0%z0' _I1-I.4.-I_I,-4o _INO%_IU)'.fl..I0%U5,-'.0X '.0U)NS'.Co'.'.0 4.-4_I,.C.-4_I_I.-4

o U.-IN_IU'.0NO_I.4UiE '.0'.0NNN0'U) _I____I_-4_I_I_-4_o SIflIA-.4C.)0'C.)NO'..IU£ '.0%0NNNNo'.U -4-4.-Co O0m_IUUNImNNS0%NrN0%C.)

o U'0.-I_INU'0NU0%NNNNN0%NSN0%N -4.-4,-4,-4o %)CA0UUU)Ifl0-4inNz U)CA'0'0'0N'.QN'.o'0'.0.4 4_I4_I,-I.-I,.44,.4,.4

o 0UC')'0O0'.N0'..4LAIANNO'0.0INC')U0U)0

-.4 N C.) U Ih'.ON 0' 0-.IN-4——

-4U)U'-4

U)N

cz NOU_I'.0C%)NU),-4..>4 UU)UU'.DU)NLANO('4 NNNNNNC'INC'4NN

O'0_I0%0%U)U)U)N_I>4 '-C0%N0%O0%O0%O,..iNU)..4 C.)NC.)NNC.)Nc')NmmC')

N'.DU).0NUC.),-4N>4 O_IO0%N_I0%O ._I C.)C.)C.)NNNNNC')NNN

LI)C.)0%NUN..40%C.)U)>4

U) NN('4NNNNN('4NC.,I('4

ONCAU_IONNNN0%>4 UCAUC')C')C.)ULA'.ØU)UUU

C')0%.-ILI)UU,%)UCANU)>4

C.) NNNNNNNNC'INC'4('4

>4

N NNNNNNNNNNNNO0%C'4CANU)C.,0U),.'0

>4

_I 1_I.4_I_INNNNNNN

o '.0U'.0LI)C')N0U)O0NE O'0%OO.I_I.-4NN0% -4-4_I_I-INC4NNNNN

o C.)0'.OC.)O'0N,40%U'.0E O'0%O.4O0..4.-l'.0 -I-C_I-I4,-4C'4C'1NNNc'4

o _INN0UO0.NNU)NN0%0%0OOO.-4.-lIA _I_I—4_I_I,-INC'4C'4NC.IN

o OUi'.0C'4O%NN'.00UII)E N0%O%O00.-4.-.4U _I_I_I.-4NNNC%INo O%C.),-I0'.0U)UC.Z NN0%0%OO0%,-4,..C.) .4,-4_I,-4.-4NN.-INC'l

o '-4-4.4NUNN'.0,-4NZ NN0%OO0',.4,.4N _I.-4--l-I4NC'4.4NNo C.)5C.)0('4N005UE '.0'.DU)S0'NU'0.-4 '1l-4_I.-4--4.-4-I-4.-4N

o U000%C.CU)N.O'.ONInN

04 N 4%) U U) '.ON 0% 0.-IN

'-4——

NU)0'.-4

>4 U)C%)511)N C.)C.)C.)

0% It) U NUN 4%) 0>4 c%Io%,-Iu)mNC.)NU0N

>4 .INNNOONC.)C.)C.)U),-4 C.)C.)C.)C%)C.)(%)C.)C.,

Ui0%mNLO_INN'.0N>4 O_I_IC.UNNr')0'_IN0%0,-4

>4 NNOU_I_INNU)U) ('4NC'4C.)C')C.)4%5C.)('4NN

.4U.-4NCAN_I0%N0%NN>4 '.O'.ONO%(%)OO_INSNUU C'4NNNmC')C.)e')N('4NN

0%UUO0%_I,4NUO>4 UULI)N0%-4NS'.ØUC') C'4NNC4C.)NNC.)NNNN

LAU)U_I'.0m_I.00%o'.U>4

N

0%'0U)NC'N50%C.)LI),.I,-I>4 .4NNLfltAUU('4OO0%— NNNNNC'4NCqN(.4N,4

o LAUO%C.)'.ONO'0mOo'.,.IZ -IN-4CANUUC._I0%0'

o IC')UO%0%OC.)C.)..4CAN0'.E _IN_IU'0UmNONN'0'0 NNC4NC'ANNNN_I,.4

o 0C.)NN_Ia'.No'.'.n.-IN-.4U'.DC')C'),-40%'.05'0

In NNC'4('4NNNN_I_I,-i

o NOSU'.0UC)lU.-4Z ON_IU'0C.)N.-40%'0'0'0U NNNNNNNN,-4_I,.-4,-4

o NCWU)NU'00%U'0E 0N0U.0C.NOLA'.ØU)C') NNNNNNNN,.4_I,.O '0N%DU0%_IC.)UNNO'.NE 0N0UCAC.N0inLnij)N NNNNNC'4N1'.l,-4,-4,.4_I

o SU)NO%C.)'.0,..4mNCAOE'-4 NC'4NNN_I.-4_I.-I_I

o C%)'0InC.)NN'0LA'00%z0

.4 N C.) U U)'.ON 0% 0.-C N-4_I_I

C.)U)0'.-4

0N

U)

U

C.)

('1

-4

0'.

'0

In

U

C.)

N

-4

0

0If)0'.'-4

C.) 0_IN N 00% U _I'.O NSN N'.O '0 '.0 LI) LI) U'. '.1) U U

NN ('ANN N N N ('4 CNN ('4

SN '.0_Itt) .-4_I SN ('40N 0% 0% 0% C 0% 0'. 0' 0% 0

N N N N CNN C.) N N N N C.)

C.) '.00% 0 C.) '.05 U) -.4 '.0000 -.4,-I_I_I .-4 -.4 N N N('4 N ('4 N N ('4 N N N N (N N

0(4 U) 0% U N C') '.4) 0%%00% 0'. 0% 0'. 000

.-4,-f_I_I.-4.-4.-C_I -.4 ('4 ('4 ('4

LI) '0 C.) ('40% U 0'. UCA LA U')'.O '00'.0 .ON

- _I_I_I_I_I_I_I_I.-4_I

SOS C.) ('4 C.) C.) '00% '0.-IN N C') C.) C.) U C') U U) '.0 %0'.O

-4_I_I_I_I_I_I_I_I_I_I

C.) C.) LAN 0' U) NN0% UN U.-4 _I_I N _I N_I (NC.) U) U U

- .-I_I -4.4-4-4 _I ,-4 ,-4 _I

C') U C.) N SN_IC C.) N.-4_I_I('4N ('4 (NC.) U U) U- ,-4 ,-4 _I_4 .-4 -4_I_I__I_I

CAN N N 0_IN U LAOS.-4_INN N C.) C.) C.) U U) U) U)

'.00% U ON U 0 U).0 _I,-'..-4.INN C.) C.) C.) U U LA LA '.0

-4 _; _I__I_I_I .4_I_I -4

SOS OC.) U 0% U SN C')'.Ø_I N ('4 C.) C.) C.) C.) U U U) It) '.0

_4 - -4 - _; -4_I_I_I__I

0'. ('40'. N '.0.0 C.) 0%_I N N 1%) C.) C.) U U U U) IAN

0 U N U 0'. N NONSN ('4 C.) C.) C.) C') U It) U) '.0 U) N

-4 _ - -.4_I _4 -4-4_I_I__I

It) U '.0000'. 4') U '.00'. 0_I_INN C.) C.) C.) U U U U'.D,.4 - - 4 _I .-4 _I_I .-4 .-I_I,-4

500'. ('4 C.) 0 '000 C.) NO'.0'. 0 0'. 0'. 0'. 0_I 0 N

_I N C.) U U) ..0S 0' 0.-IN—-4—

99

2>4IA('4 (NN(NNN(N(N(N(NN4,,N

2>40(N

2 '0NIA00I'0N'0NN>4LA.4

2 '0NN0'NIDmO%'.DC')SN>4 ,'4000010,-4040%O%010%O

2 0%0'O'0LAONIAm0'44)>4 NN0,4OOClLA NNCN('1N44)mmC',(N(NCN

2>4 IDNNN0%.-40OCl'0 NNmmr')Nc.icN

2>4 LA10IDIDNOOQClC', ('4c4N(NNcNmmmc4c'4N

2>4 m40'.o'.o'.DS01OS01Cl("4 NNNNNNNC')(NNN(N

2 LAN'0LA0%0'0"QID'.0O(N>4 0'NNmNmLnN.CIAo'.o%.4 .4NNNNNCNNNNC'L(N

o£0) .-INNNNN('4NNNNN

o NO'I,-4NLAN(NOIDz '00101QN'0(NmN-'.0 -4,-4,-4,-4_INNNNNN

o 0'-4OLAm'.n0%'0.-IN£ mNNo%0),-lmNcqNr-.LA -1.4_INNC'4C'I(N(4

o 01O.-4NNInt-(N.4mr-£ NN40r-OmNN40400 .-4-4--4,-4,-4.-IN(N(N(NNCN

o 0%O'001N0%'.Dm£ -4IDLASIDN01r'l,-I.-IIDIO('5 '-4-4.-4.-4_I.-4NNN4

o .-4_ILA0O0%'0N00'.£ OLA'O'.D10400'..-4_I..-4LAiA(N _I-4.-4NN(NNNo '0m'0LA.or—su)mNIDo£ OmmLnm'0r-o,-Iornu,.4

o 01_Im'0.440m'.omoE .40144"004OLA04450 .-40,-4.-40_I_I.-4NN.-IN

N 44) '0LA '.05 010_IN-4_I

IAIA0'.-4

2 '0IA_Ir4'.OIAIflO'.(N('-IA'0>4 010%-I.4O'.0'.01IANOIIA('4 e'4Nc'4mC'5c'4c%I(Nc.4('.s(N

2 OO0'.O_INtn0'.>4 04OO0%0%0%N0%QooN NNNC')44)N(NNN("I(4)445

2 _IO44,N0mme'5_INm>4 0'.O%00%0%04O0%0'.,-u.)145

.-4 NNN44)(NN(NC4)(Nc44)e4)

2>4o,4 (NC4445C'I(NNC')44)44,44)m4,)

2 N00NmLAOLA'.0IA>4

U) c'4c'4mmc.'Immmmmmm

2 IDLAO_I,-ILAO'.'0,-ILA,.IO>4 _IO0%ONLA4')LALAN'0

2 m4,I.-4%oanNLnmr-mbN>4 _IO0%0('540e4)LAu)40m r'Nmm(Nmmmmmmm

2 '0OOO.-I'0ma%Nuo%'0>4 NN4450O'040LAU)1040N NNmmmmmmmmmC,)

2>4 LAN_IC4)OO%'0NID4001.-4 NNmc')mmmC'5o (N0'.400%'00%IDN'.0NE IA1D0'CN0'.C'4LALflU501040' ('4('4N")Nelmmmmmm

o LAONr-.-4(NNc')O'0ONE '0IDNO04SOm'0'0o%o'.0 e'IN('l5e4cIC'5mrnc')m'0

o mb01oN.-'INN40.4LAZ '0LA100'.DQ%(N4,)mOIA NNNm('44%4('4m44,mm'0

o _Im0'_IU)N01_I'00%mE'0 NNN44CN(Nme')mc",m

o 0%04N4NIAC%'.00%'S_I£ 4,5'0'0LAID0'_I_ILAN4') r'1c4(NNCIINC.INmmmc.)

o IDLA0%C')S.-4_IU5'.001'0eZ I4)'0(4)NNIAIAOOc4)'0N NNNNNN(N(Nmme')m

o _IIDNIA.-4'.0e'aN_I'0oE I4)N(N%OU)'0e4IDO_I'-4 CNNNN4,NCNCNNNmC4)

o '0Om_ICn'0(NONmo£ NID-4'.D.-4N.-4LANLfl,-4o N-4NN('4('4.-4(N('4('4N445

N '0 U) ION 0% 0_IN-4 4-4

IDIA01-4

2>4IAN '0m'0'0'0'0IflIno2 IA0NU)S0%N'.0Ne'>4o('4 44)44)14)4,5

2>4IA.-4 NN(NNNNIm'0,'%

2 1DU)NN..4NQO'040r'.>4 or'4,-4NmU,m'0NLA'0o.4

2 N'.0O0'0m'.0r-om40>4

LA mmrn'04414-,')m,')2 IDID0010%ID,-4N40N40>4

'0

2>4 cNm'0LnNNInN'04')

2 _IS_IIAN'0'00%0%0>4 Nm'01DIDr-1Doo'0NN mmc'5mmC'5mrn'0mme

2>4

.4

o _I('INID_IO0%IA'0O'0IA£ 44)LALAIDIDO%Q%(')'0(NLAN0'. mmmmmmm'0'q'0me%

o '0IDmo'.meINUS(Nou)r-.£ 441IA'0'0LA04I)4,_IIAN'.0

o '010'.A.-40'40NU)UCLAO£ mU)m'0'0N(N(NoLANIn mmmmmmm'0'0'0mN

o NIAS_ILA0%ONIDO£ mu,cNm'0'.D,-4,-o.'.A'0

o NN4O_INI')G.-4'.O.-.£ (NIA,-IN'040N0O0'.iflm

o ClN0%IAN14)IDNIAN£ -4'00Om'.A.0o.u)N mC'544'mmc%

o 0%C4)10-4N_Ir40.4E om0%oNmmusIOLA_IN.4

o U)0%—4In'04100040'.cZo mNNmmNC.INC-'4mC-44N

("44,1 '014)105 GO_I N'-44-4 .

NUS04-4

a(N

2>4U)'-4

0-4

U)

'0

2>4

N

-I

0'

'.0

U)

'0

N

'-4

0

'0U)-4

'04,) _I01'0mr- 44500)('440O0101C'ID44)44)44)445 ('INN ("IN N (N ('4

N '0N-4 NON m ON_I.410 N S LA ID U) InS 0%(4)('ICNN(NNNNNNNN

44) SN _IO'0NO% m 0.-4'.OON 010'. N("-SN0'C') ('4 N N (N N (N (N (N N N ("'1

0'01 4,1044) mm LA'0'.0 N%00% N 555(N N N (N (N N (N N ('1 N N N

LAN mom 4440% SOC'S NLAID '.A 44) N LA U) LA LAID 405

(N NN NN(N ("I ('INN (N N

0 445 N '.0 N U) LAN U) (4)010%'0N .4010_I N_I_I '0 '0 Il)N(N (N _I (N NN (IC ('INN (N

N It) '0(4 4 4') N Nb N_IN4001b,N09r.N _I_I_I.-4 I_I .4_I N N (N

N '0'0N0% smm 0 '0'0U) mm (IC N_I .-4 N U) LA LAID

0mm m _I N '.0 mos('5 (N (N .4.400_I m'0'0'044-4-4-4.4.4.4-44-4.4-4-4

01 (Ne's 040 '001.40 400)-4-4.400% 0 CNN (N (N

- -4.-4_I000_I_I.-4_I_I

(N 010 LA .-40".DN U) e's ,-4 m-40-4 0% 0' N 0% .4 (N N N.-4.-4,-40000 0_I_I_I_I

IA LA 100' U) ('5 (N '0O%NNN000 N 010_I_I_I_I.-4_I00000,-I_I_I_I

N.-4 4,5 '00501044) (N (NO0100 '.Db0%0_I_I.-4000000,-4.-I,-4_Io 0 '0_I40b40 IDN'0010'ONNIDS 0100000-4000000_I_I_I0' N N 50' (N '.0_I 14)0'.

01 04N'.D IAN 0101010101

000000 0000IA'0O'. N_IN '000 ("410 No G'10 IA'.Q IDm 015 01

00 d 000-40000.4 '044) .05 010_IN

-44-4-4

100

lx>4 NIn40In(NeNN00%0%In(N

lx 4(N4DIn0C4)mN0%.-40>4o(N (%,0(4)C%,C4)C4)(%,

lx 0%In-l0040e'O0>4In-l 44)4%) (4) C') 4%, C') 0(4)44)4%) (4)44)

lx>4 InIn0O4040LflN404040o.-4 4)m(4)C')4)4%)

lx 04040(N0%00In->4 (N(N00—4.-4In0%r-0In

lx N0%NIn404%%00%C,—40>4

0lx 40,-I In 00% (N (N In 40 In N>4 00%NIn0%In0%0%0%NQ(4) C'%(NN(NCN(4)C'4)C4)(4)O

lx In0%0(N0(N40Q00%>4 0%4040('%m00%r—oN (N(N(N(N(.4(NC4)4%%c4)r')C4)C')

lx>4

4O 0%Ln-44000e.4inm4.Z In.-4040NN(NInc.)In00% (N(N(N4%,mc-(')mo 0N0-4,-40In(NO00%E (NN004%%InOU)(NIn(N0 (N-4-4-4-4-4-4(4)4%)C4)r#)44

o In40ON0%40,-nooE 0NIn0(N.-4C4)0n,.-4In(NIn CN(4)C')C4)C')o NC4)In(4)0CNON(N0In40z 4000In,-IIn.-I0o 0%.-4400NNe'04040E 4004%)N04,-NIn0In0(4)

o 0N0,-4InOflC4)ZIN 00,-I(N(4,N(4)(No ,-I0%00Ina%In0mC4)400.-4O.--4In40C'%N0N40,—4 000Nc-.ac.io (4)00O00u1Oe4)4.)In#-IO%.-40u1c%,OOo 0.-400O.-4,-,-40(N

(N 44)0 In 40 N 040 .-4 (N_4 (44

(1%0%-4

lx>4In(N

>o .(N

lx NN0%..-440(NN0mIn>4In

lx 0QLfl0%C')N.-4b0%>40,-4

lx 0%mCNLn0c%40C')o>4

In om00000.O.lx>4

0lx (NNcNOmInc.4a%o0%o>4

(44

lx>4

('4

lx bD0InODfl0%0(N>4 N040%04-l40InLn(')0%O,-4 '4)C')')')'00000oInO (N(N00,-440bC4)(N401n,-4x N400%0%,-4.-40r-.0%400..04 C'%')C%,C%,00Ln000In&n

o InC')O0m00N40InNZ InIn('44%%,-40%0.(N40

o 0%N40C'O\CNNZ 00Nb(N0N0%40,-4(NIn (4)0000In041Ino InN0r-0r'-soInE 4040(NN40N.-ir.0o 0%(NC4)14)40InC4)c-40(Ntoz(4) C')C"C'%C')C%,OInOu%In

o (N040CN0(N0%,-4(N00E(N 4%%000u%0

o 0In400%Q%N0%04,.4%0%0NNNNN0%0(N4-4 N(N(N(N(NN(N(')44)%,oO

o (N0000N0C4)0NEo NN(N'-4(N(N0

.-4 (N (440 In 40 N 040 -4 N4-4-4-4

OiIn'-4

lx 040In,-4,-lmOONO,.4>4 00%0NNO04NC'-In(N 0C')04%)4%)44)0C'%c')f')m

lx LflN.4(.4(N(NU,0In>4o .(N Inmmoo,,lx>4414

,-4 O00e.-lx>40

lx>4

It) O000004%).ilx>4

0

lx ,-INC4)OmOInF--040.-4r-.>4 C'4NLt5r-4InN0504,-4(.C') 0'0(4)(')C4)(')O(4%

lx DN40Q440O,-0'>4

(N 00000OC')C')c4)(')(')m

lx>4 N04A00%(')D('4(,—4 4)00C')4%4)c.4(4%c4)(N

o In(NOOr—0r4)40,-4040(Nz 00r-(N400(NN0% 00C')004%)(')(N(N(')(N

O 0%40(N40404000NIn0,-4z 0440,-44%%,--40%0%D.-.14040 oomoC')C'4(NNr4(.5(N

O C4)0N0%0000.-4.--4E 00N4%4(NOO,-4O,.40In In0C')0()mt(%,(No 0tD.-4(N40(N40m00oE 0InNNInO04(N000400

o r-I(NN0%,-40%4014)N40.-4r-Z 0%NInCN0NN('4(N0In€4) 04%)(4)(N(Nm4%,(Nm(N

O ('4C')(N.-4(N.-400%-40,--iZ 0Ine')4-4InIn004InIn.o(N OOm4)C'5(N(N(N(N(N(N(N

O ,-I0000NCN(Nfl40NN00%OIN000,-I0%(N.-4 (4)0(Nm(4)(N(NN(N,-4(,(N

o 00(NInIntn0%(Nor-z (NIn0(N40040(N(No

v-4 N (4)0 In 40 N 040.-I (N_4 _4

000%-4

101

lx C"€4)N40InN(N40eIn(N>441%

N

lx 40ON00NO%CNONN>4 040%N40InN0%NQo(N

lx>4 00%N0%Q000%Il)'-4

lx 40InInrN-4InInN(N(N(Ne>4 00400000000.-.4(No_4

lx InNNN.-404N(N(N(N040>4

1) 0(4)0(4)00000000lx>4 00040C')(NOO(N(N0

lx NInIn00000In(N400>4 0%Nb0c')(NOOOOC')

lx In(NIn(NLn,-4.-4mc.0>4

N 44)44)44) ('54%) (4)4%) (4)4%) c')4%5 C')

lx>4

-4

o 0InN,-40S0OInN40z 0.4O(N(N(N(N0C')4%4(N4%)0% (4(')C')4%)C')(4)o 0In040In.-0,.-0Z 00-400oe.-4,-4..4(N0 c.sc#)o N0N0O4InInE N000%u.-40e,.IIn (.N(')N(N(N(N4%4(%,(%,4%%(')4%)

O 0N.-4(.4cN40In40N(NE N04IflN40400%004O%,.I0 (N(NN(N(N(N4%4(N(')(N(N4')

O ONNInNN.-4(N0400440E(4) (NNNNN(N4%4(N(NC'4(N(4)

o 00%00%0%%fl40N,-((NO

N N(NIN(N(N(N(NcNe.c'I(N(N

O 04(NU%(NLI%0e')0eo.-o.-4InmmC')(')(Nf'5.-4.-4In40

.-4 N(Ne4NINNe.INNO b(N(N00(N004oo(NEo CN.-4-4.-4,-4,-44%4(N

,4 (N 4)0 In 40 N C' 00.-4 ("I4-44-44-4

O'4LAU)LA0NC'OLA40>4 (')LA04a.a.No,40LAN 'a'aC.)re,'a'a'a'a'a'a

>4 OONQ'.O'.ON 'a'0'aC')c')r')..)c")ENLA0.0.40ONC')00400

>4IA4

>4 EN-1,-4ENN,-INENEN0

>4 C')C'),C.jENr')ENC'),4QLA

.-)0'.'aS0OC')N'aØ'.0.>4

'a

0'.'aO4EN4040,-4UCNCN40u)>4

C') 'aaC')'amm0440C")40LAr4,40'.O

>4

EN

>4 40EN4EN'a,-400,-4,-4,-4 C')C')CflC")C')C')C')

o 'a,-'4ENO40'a0(N40'aa.E 'aENO'.,-4,-.I0'0EN0'.00.0'. C')C"ENC')C')C')mmNENEN

o 0LA0O40O'.40LAQa.40E C')..4N004Q'.C')N0'a.o'.40 C')C')CNC")ENNmC")ENNc,)EN

o NEN0'(NCN'aa.,XLA

o O0.'aN.l40m5,-'a C')C')C')ENENC')mc')C")C')'4C')

o NC')'.0LA'a0'0NEC')

o 0.CN'a'-4,-ILAr.J40,-4SNNE O'.0'0'.a.Oo'.a'.-.o'.EN NNNNNENC")ENNENC.')N

o N0.40(NC')ONNN(')ENN4040NN0'.NN40O'.0'.,-4 (NNNNNNENCNCNENNN

o O'0LANOe.)NLAz NN4040-.4a.LALA40040 CNNNNNNCNNN(NNN

,-lN C') 'alA 0 N 040 '- EN,-4 _4

N400'-4

(')0NQ'.U'4LANEN'a'a'a>4LA(N '0'a'a'a'0'aC')'a'a'a'a'a

0'. LA'aO'. 'aC') N C') C') N LA> a.a.04000a.o'.a'.a'.o,-0N C')C')C')'a'a'0C')C')C')C''a'a

>4 O'.COl00a.O0%,4NLA,-4 C')'a'a'a'a'a'a')..'a'a

CNC')40WN0\,.-40'.OLAC')5>4o,-4 'a'a'a'0'a'0'0'a'0'a'0'aI>4

LA

>4

'0

NNO'a,Ofl4040LA>4 0'O0'0OOQ0.-4O,-lC')

>4

EN

>4 000'NNNLA40.NN0'..-4 N',C")r.o 40ENN00'.0S,E 0'04N,-'4O-4'a404040N040' NC')C')C')C')C')e')C')

o '0LALAO0,-4O40.a.z O'.O.'-4,-'IC-4LA4040NN40

o 0ENENLA40')40,..5z 00'N,-4,-4,-lLA4040NLI)

o 04NNLAN(NO0a.'0LAz 00LANENLA.bN'a

o '4b00r .-4'0 N 40 CNNE '-tO'0,-IN('9LflLA40NNC')

o O'.—404.-I0'C',LALC'aLA,-z O04EN0.,-40'0Lf')'040N40N C') (NC') EN C") C') C') C') C') C') C') C')

o 4O0.-f,-4NOENE 0.LAO0.EN'aC")'aLALA,-4 ENe')e')m')C')C')o LANx 0.0.0'.OLA(NOLAo NN.-4CNC.4(NNC')C')(')C')m

,-4 EN C') 'aLA 40 N 0.0,-4 N-4 ,-4 ,-4

4.)

04-4

0404NC')m4040,4U)ENON>4 000 ,-4,-l,-IOLA(N 'a'a'a'a'a'a'a'a'a'0'a'aX ''4a.N,-IN,-404C')O40m>4 O0.O-40O,.40,-4,,.4oN 'aC')'a'0'a'a'a'a'a'a'a'a

EN0N,-4O,-4NCS,.4N>4 N-4.'C'4,-4NNLA.-'l 'a'a'a'0'a'a'a'0'a'a'a'aa ,4C04C')N,-4,4040C',O>4 C')C')CNNE'.INCNC")C")ENNC')o,-l '0'a'0'a'a'0'a'a'a'a'a'aa: 0N,-400%04400'40O0'.>4 C')C')C')C')ENENC')C')EN(NNNLA

,'4'a 'a (N .4 N NON 4)>4 C'91'CCNC')NEN'aC"CENN,..IN

'0'E'0'a'a'0'a'0'a'a.X>4

C')

C')Oa.OLA0"aoa.ENNN>4 0-4NENC0'.0'Oa.o,..0EN '0 '0'aC') ma m'a'a'aa: (N (N'a C'4 .-140 C')'aO 'a'am>4

'-4

o 0'0%LAN.-(NQ'.LA40,,-04z O'.O0.00404ENa.0'. C') C') 'am 'a C') C') C') C') C')'a C')

o 'a4040O'aSr')LA0a.,40 C')C')e')'aC.')'a'a

o -440LALAC')400'.0'.O'aa.a.Na.0'.C'),-4LA

o LALA0.04(')C')N,.N(NNE N0.a.N4040NNa.04N,..

o LA'a0msE NN40u)4040,.,0C')

o N0,-444')ON, 40N40404040'.O40'.ONOON

o a.400.'aNa.'aUI40CNLAO'aLA'a'a41)LALALALALI)a.

,-$ rnmmC',I')me')C')C')C')

o LA,-INOEN,.4Oa.NC')C')LALA,-CONQNLAC')m,.l

O C')C')f')C')C')C")

, N4') 'aLA '0 N 040,-I N-4

'040a.'-I

102

Oc)LA'aLAm40'0LAOa.EN>4 '-4.-INNENONLAN 'a'a'a'a'a'a'a'0'a'a'a'aa: 'aOLAN,'-4'aN)a.a.m>4OEN 'a'a'0'0'a'a'a'0'a'a'a'a

OO,'-40C')O>4 EN-4NN.-INNNEN,-4C')C')LA,-I 'a'a'a'a'a'a'a'a'0'a'a'aa: 0'.LC4'a0'N400.NNN0.>4O,-I 'a'a'a'0'a'a'a'a'a'a'a'aa:>4 NCNENENC")NC'4C")'aLI)LALALA 'a'a'a'a'a'a'a'a'a'a'a'aa: EN0NNmLANNN,4Q'.40>4 N EN N EN (N EN EN ma LA LA N'a

a: LA040'.'aO400.C')N,-O>4 '-40"4'-4.-IENENENC')'a40OC') 'a'a'a'a'a'a'a'a'0'a'aLAa: Nm0N0NO,.4>4 COC'-4-400.-4NC')'a40EN(N 'a'a'a'0'a'0'a.O'a'a'aLA

NSENcN0'.)Na.,m>4 OOOO0OC')'aLA,.l,-4 C')'a'O'a'0C')'a"a'a'aLAo N'a0.40,-4C,-40(N,.0.a.—400,"4o.o,-4'a'a'ao0'. m'a'0'aaC')'a,.'a.a'aLA

o 0LA -4 '0 '0.-I NNNEN,-I,-4O0EN'a'aLA,,

'0 'a'O'a'0'a'a''a'0'aLAo LAN,-I40C')C')N..-N'aNO

LA 'a'a'a'0'a'a'a'a'a'aLAo 0'N'aLA0'.NNLA40'a40E .-IN,-4,'-400%O(N'a'aLAO'a 'a'a'a'aC')'a'a'a'a'aLAo LAC'440.-4,-4'aC')N'aOSz O'-4O'-40a.Oom'a'aNC". 'a'a'a'aC')'a'a'a'a'a'a

o LA'a'aO,-IEN00440C.)NE O"-4000040.O,-iENm40N

o a.4004'a40LAC')O0%O'aLAE O040'a.04a.O,-4LA,-4

o ,- '-40 (N 'a .-C -I0 C') LA N IE '040NNSO40C')0 rnCC')CC')'a,-4 N C') '041)40 N 040-4 N

-4

LA400.-4

>4IAN

>4oIN

I0%DNN,-4.00%OOm>4 IN'LflLAINIANLAIA.-4 00000ou.o.o

0LAN0NANNInOQIN>40

>4 NO0N0%C.)10In CIAUU1IAIt)0.CLA.0

OO0%NLfl10mIANo>4

IAiAuuU.

>4

m

>4

N LAIAIALA&AIALA10IALA

>4 ONOONr.)0mIDN0%.-4 IAIAIAIALAIALAIDLALALA.0

O '0.4Cl00OID0%10Z ClC'O0,-4,4N0%ClID.00%C, IniALflIAIniAUUU.

oE 0%N.-4mLAN0ILAOID '0IAIAInInLAIAtDLAIAU

O LAN DO N.-400.-iE 0%NNIncoe..-.410INLA UULflInLflIn%0IDIDULfl

o DNN'.-4ID.-41010ONE0 InInLAInInLAIDID10LAIA

o C,IAID0%CN00NNAlIA0%0%NNN0%IN.0N). LAIAIALALAo LAID NU.-4O%..4mtt)sE NNID01N1010NON

O 0NIANNNNINNInNIflNN10O%Nf.e— OOCIAIAIALA0

o 000I,-lNCl.-4O.010O0 UC

— IN 11% 10 N 010.-I N— .-4 —

lblbCl'-I

IDNLAN(.OI N>4 NN.-4ID 0LAIN '0U)LflIAflIA 0

IA0N..I.4mON>4 1LALAN%DNOINLA).,oN

>4It)—

>4o—

>4

LA

I00%.-INIDON4>4 IDI0ID0%NNLANNO0 LAIALALALAIALA

>4

C'.

IALfl.-4CN0%LANINIDN>4

N

NO.-4ONOIANO,..>4

'-4

O NID10OIA.0z IDI0.-4.-4.-4IDLAID10N90%

O LANON10r-.IDNQ10 000IALAU)LAI0ID10O 0%INNIAc'ioE IONIA

O NN0c.ic.00%InE 10N1ONCN)NIDO0

O OIDDLfld%N.O0%E IDIACNN4.')100C.IDC,() I0UUILflIALA

O InOIDN'010,-4NN,.ZN ')I000U.0IAIAo In-I.-410N%D,.4IA..N0z I0N.-I0%0NIAO%— 0.0.oo.o.oo NN.-4.-IONOl0%.CONOZ N0%0NIN.0LA0%o

— Al C' In ID N O'0.-I N.-4 — —

NIC0)

IA 01Ifl0N.>4 LA ClNO%0%INClbU,N

>4 0cNAl.-40%.-4O..4.)NoIN LAInIAIA0InLAIAIAIAInIn

NC'4)LA0NInIDLAN>4 I0OIN%O0%NLAIA,-)

0O00%0%OClc..>4 NN.-4NClInLANOU,o—

>4 l00%0%ONIAIDNNO%NIA InLflIAU%IDLflIALt%4AULA10

a ',LAIDNNIDIAIDN.-INO>4 ID0%NIA1010NN

LALALflIALALflIALALALflLAI

>4 IDCl.-4IAIDLAIOLA10mC) LAIAIAIDLAIAIAU,InLALA10

>4 I0O,-iONInmIAU1..,LA(N IAIAIDIDIOIALALALAIAIAID

>4

.-$ LAInInIDtDIDInLAInIAIAID

O NN,-4NC)O0INLALAE NIDC,0ININ.0AN0%100% IALAIAIOIDIDLAIAIALAIAIQ

o O0%.-4In,-40%iAIn.-zID IflLAIt)ID10I0InU,LflUD10

O INmNIA'0%In,-,-4.-).-LA0'DNO'CONNNNOClLA IALAIAIDIDIDIAIAU,In10ID

o I00%In0010..4InINNE0 IAIAIDLAIALAUUILAID

o IDOlClLA0NNIDO10In,.Em InUIAIAIDIAInIAInInU

O ID0NN0%,.ItnNOz oe.No%.oINf.,mNC,.IN IAIAIALAIAIAIAIAIAIAIAID

o NIDIANI0N— Ifl&flLflIAIntflIAInIniAlb

o 0NNID0NNOIANE 10In0NNClID..INClo LAUulIA

— N In ID N 010.4 N— .4—

IDCl'-I

103

>4 OlINU,N ml!,

X INNLflmN'0IflOIn.C>4 0ClOID0%U,NNoN IDIAIALAU,U,IflIn)AU,InIA

>4If)—

NIDNIDInONOLA>4 IDLflIDC4N.-4..4Cl0(.IDo— IDIDIDIDIDIDIDIAIOIDIDID

>4

11% IDInIDIDID10IDIDNID10

ON.)flN10N..4>4

0 IDIDIDIDIOIDIDIONIDIDN

D>4

C4 ID%DIOIDIOIDIOIDNNNN

>4 NC, N LAO (N — — N ID NN %DIOIDIDNNNNNNNU3

>4

,-4 IOI0IDIDIONNNIDNN

O OINmONOa,Z 0r'mNOIO.4NID.00% IDI0IDIOIDNNNNNO N.-40NClOIN.-4NID

O OCl.-41m'000N0%o,-4EIA

o 0%N00OU,U,,-ILAE NNNmNIAIDIAOInm' %DIDID%0lONNNNO NNN0IInIN.0C,IDZ NIn0INIAID'0O10.IN,..4m IOIDIDIDIDIDNNNN

o NmIN1010.o0%NZ 'CIA00%mIDInONNONN I010IDInIDIDNNNNo lb.-IU1.0OmoU,omE 0m.-LAm..-IooIDm.-I ICIDIDUIDIONNNNN

o NID.-Imo%.0.0oNooE mIDLAIA,.4NOCl0 LAIAOIDLAU,IAIDNIDIDIA

— N t) LAID N 010.4 N— — —

ClIDCl'-4

>4It,N

X 0%.>4 %DO ..D O4 >4O . . . . . 0N %0%Ø It)In N

N-4ND>4 0O0% N.D >4It, . . . . . . It,

Lfl 'O D

N%DOSN,n>4 0O >o 0NbbN.. ,..4

O NUNOc,>4 >((4 NNNNNN4%0 ((4x OCI(4O%0fl>4 >4

NNNNNNNNNr..Ø

>4 NC4Nc >4.4 NNNNNNr-.r-N 4.,

bO.-.I0,->4 0%0%-4 >4N NNNNNNNND%DLf ('4

Z NO%N>4 0%0% 0'NO'0,f, >4'4 NNtI -,o %DOf.)SN0OD-4.4O0% N SN InU 0%

o 0%mNSbOO 0Z 0%NNNN0 0o 0%0N NU—R(O 0z O In0%0%Its '0NNN In

o 0-SON l0NLf,00%Ntio ONNN4.sm-lo4.) NN'0Lfl. 4.4o NNO% ,'0 0('4C'9N0 XN NNit, No .-IOON4.).-4.-SNO,.40 0

Z-I —

o InO0%NN,SL(40zo 10 LtUl"C) 0

('4 (.4 In 10 N 0% 0 -l N_4 ,-4 -4

-IN05'-4

o '-i5(40 NsO 5050

Ifl05sO($51150

050% ON 115 sO 0' sO SN 4.) (.4ON N In 4.) LI) 500 50 In

'.00 In 4.) 0 In ,-l0 4 0 '-4 N N '.050 (0 '.0 '.0 sO N N '.0 '.0 '.0.0 '.0

,-4 N In .-4 In NO Os '.0' '.0 s-I N '00-4 ('44.5 N N0

'.00 '.0 0'. 0500% 4.) 5 (.4((5 N 50 50 05 s4 s-I C sl N'.0501(5 '.0 '.0 N50 '.0 LII '.0 III

00 UI 4.50 s-40'. '.0 ('54.4C4N0 4.50% C') ' 4.)US In '0(0 sON N 415 LII US III In

NO s-I 50 ON '0 ('4 C'4,-4 ,45150 '-4 s-I N 4.5 N '.0 N N s-I

'O'0 In US '.050 In LII Its (15 In

4 - '.0 NO N 05 ('4 N In 0% 4.50' 0' '0 '.0 s-I N US '.00% In

Lfl

'0'0 '04.4 415 ('5 s-I '0 s-lOSs-I N 0' '0 N N '0'0.u ('54.5 '0 '0 '.0 '.0 LII LII '0'0'0

.-4 ('50% UI ('1 In ('40% N '.0C') '.0 SN s-I N In '0 (15 N (.4

o'o '.0 '.0 In (15.0.0.0

s-I '.0 In 4' 00'. l (15 ONN N LI) LII US s-I ON 10 '.0.0

N '.0 NO, '0(15 N III NN10 '004.) '00(11.0 III'04.) ('5.0.0th 415 '0In'0 '0.0

s-I 501150% ('4 , In ('5.0 ('5 ,-4.-4 US N s-I '0 ('44.) ('S S In In Os

0% '.00 Sf1 ('I 50 N '.0 ('IN N.-4 C')U5 0 '0 ('4 '0.0.0.0 ('5.0'04.45.4 '0((4 US '0 '04')

0'4') 100' N ('4 QS (IS '.0 N'00' 0.00% US S '0115 US so ('I

5•S 4') ('S ('I ('5.0 '0 '0.04.) 4.) ('S

,-4 N '5.0 In '.05 050-4 Ns-I ,4 s-I

N '.0 N'0 U') '50 '.0 N '.050 NO I') '.0 US '0 N ('I N '.0 445N '.0 '.0 '.05050 '.0 SON 5050 '.0

so ('S US ('4(0 N 00% In'0'.0 US '05.4.0 '0UI '0 '.0.0 ('S'0

sD

.050 4') s-S ('4 '.00%N ('9 '000'04.5 '0 ('4 C'S 4.) N'0 N ('4 - N

N IfS In 0', '0 In 04.) '0N('S s-I ("5 ('4 s-I s-I 4.) 4') s-I N s-I ,-4

'.0 N ('40% s-I '0N 415 N ('50% N0% ('I ('4 0'. m c r S's.0L In '.0.04 5155050 '.0 '.05010

'01150% 4') ('4.0 ('1415 '00% 050050% N s-I 0 '.04.) '.00% 'ON

'0L (IS (151(4 15 (15 US In In In

N '0 '.0 It) 0.00% 0' 'ON s-I NUS sO (15 0% sON sO 0% N '0 '0ULtInInON00 It) N ('4 ('4 ,-I N ('5504.) 1') N N '.0.00% 0 sONS

0(1 445515

s-I SN '0(15 US .—4 N US 0%4') ('40 0 N '0(4 OS '.04.-N

'0£! .0.0 U) 445 In III In US In

O LII SO ('S ('4 ('5.05104') US In('4 s-I 0'. US LI) s-I 4.) 445 U5 S

LII U U ($5

N '05'S ('5055 LII OS N In s-IN OS N ('5 ('4 '0 1') '0 4') (450('S C') '0.0 '0.0 '0 '0 lAS US US U'S

N -400" N ('50 '.0 ,4 '050In It) ('4 0% 4'50 II) 050% 4')

sON '.0 '0 ON N NO'0(4 Its ('5500550 'OsO 0' . . In'-I so ("50 -I 0' ,-IN (II 0',CN'0 ('5 ('5 C'S S'S C') '0 N 50 '0 N

('54.4 . 4')'0 '0 '0.0 '0

s-I ('I 4') '0 Ii) '.0 N 050-4 N-4 I .-4

0%(45 4LflNsO'00

(05(4 N '00% '.04') 0.4500%DOs 050 ('4 ,g (4) _

NO C') 500% 45 4150%N US '01(50% 0% 0-4O '0 N'0'.O '.0 '.0 '.0 '.0 sO S N N SN N

In N 50 ('S '.0 .-4 4') s-I N 4150%I') US LII US 50 ,-4 N U') US'0

0' In 0(40 '0.-I '0N NON In In 50550 .-IN US 50 U') In

5050505010 SON 5050 '05050

Os NN '041)4') '00 (45 .-I'0('540 In 445.0 US ('4 US ('5 ON '.0

('SN In '.0-40% ('4 US 4.) 050'.0w In Ifs ('S '0 '0115 s-b US US

N5050b5050

,-4 N SO US SO '00% 0% '00 N('4 NNs0 SO In '.0004.)

'0C'S 054(40% '0.0 '.0 ('5 115 N500% NO, 1') sOs-IS'S 0%.O In

50505050 '.0 N NUSsOsO

'04.) N 0% s-I N N ON 00%s-I'0 I'S s-b 04') ('4SOO,4aa)rNLII s-IN LII 50510 s-IsO 100% 0'04.50'. .'IO N 0'. S'S N US In

'0N 4') ('SN '04.) '-40.0'-I '0(90 N 0% NbC') ON N

'0N S 0% 00' ON s-I ('450.-I I'S 00% ("4 N 500 US ('S

s0'.ONsOSN N N5010 0" ('4,-I 0 '00% '0 US0505 In C') NC') N ('4 US '.0 N

US 10 SOs,ONN SN N N

'0.-I N ('S '.0 0% 0 US '.0 USsO N 4') s-IC'. US InN '0(4.0USIn%0'.0S0N10SNN 0% N In 0010,-IC'. '0445 ('5N '0N 0% 0 4') 05 '0445 N

In US In 51) SONS NsONN

s-I ('4 5') '0In SON 050-4(5*s-4 .-4 .-4

445(SI

0('4

0s-I

>4

1(5

'O

(.4

>4

N

>4

s-b

0%

'.0

UI

'O

C')

('Si

0

NN0%'-I

0('4

0-4

(IS

'0

('S

('4

'-4

OS

'.0

US

'0

('S

N

-4

0

4',NOs'-I

0N05'-4

OL+

0N

In

>4

N

>4

I-I

0%

0

U)

N

-4

0

0% 5 N0 N>4 N- 40 >4U) . • U)N -I 40 N

-440

>4 II,mNOo%.c,bNmNN >4O • 0N N

>4U) • In.-l SS ,-4

5400>4o • 0.-4 .-4

>4 100%NNN0%00Ln0U) u

X>4 0S0N.-400N0c)0

X 400%100U)N>4 b-40%UO0% >4m NNNNNN 4')>4

N S%0NNNcONN N

>4

4 N0NNNNN40 -4

o 00m0%0% 0E N U0%SO%U) E0% 405105555404040 0%o 0-4 0Z 0%N400%NN Z10 40

o S40Lfl.-4SO0NO40N 0x ,-lO%000 -4NO ZU) % DNNr 040 11)

o 0041 Dm040N0Q% 0z N04040 40NOLn E0 NSN40401t) 0

oZ NOOLnONmONID4') 10 40U)40NNN4040U) 4',

o N0%O-4N0O%NN400,-4O0%0NoSr-OoN % %D%DLfl40U) N

o 00 0%x1-4 In InU)U) lo IOS0%U)S

000N40U)0%SNNIDO%o Lfl00 0

1-4 N ') 0 U) 40 N 040 .-4 N

0N0)-4

NO 0 ID 00104') 0% 5ON 0 4 4') N NO 000 NNNIn.-)N 0 ON U) 040% 40 InO 4') ON 0 N '-4 00 .-4 N

000% In 0% .-4 4') ON .-4N NO ,-4 N .-4 N N .-4 N In N

400 0% 4') 4') '0 0 0% 0%4' 0040%0% ON 004 N ,-4

N NS S N SN

In SO N 0% NO In 4')N 0% 0% 0040 N 00 NNNNNSNN 0N 4') N 00% 0% In 00% N

NNO% S 4') 4')0400NNNNNNNr—%04040

540 N N 00% SN .-4 b 0404040 11) N N 0 C') 00' 40.-INNNr-NNNN 40404010

N 0% C') N 4001)04 Q% 00In 0% OS ('40 10 C') ,-) 0040 40 40 N S N N 40 40 40In In

0400% '-400010 N In C') N0% lOIn 40 ,-4 500 04 40 00

010 N%OIOIOInLnInIn

(4 .-4 In S1040 00 ('4U) 4') ('4 .-4 U) .-40% 00%

40 0 ID U) In /4 0 0

SION U) 0 08100000 ('400% 11) ,-l08 00%11)4040 U) 401040 In U) U) 00

4') '0 N 000% ON N N4')0 N U) 4') OhS0

41) 44' 40 40 40 Ui 4)) 1/4 0 '0

0 ,-4 In N 04') 010 InNN In I)) 0 40 C') 40 r-

In U) U) In 10 II) 114 In In U) 00

0-40% S .-4 04') U) C') ,-4 InOQ 4') .-4 .-4 U) 4') ('1 ('4-4 U) Li)

InLflInU) In U) '0.0

0405 In U).0 4') In N U) ()In 40 U) C') .-4 .-4 0 04').0 '0 In U) U) It) Li) .0.0.0

('41051010100% 04') 04 SNU) 0 N ('40 ,-440 C') N

N4')'0 114405 040-4 N.-4 .-4

04040 5 4') 404)) 0.>4 .-4 405)14 0 5100In .Cl SSS Sr-SZ>4 0%04O00N,-44')0 .N

>4In.-I

>4 O0O0Ot-S0%O0 ..-l NNNNSS>4

It) SNSNNNSNNNSNz>4

'0 SSNSNSN5SNr-5.-4mIn4oSoo.-4e9o.O

>4

4') SNNSNNSNSSSSx>4

('4 104040404040404040555

x>4

'-4 104040104040404010N55

o InN)O,—4mInE 0%O5040%04')In4'4N0% InInInIn10L1i40%010N55

o O00O,-4In0Lt)40401010InO40NC')040O0O40 4fl4i)LnU)InIn40IDIDNSN

o C')ON40.-44')0U)4'50OE InONONSNO10000%In U)LnLnInU)U)%OIOIONSID

o 0C')0%0%4000,.4E -4C')U)40O4')40.-40%04'0 LnU)InInLnU)4040405040

o NNC')N10%0S0%OE N.-l04.-400.-I4n0451') Li)In'0IflInInU)4040404040

o 0NC')40.-40OS0%0%.-4E 0%04NNN10N.-40040('4 0'0'00InInU)InIDIOIDID

o .-41fl.-4IDOSU)U)0%1000E 40%0U)In004C')qSO0N,-4 U%'0IflInU)IDIOIO

o .-lO4')O10U)InSEO 00'0InU)U)40U)U)

4', .0 U) IDS 0.0-4 N'-4

NN0)-4

C') NON ('40% N 40 —04 IDS U) '0 U) '-40440

NS0%oso4r-N('4 .-440 4)) 40N 04') N '010('44') 05 N 5504 N 4') .-4

C') ('40440 0 '0 N U) ,-4N .-4 C')40S U) NO. IONO. (8) .-4 C')

N C.: F ('-SN 54040 0 N ("4100% 08 C').0 0

04 N N404)) SN U) U) 40040 0 N N SN N NNN N

0% - N C') C') C') LI) .-4 NOS0% 4') N 0% N N4040040 '0 NNNNN NNNN

0) N Cl) U') C) C) 000 if) 0 C)N0% 0040% C') 0 ,-400 C) ('44D%ONSN C) C) C) C) C) SN

0 LA 0(l) 40 ('4 01') N N N LI)SNO N In .-I OC) 08 04')4DIONNNC)C)NNNNN

0080% C) 0.-4100'. 08 C) C) NC') C') U) ,-4 ,-1 m U) 08 0 C) 0

NC)C)C)C)NN40S

U) 4 C') C') ,-4 C)'0O% U) C) N 04U) 000 ("4 C') .-4 C) (' N .-4 0'.

4040 08040% 080% C) N hID

SN001')U) 0" bOO'S N040"0,-40% ('4 N 04040504040 0%C)08O%OO8NNN

-4U) 40 U) 0.-IC) N 40 U) N 540.-4C') 005114 C) 4C') 40400

C) C) 040085 C) S-4

C) 4') 0O%S U) N 404') 4') 4') SC') U) U) C) 400400040.-) C')NoC)C)C)r'

-45(4 In C) ('44') 40 05 .-1010 C)100 In ,-$ .-40 C) U) 0 C) 0'. U)

NN%0%C)N04SC)r-b

C) 40050C) C) 0500 N '-44')5 In 40 N 40554') 4') N NONN08C)NN04%0C)SN

U) .-4 4') N 40 C) 04') 0 C) 0' ('4ID C) C) 05 C) 4004 ("4 0% 500

55 C) C) SSN0% U) SNN

40 C) 4 C') It) It) 051050.-I4') .-40 In U) 400.'-4N 4') .0 UI

55 C) C) SSI"044040N40

.' N 4') '0In IDS C) 080,-4N-4-4-4

'0N0)-I

U)N04-4

105

>4 ONOOSa)0U,N

40ONO.-40'Lfl>4 U4 *t4 '0*00*0*0 *0 NIn 0000=000=0

'-4 q N OS *0 .-4 -N r'.NN U, S *0 N *0 Ui '0*0cx a,

N 0 5 ('540 N ('5OS 0 .-4 -I ('5 '5 ('5-4 US N

*0 OS a) ('5 N N .-4 ('5 N *0OS 000 ('5 N N ('5' ('5415

NN .-4 OS - N OS a) N OS *00 OS 0 OS N ('5 N 041505 '-4

SON a) a) = a)

OS *0 OS N OS N a) *0 a) 0 ,-4 S('5. 50 .-I *0*0 US OS ('50NNNa0cxsoso

o OS*00'00mDNC-,-IE NOS5'0a)O,..OS

-I -.4.-4 OS 00s00 .-4 ('50*005('5 iS OS ,-4 ('5*0*00 C

SN N NO 00 0000-I -I

N 00 US a) OS ('14150*00N 0 N m a' 0 .-i -iNN N 50000 OS000-4 -I -I

OS NO ('5 ,-4 4(5 N OS OS OS.-I .-4 N *0 a)N .-400ON N N N N a) a) OS OS OS 0

'-.4

O ('5*0 N 15 ('5 415 ('50 'a' OS0050 N 'a' 05*00 N U'S -IN*ONNNa)NNasosc,So-I

0 Oa'NO'5NONUS,.-4Z ON*0NOS*ONIUON,-IN 4040IO*ONNNOOSOSO-I

a)N0*-4

O 'a' ON a) *0 a' ('5 ('50 145 'a''a' 'a' N 1(4 OS N N ,-4 00

*0*040*0 * *0*0500 OS OS

OS 0 *0 ('5 a' 0 US OS 1.154(5 ('540 OS ('5 N '000(505

—4 N ('S '0415*0 N a) 050-4 ('4-I

OSNOS'-4

*0(4) O OS lii ('5-4 '0(4)40415*0 N a) 0*5 N *0N ('4 ('4*000 = a) a) =00' OS OS

a)OS N 0 ('4 N OS-I N OS 0' 0'*00=0000= -.4*0*0 OS

05 05 00 (0 05 OS 05 05

a) (45=0 US *0 N a) '0*0 ('SNSOS = ,-4 a) 1.15 005 a -

OSOa)0a)OS00s-I('SOON US 0 ('5'00 0' NO'0*04150 US '0* 0000 4(5

00 0=000 OS-I -4Ui 0= 0* N -4 *0 Nm 415051.15N ('ION 4 N ('5 N a) ('SNr a) a) a) a) = a) 00' 0'

'-4

NO.0500 ('4*0 NSa) a)USN *0 N '0 ,-4 '0 *0 US 00 ('4

-4 —I -4

OS *0 N a) a) '0C'4 N N CS ,-IN0.05 OS'0O '0 ('5 Lr N ('4 NOS OS OS OS 0' 00N00

4-4 -I-I-I-I-IUS N OS (')0 OS C .-I l'i'0 US 00(4505 ,-4 N N 0*05 SN NO00S00*OsO*00(..lN,--I -4 ,-4 -I --4 -I -I ('-(ON *0 N OS N *00 N ('4 445OS ('lOSeS N a) Lt) N 15 N0*00500.05 OS 0 ON N N-4 _4 -I-I-I-I-IS 1(4 '0*00 -I a)'0 ('4 N -I ('4000505 ('50 N *00 Na'O'00O50O'OOrnNN-4 _4 -I-I-4-I-ION a) 0 ('5(44 N N a) a)ON 000 ('50. 0 Ui NONOS Os OS 005 OS 0.050 ('5 .-4 N

-4 -I-I-4('4SO. S0 ('41.15*00'00USUS *0505= ('5*15 OS '0OS US

OS0c'sNr)('4-4-4-4

('5*0'0 US N 0 OSOS OS '0*0'0*0*0*0 N .1 #-4 ,-4 ('5 .-4 0. '00' 050504 OS OS 0.00 N -IN

-4 -I .-4 4 -4'00 0 00(44 NO OS '0 N ('4*0 15 415 US = N N OS OS (4 N ,-

O.05O.05OSOSOS.-4ON_4 -4

('S('5 '0415 *0 N = OS 0 I N-4 -4 _4

*0 N *0.-IN 004*0 OS 0* 1'S41) '00*00(450- *0.-Ia) *00

• Ii)('4

.-4 N In N 41) *0 N ('4 N US '0 ('4051') '05*0 '0145 N 4') *005(45• 0OSOSOOSOSOO*OSOSOSOO N

O0415*0NI45NNOSa'.a'>41.15

.-4 OO,-4000.-4.-404N04-4 -I .-I 4 -4 ,-4 .-4 ,-.-4 .-4 .-1 -4

S*O('5NOO*,-I*0OSO*0,I>4 OlJim.-4OSa)*0,-Ia*oa)osO.1 N,-400.-4,.-4.-404,-4,-.$

.4-4 -4 -I ,-I ,-4 ,-.4 4 -4 -4 IZ 0('),-4OS*0O4*0'0ONOSUS>4

445 000004.—I,-I.-4-4 -I ,-4 .-4 .-4 .4 .—$ .-I 4 .-4 .-4 .-4

40OSNmOSNI150*0NN>4 '0NOLOUSO**OOS.a'OOS'a' 0000*O-I.-1('4.-4,-i-4-4-4-4-4 -4-4-4-4-4-4

05050 moosm,-IUSOO>4 00NU,NN'0NOS('S 0000SOINc's-I-I-I-I-I -I-I_S-I-I-I

*0*ONUSNIflO,-.45Nm*0>4 ('lOC'S&1SOSN OOS0OS05.--Ic,xN.-4-I-I-I -4-4-4-4-4o *OUS¼OmNOUSOOSU4S>4 005 *0O.OOS*Dm'0OS.-4 .-4'00O5,-.4.-40,-4-I-4I -4-4-4-Io -I00*0Lt4'0SaS,-INOE N05,-40USN'0505OOSOS 0'0O0005Na',-4-4-I-I-Io NS,-I'0C'5OOOS40mSNE ,-.4SN'0OS*0-I0m*0 N415N.-400s-INmuim-I-I-I-I -4-4-I-I-Io O'0.--I.-IO*OON.-IONZ NUSOS'0O.-4C4U505*O4(4 N.-IOQ%,-4c9c'5'0('5-I-I-I-I-Io N0SOS'0'0c'5SN'0

o '00 0l'a'*0OSOSz(45

,-4-I-I-4 -4-4-I-I-Io N1'SSNOSOS.-IOUSO40SX('4 N'0USONOSOSO,-4(45u540

r-4,'-4,-4,-4 -4.-4,-4.-.4

o OIIn.-4NOO*OS4'54'5O40E 0Lfl*0*0OOS.000siIS,-I NONOOS,-IN'0a'o mOSOSONOSNSmma)z ONOCNNO'0NOSN*ØO -I('5('5OO'O40000mO-4_I-I-I -4-4_4

('1(4) '0145*0 P-a) OS 0 .-4 ('44-4 -4 -4

0a)0*-4

a) a) '00 ('55 US N P-SN NN ,-4 1') '00 OS a) -ISO ('SN

S NO S In US -40*0 US ('1 ('S0 ('SO OS 0540 Ci .-4 lii .0

000000 SOS NO 0-4

*0NO'0L1500SO*0*00>4 SO'0*ONUSS,-INO*DOIn ...4 '4NNNNNN'a'145'0N('5

-I -I -4 I -4 -4 -4 -4 -I -I -4 -4

>4 .-4US('5,-I40NN415O0".4 NNNI'5CSN'5'a".4'Nm-I -4 -4 -I .-4 ,-4 ,-4 ,-4 .-4 .

>4 OS'0mOS1*S.040NUUS N('SNN1'5'0'0.0('5('5

-I -4 -4 -I I .-4 .-4 .-4 .-.( .-I .—4 .-I

z NIUiOOSmOc's'a'uso>4 ('4O-II1SLfl'0,-.4'0*0,g5OS.-'0

I 4 .—4 .—4 .-4 .4 ,-4 4 .—I .—4 ,—I .—4

Q>4 9NN*0CSN*0mN*0O('S Nrn(*41'5'0u5US.,..-I -4 -4 .-I ,-4 -I .-4 -I .4 ,-( .-4 -I

N .-40N0 1'5 .-4 C 0 --4 0>4

('1 u5'0m.4 ,' .'4 ,-4 .-4 .-4 .—I ,-4 .-4 -4 .4

X NmO.-404.-4O*O40N.-.4>4

,-4 ('i('9('S('5C'5'0LflaSStjSr.SN'a'-I -I 4 -I -4 -4 -I 4 -I ,-4 -4 .-4o 0OS'0USO.a'4OmOa,05Ifl

05 Nm4'5'0O4f5,f5u5.a'N.a'-I .-4 .4 .-4 ,-4 .4 ,-4 ,-4 ,-I .4 ,-4 .-4

o 0400m*0Noa'E N.-4*00U5*0Omc')oa)*0 ('S'a'N'0'a'*0N5u'S,-4('5-I ,-I .4-I 4 .-4 ,-I ,-4 -I ,-4 .-I .-4

o ..4ON*0O%,-4.-.4u'SNNNE 05m14S*0,—4NNos5*0NUS ('5N0'0*C5*0'0,.-4('5-I 4 4 .-4 .-4 .-( .4 .-4 .4.4 .-4 ..-4

o '0OOSOLflm'0.-,Os5'0,..4Z -.4'0f'5USO400,-4541)uj5'0

,-4 -I .-4 .4 .'4 .-4 .-I .-4 ,-l .-I .-4 .40 N05*O'0m40'0'0*Ø40Na)E *0N'0.-IUS*0N'0'0NN('S '0'0NU5'0'0U5*0u5

-I -I -4 -4 .-* -I -4 -I — -I -4 -I0 N*0('540504U,004N05E 1')ONO'0*DN'0NsinN 41541 4*0US'0U50u5mO.i

-4 _I ,-4 I -I -I .-( -I .-4 .-I -40 *0OUSmOON*OUS'0.a'a)

4-4 In'0NIn40'0415u5'0moo_4 -I .-4 .4.4-4 .-4 .4.-( -4 .-4 ,-I

o SOONNC'500.a'IJSN*0E OS*0.-4UiNOuS'0405NO me'5mN*0'a"041,4N00

4-4.-4-4-4-4-44-I-4-I '-4('4(4) '0Lfl40P-a)Q*Q 4 ('4

-4 _4 -4

-Ia)0*'-4

>4 ONN'0.-4OS0',-400m0US U,("I 0O0OSNa)a,a) N

USO.4*ONNNSN'0O> >0 0 0N 00 ('4 a)O00Na)0' ('4

>4US-4

>40-II>4

445

I>4

0

I>4

US

'0

1')

I>4

N

-4

I>4

N 000\Os000000SOS-II>4

-I

CS

0'0

0xUS

0E'0

m

('4

-I

0

-I

0

106

CM .-4 .-4 0 Nb 0 0 '.0 .-4N .4CM '.00% CM

N 000% .-4 00'.0 '.0 r-_4,-4 __.-4'.0 .-4 00% 0'. 0 4 m .0 in .-lin ifi '.00% '0 '.00000% 0' 00% 0% 0000

0' N 0' ,-4 .-4 CM a' m .-40% in '.0 m N '. '.0

-4 - ,-4 ,-4 _l -4 ( -4 -4 .-4m N 0% in 0 inON C'. '.0 '.0 in 9 '.0C

N CM CM N CM .-I 40% a'. o-l___-4 -4

C in N .-4 '.0 in 0' .1 N inN in 4' in '.0 In '.00CM .-40 N '.0 N N

-4 -4 -4 -4 _.I -4

.1 (Mm It'. '.0 N 0'. 0 .-4 N.-4 ,-4 -

'0 '00 In '.0 '.0 00 .-4 min 0% .-40% N .-l 0% 0

t '.0 Nh 0'NN 0% O'.

in in '0 In In '.0 N '.0 0' '0 '0It'. .-4 0 0 '00 ..0 0'.

.-400a'O0..-4.-40000_-I_4,-4 '-4,-.-i'.0 '.0 '.0'0 4 m '00% NU.'. 0'. N N 0'. 0% '.'. .-4 in

CM CM 4 .-I N .4 4 N N N CM N_I 4 ,-l .-4 -4 -4 -4 ,-4 .-l .4 .-4 ,-4

00' N 0'. N'0 0. '0m 0m ma. N (MO '.0 C".'000'. 00' 0 .-4 — .-4 .'-'. .-$ I .-4-4 -4

m '.0'0 In '.00-4 .-4 N In0'. 0% 4 a'. N CM C 4 0.40 r'4 0 .4 r4 CM ('.1CM CM .4 Nr4 .-4 .-4 .-l .-4 .-4 -4 .-l .-4 .-4 .-I .-4

NO'. 0 CM in 0 N '.0O '.0 '.0 m 0% ,-4 .-4 CM 0' ON

0' 000 .-I CM CM .-l CM N— _-4_-4,-4,-4-4m C". CM CM N '.0 Na' 0' CM Lfl4 N N in 0'0-4 '0 0'.

40000 ,4 4 CM r4 ,4 ,.4 ,-'.r4 .4 .4 r4 r4 4.4r1 ,4 ,4 .4

h .-l 0'. m '-40'. '' ON 0'0 CM 9 '.0 NON In NOO 0% 0000.-l.-4,-4 .-I ,-1 ,-4

——-4———————in '000% CM In '0 '0 NO'. N'0

'.9 '0 '.0 C". In in CM '.0

000' 00 '-4-40000-4-4-4 -4-4-4-4-4-4-4-4in '00CM '0 '.0 in ONC". '00 in N 4.-40N0' 0' 0'. 0% 0% 000000

-4 -4 -4 -4 -4 -4

CM '.0 N 0% '.0 N N '.0 .-4 NN ('40 '0CM U') N '0N N 0'.

oao'.ocoa'o'.oo'.-4 -4

O 1 N ('40'. N C". in In C".'0 it'.'0 .-4 U'. CM in CM'0 '.0 '.0 N

0%0'0%O'0O0'.0%O%0'.

0'. CM it) 0'. '.0 N Nm 4 .-4 '.1'.N in CM it'. N'0-4 C". U". '0 '.0 N

0'. 0% 0% 000% 0% 0% 0%_4 -4

C". 00 m '.0 0% N NO' .-4 0%'.0 '0 ('1 in .-4 C". 0'. OC .-4 it) '.0

o0o000a'-4

U". (MN N moo .-I N 0C". .-40 ma'. (MN NO'. m in

0'. 0% 0'. 0% 0% 0%

'0'0'00% m in in .00 In mm.-40% N.-4N N CM If) 0'. 0

o .-I0inm'0b'.0NZ oa'momN%ONN'0..-4.-lo 0

C".

0'-4

.-4 CM m '0in '0 N 0' 0-4CM-.4——

X>4 4%0'0mNo0'.u')NNo'.U, .CM 0O,-4.-4C".mO%N,-400%0%

-4-4-4o.-4m.-4a'.4CMmm'.o>4 '00iCMinNin.-4cMNØ.No

CM -4.-4NNeimN4ooe-4 -4.-C .-4.4.4.-4..4,-4.-.4

l-4%0CM0%>4 ,-lLfl'0a'O.O'.NN'.omoa'in.-4 NNCMCMCMC'ICM.-4CM.-4

_4 -4 -4 __4_ -4 -4 _4 -4-4-4inN in0%'0'.0'0IflN

>4O ..-4 mCMC'lN4..4..-4

-4-4-4-4-4 .-4 .-4 ,-4 .-4.-4.-I ,-4

'0mho.-lma'o%'0'0>4

it) Nm'0mCNCMN.-4NN-4 __4 -4 .-( -4-4-4-4-4-4-4

Z c'40%bmN,-4'.0o'.m>4 O0O'00'0%.DNO0'0 CMNNrnC".mC'.INCI,-4NCM

-4 ,-4 -4-4-4-4-4-4-4-4-4oinmCMr-.'u')in'.OLnCM

>4 0it)U')C". .-4N4N('4,.-4.-4,-4

-4-4-4-4-4-4-4-4-4-4-44.-.4-4N'0m

>4

N ICMNN.-1r4r.4-4-4-4-4-4-4-4-4-4-4-4-4

0 0'.-4r-4OONN.-4m>4 N%0'.0a'OOifl'0.-l.- O0.-4.-4mmNCMN.-400

-4-4-4-4-4-4-4-4-4-4-4o N0%CMNm.-4'0O'.Oz 0'NO'.N0CMCMiniflOO0'. 0'00..-4c'4m('IC'.l.-lOOO-4-4-4-4-4-4-4-4-4-4-4o ONa'a'mmr-CM'0c%Ime.4b'0'.OON.-lO'.DCM'.0 0'.000C'4CM.-I.-C.-COO%Q%

-4-4 .-4 4-4 -4 -4 -4 _4 -4

o '-4'0O'.D.-4NU')0%NCME '.OOinN.-40Lt'.inOO'0',OU') 0'O0OClN.-4.-I-4OO%

-4 _4 _.-l_ .-4 -4 -4 4oZ '0O'.'.ONOInOmOoNC".'0 0%0%00-4.-lOO%

-4 -4 4-4 -4 -4 .-4 -4-4

o '0InO.l0%0%inin(MCe'. a'a'OO.4.-I,-.40o%a%

4-4-4-4.-I -4 -4 -4

oE (MN 4'.DUl%OOr-(N 0%04O0OOOOOO%O%

-4-44-4-4-4-4-4o CMCMNmo%%omoN'.ooocMr-Nv'.mN'0o'0o— 0%0'0O00%U)

4-4-4-4

'00'.-4

N U'. N '.0111 C') N (MN NO0 (MN '0 '.0 N 41)0'0 C".

4-4 -4

Nm'0 411 '.0 N 0% 0.-C CM4-4-4-4

I,)CM

NO0%CM0%'00%,-4.-4in>4 oinmLn-4('4in'.n'o.-lu')cMo('4 O.-I.-4.-400O0'N0'.a'

.-4 .-4 .-4 .-4 .-'.'0ON4D'0iDOCM..-4NmN>4 '0r'lm.-INOO'.No'.ONN

U,.-4 .4CMNCN.-4('4,.4N.-4r4.-4.-4-4-4-4-4-4-4-4-4-4-4X mo'0c-o'.oN'0oine>40.-4 -4N0.-I.-4.-400'.-4-4-4-4-4-4-4-4-4 .-4 -4

>4

in .-4NNN4,-4.-IOØ'.-4-4-4-4-4-4-4-4-4-4-4If) ('4 '.00% '.00'. 00CM U'. 4>4 '.0mNNLt)C'4.-4N0%0N

'0 .-4NN00.-l000'.O-4-4-4-4-4-4-4-4-4 .-1

Q otn.000'mo'.ca%mNo>4 moN'0m8no.NCMmm -..-ooooooo.o'.

-4 -4 -4 -4 4-4 -4 ,-) -4-4-4

>4 a'Lnma'Nc'4'.oemN 00O0'.04O%-4-4-4-4 -4

-40' in .-I CM N '.Dma' ('40% C".0% in C") in a. .- N C'4'.0.-4

0% 0000' 0'. 0' N-4 -4 -4

'.0CM '.0 'ON'0O Oc..'.In ,-4 0" '0 '0CM it'. ('lIne 0 '.00% 00'. 0'. N

-4

('40CM N %0'0 0' '0 N in'0 N0'.0 '0 N it) 0.0 a' in0% 0.0% N N N N N N N

in '.00. '.0 ("40' inN U') CM '.0'00% 0.'.0 N in'00'0'NNNNNNNNit) in N 0.0 N (M'CIn in 0'0in in .-4 N %0'0 0. Nit) in '04".0'0.NNNNNNr-N.-4 C". .-l 0' (M'0*OO C".U". in0 In in C". N Ifl'0'0 '0in

0% 0'. N N Nb NN N N

o in a' mm 00' '0 '0 in N'.1'. 4". N N In N in (Mm'0 0%

0'. N Nb N N N NN

0 in '.0 in (MN 0 '0 '00'.4". N CM 0' N .-I ('Im 4". 0

NN N N NN

o 0'NN'00'0%Nin(Nin'.00'.Z C'4t-'.0N%0'0mQ0.'.o.-40 N'.ONN'.0'0.ONN.OinIn

(Mm 'Oil'. 'ON 0'. 0 N-4 -4 -4

0CM

0-4

in

>4

'0

it) '.0 N N m '0(1 in 00% Nit) NO N 111 . in 0'.'0 N '.000 0% '.0 ("0 '.0.0 '0 It)-4 -4

>40N

>4in-4

>40-4

>4

in

>4

'0

>4

>4

N

-4

0'

'.0

U'.

'0

m

CM

-4

.-l N in .'-I .-'4 in '0 in N In in

..0,-4N Lt)'.D0 '.0 m 0 0.-4mm mm m'0r',m ('4-4 N-4-4-4-4-4-4-4-4-4-4-4-4-4.-4a' .-s C') N 04.-I C". -400'00% in N ,-6'.0 ("4 '.0 '.0 0

mmmrl.-lo-4o-4-4-4-4-4-4-4-4 _4 _,-4_x 0'.'.0.-4LflNLOr-40%N,-INO>4 ONNN'0NON'0.-4C". mmm'0C".C..4,-4,-4.-4

-4-4-4-4-4-4-4-44-4-4-4-4'0in'0'00b'0.-40>4 0".O.-IN'0'0'0NOO'0c.i

N mm'0mC".'0rnNCM.-loo-4-4-4-4-4-4-4-4,-I .-4,-4,-4'.OinN'.ON'ONNOO%O...O

>4 N'0CMOO0'.c..4'0C')0'.CM— O'0'0'0'0C')mN,-4O0

-4-4-4-4-4-4-4-4 _4 -4-4-4o %ONNO%CM0%'0O'.ONIfl'0X N0%ONinininU')O,-4'.0.-40'.

4-4-4-4-4-4-4-4-4-4-4o O%0N'.0in'00'.'.0m-I.0'.0 '0mmmN'Om.-4oo.o'.

-4-44-4-4-4-4-4-4-4o ON%0Nmc'INNIn.-4.pNE oC")oLnNin-4'.oO'.'0minin '0'0'04')C'4'0C".,-4O0'.0%

-4-4-4-4-4-4-4-4-4o N.-4'.D.-4C".,-4r-4mE 0N.-INinIn.-ImN'0InIn'0 '0mr'4'Om,-400'.o'.a

-4 -4 -4 ,-4 -4-4-4-4-4

o inmma%.-40'0.oinr-'OoE NNO%'0m'0'0CM'.OON'04". m'Ommc.a'0e.4oa.a'o%

-4 -4 .-.4 -4-44-4-4-4

o (M-4O%0%%0'O0%,-4'0'ON('Como%o%-4%ooN'.oN('4 mNm

—-4—————o mo'..-4a'oo,-lmNmo%oin .-I it) C". N N '.0.-4N 0-4.-4 N('4mN.-4c'4a'NNN

-4-44-44-4-4-4

0

>4

-4

0.

'.0

in

'0

C".

('4

-4

In04-4

107

N 'U LA N to 4 to 0 UI 00>4 UI N N 'U N _I 0% UI NO ,-4 >4 0 UIlit . UIN to tO to lit 'U N N m 'U 'U mm N 'U 'U

'UmUINNmo%moatot —>4 40%'U0%IAtONmLnfI4N >4 at0 • 0N NNUt0,-40%0%N0%0t N_4 _4

_IUUltomatto'UOmNN NOt>4 LaImc4tONNatat'Umm >4 tomUI • UI,-4 0OtOt0%0t0tOtO0 _I 0%0t

-4—eNotoo'UtnNNtcmm

>4 UINattoN'Uo%0mm >4 lAm0 • 0_I 0 N N N 0% ,-4'-4

_I N to mUlCt 0%>4 UI N mm . UI N 0% UI 0% 0 >4 'U 'UUI 0%NNNNNNNN UI NN

ate to oat oe o N -4 m>4 m tO atm UI N N Otto to >4 NO'U 0tNNNNNNNtoto 'U NNC 0% ('4 .-4 0% NOt 0% 'U .-4 to 4 m>4 00' UI m 0% UI N m UI 0 >4 ON(_t OtNNNNNNNNNN 11 NN

0% ('1 'U m UI 4 NO N fI>4 m _I .-t too 'U N m 00% m 'U >4 N UI

N NNNNtONtotor- ('4 totO

tOOt_I .-l N 'U N to_I 'U 'U to UI>4 'Umat UI .-l N ('9 _I 'U >4 0 m— NNNtoNNtoLfltotOtotO — toto0 LnmN'UUImototo,-4'U 0 NmE to'U0to.-ttD'UNO0%O,-4 E 0_I0% NbNtoNtOtoUItoUIulto 0% tOto

0 lN4C'l0%'UNtOtototo 0 tOOZ %OmUI'U-4OUI0%NtoNto NNtototototOUIUIUIulUI to UILA

0 m 'U 'U Ot.-4 _I NON UI 0 4'UN'U'UNNO%r'ItototOUI NNtDtotDtOU1UIUIUILAUI UI UIUI

0 toUt('9UIQ,NUI('4NOt 0 OtN,-4'U'UNC9O%,-lto'Utoto Z toto

'U N N to to tO to UI UI UI UI UI Lit 'U Lit UI

0 0%UINtototo0tOUIm 0 .-4NZ NN'Umtoc'9ee4lrtmtoo Ntoll NNtotOtototoUILflUILitto m UILA

0 m atm to UI m at 0 Itt 0%'U'UmNUI.-4,-4UIm'UUIm LEl

('4 N N to to to to to UI UI lit UI to ('4 UI IA

0 UI Otto N UI to 'U In N mat UI 0 mmE _I .-4 'U_I m 0 N N mm N N UI,-4 NNtototOtOUIUIUIU1UILA — UIUI

0 -4mcq_Itor--'UUItoN 0 m'UtI IAN _I ate ('I 4 to 00 N 'UO tototOLittDtflUl'UUI'UUI'U 0 IIIUI

_INm'ULntONoto_IN-4-4-,

NUt-4

108

SD U) ('500 N sO ('S UI s-I NN N N CM CNN CM N N S'l (5)

N N N N CNN N N N CNN

00% Na)a) ('511) ('10000-40000-4-4.4-IN ('SON (MN N (N ('INN N N N N

00' a) 0% a) NO a) N N S'SSU) 5555 U) U) S U) 050-4s-S .4 s-S s-Ss-I s-I s-Is-I N N

U) 0 SO s-lOS ("5(5) UI 0' C'S CNN('50 ('50 (.100 ('SC'S UI sO N

s-I s-S s-S s- s-

50 U) 0005 NO sO 0 UI N 50(MN N (SN C'S ('IN (.10511 sO

-4-4.4-4-4-4-4-4_s-S-I-I

N 005 SOs ('5 U) N 50 (MNs-S s-I s-I s-I s-I — (Is-SCM ('50 UI

s-I s-I s-I s-S s4 s-I s-I s-I s-I s-S s-I s-I

051)00CM ("10 N 1(5 U) NO050505000-4 s-S -4CM ('5(51000s-I s-S s-I s-b s-S s-I s-I.4 s-I

('4-40 U) ('5 ("IN 111 U) U) Os-INNN N U) U) 0500.-I N s-S

000000 -Is-S s-S s-I s-I

UI LI) U) s-S N 'OSCM 5005005050 SDSNN U) 00.-I s-I 000 000 0 0 s-I s-I s—I s-I s-I

0' N s-100 U) ON NO s-S NUI 141 sO SON 0 U) 050-4 s-SOS

0 00 000 0 0 s--I s--S s-S 0

5000' s-SN UI N UI-I 0' 0% UIUI US LI) SD 0 '05050000%0 000000 0 s-I s-I s-ID

Ns-I 5115(51CM ('IN U) 11 SD (.1UI UI UI LI) SO O NO' 05000%

000000000 ,-I -40U) N s-I N NNNSO UI ON s-S00 LI) UI UI LI) sO 0' 05000%

000000000-1-40('I NOUI NO' 0' 500' s-S UI 0000000 Sf1 N 0' 05050%C 0 C 000000N ('5 NO 0% ('10' (s-SOC'1 s-I 0S'S ('1 ('1(51CM ('1(51 UI S N 0%

cCCOc00OOsO 0' (510' 00' 5050CM UI 0% ('1s-I s-I s-40000 s-I UI UI 0%OOs-I N (.10 UI sOS 0' 050 s-IN

-4.4-4

00 U) 0'005.4-4.4-40 a)00(5) ('5(51 S'S0000-0 (.1N N N N CNN N CNN N N N

0'500(5)41)00'0'05-4 N U)s-I s-I -4.40 s-S s-b s-I s-I N s-S s-S

N NN N N N N (.51CM N CM N

0' sO ('S s-SO'. 0550.4 N 500% ('SNNNS 1155050' 0' a) SNs1 s-I s-I s-S.4s-S s-I s-I s--I .4

500 (MO U) a) UIs-IN 5005 C'S505050500 UI sO N SN 5050s-I s-I s-S s-I s-I s-I s-I s-I s-b s-I s-Is-I

.40050' S UI C'S 0' s--I UI 0% NUI UI 0O(')051I UI 5050141 UI

s-s-s-Ss-Is-N ('S (51 C'S U)s-SN S'S 0' .4500('IC'S ('SC'S N ("5(5100 UI 00.-Is-Is-Is-( s-Is-Is-Is-Is-Is-Is--Is-I

U) s-IOU) NUISO s-I 0' 0' 0' 0'0.4s-S s-S s-Is-Is-I (M N N CNN- s-I s-S s-I s-I s-S s-S s-Ss-S,-S.4 s-S

(MU) 0' 00% -40 UI N N ('SO000 s--SO s-I s-I ('4CM (MN

s-I s-I s-I.4 s-I s-S -4.4-4-4-4-4

500'S') UI Li) SOO055050NsI0.05000000 s-S s-Is-I N

O s-S s-S s-I .4 s-S s-I s-I s-I

ONs-IO0 LI) (MN 0 C'S '0 0%0505000000 s-I s-I s-Is-S

Os-I s-I s-I s-S s-I s-S s-I -4 -4 s-S

CM SOON ('SC'S-I It) CM s-SO U)050,000000-4-4-4-4O0 s-I s-I s--I s-S s-I s-S

s-I UI 0' 0CM ('40' 0005CM SO0505050000.0.40 s-b s-I000-4-4-40-4-4-4-4-4

CMOS 05000' (.10% 0' .400.050505000' 000-4-40000-4-40-4-4-4-4-4

SON U) 0' 0% 0' 0' 500' 0' Os-I0505050' 05050' 000.-Is-I

000000.4-4-4-4-4ON 00' 5 UI CM ('I 0% s-I s--S 500000505050-40.4-40s--I s-I -4000 s-S s-S s-I s-I sSs-I

s-I N (.10 UI 5050' 05 Os--IN-4.4.4

NO '(I U) UI a) 000' 0'NOI'S ('S ('5 ('54'5 CNN N.4 s-Is-I s-S

N N N N ('IN CNN (MN (MN

-40 N N SONO 0' 050' 0' 0-4-40000.0% 0' 0' 0' 0' U)CNN N N N .s-I.4s-I.4s-Is-I

N U) C'S ('lOst) N 00-4.405050505050000 000s-I -4-4.4-4-4.4-4-4.4-4-I

N a) ('5'5oa' 0' -4-40UI UI UI Lt'I UI ('SC'S CNN C'S Is-SC'S

s-I s-S s-I s-I s-S s-I s-S -I s-I s-S s-I

500' (.1 ('500050' 0' Os-S s-I00000 N s-I s-S s-I N N CM

s- s-I .4

UI 50 ('5 (.1-ISO NON 00(5) (5) ('5(5) (.1-4.4-4-4.4-4-4

s-S s-I s-I s-b s-S.4s-I s-s s-S -4 s-S .1

UI (.150 ('10 0' 0500% s--I UI NN CNN N N s-S 0-40.4-I s-Is-S s-I s-4 s-b s-b .4 s-I s-S s-I s-I s-I .4

N 000(510,000 s-I SONN (MN N N s-I s-b s-S s-I s--I s-S s-S

-4-4-4-4-4.4-4-4-4.4-4-4

O 50N 0' s-bOO 0 s-I s-I N NN s-I N .4 N N s-I s-I s-I s-S s-I s-I

s-I s-I s-I s-I s-I s-I s-S s-I s-I.4 s-I s-I

0% UI .-IN000 0% -4-INNNs-IN N s-I 9 s-I s--I s--S s-I

s-I s-I .Is-S.4s-4s-(.4 s-I s-I.4 s-b

0' 0050050500% s-I.4N Ns-I s-I N s-I s-I s-I -40 s-S s-S s-I S

s-S s-I s--I s-I s-S s-4.4 s-b s-S s-Is-S s-b

5(51051150' 0' 0' 0' 0050.-Is-I s-I s-I sI s-I s- 99 s- s-I.4 s-I

s- s-I sI s-I s-I s-S s-b s-b.4 s-S s-b

UI (.1500 UI (.1 UI UI 0' SN QIs-I s-I s-I s-S s-S -40000-40-4.4.4.4-4-4.4-4-4-4.4-4

('5 Os-IN 0' N 500' s-I 0' s-I C'Ss-I .4.4 s-I 00 0% 0' 0 0% 0 9- s-I s-4.4 s-S s-IOOs-I0s-S s-I

0550 C'S 00.40,500' (MN ('50-40.400' Na) 0' 0' 0' 0's- s-I s-S sI.40000000

.4 N ('SOUl sO N 0' 050-4CM-Is-Is-I

U)CM

U)N

I>404.4

'N• 0('4 N

1>1U)s-I

0.4s 415N .4

1>40s-S

0a)• 0s-I -4

I>4

s-IoUIUII>4

OsN0 s-S 0

1>1

0-4

('S-(.,I>4

UIas

(MONI>4

NN0

0Z UI 0sO E0,00%0E

a) 0sI) X500500Z LI)

UI

U)OUI0 NEU)000X0 0OC'S 0 ('50N

NOXOE0 N

0X

(NC's-40-I

0E s--S

0 0 0N-4

I>1U)N

I>40N

I0-II>5

UI

I>4

0I>1

("5

I>4

N

I>5

-4

0's

0Zso

0ZUI

0E

(5)

N

-4

0

a)as-I

0(54

0-4

UI

-0

('S

N

s-S

a'

so

UI

0

(.1

N

-4

0

0%0Os-4

a)I—SIU) N

U)

o I

050so

r_, '—.I—IZN>S-I

ox04 E

—.0(V OZI—

SO NO 00% 0%s-S -4

109

ONNNO.-4>4CN

Na'O%a'ND>41(1 .I NNNC4c,9NN

>4 4ULflo.-4 NcN.lN>4

u• C'4NNe4

00O.-Im0NO%ND>4 NN0000a'a'000..-i

mNN4a'0Na'>4 OOO%Om

NflN-1N>4 NNOOON 4r4,4(4

>4 NNNNa'4 -4--lo OO0¼DNE NNbN¼Nroa' 4,44-4

o 0a'OQ0ua'E I1NNONNNN.O

o 0a'Omc'Z NNr-.DNNNNIfl

o NNN-4,--4Qz

o, 4ro c-,cT.,o: DUflNN -1,-i,-

oE-4

o OOu1uz NNOijo 00OO.-4,O,4,4

.-4 N f' 'U D a' 0-4,-4 .-4 .-4

'-4U,a''-4

>4U,N

CM,-IN'I?>4 Nr-sr..0c r4Nr,N

> 4ONNNU,.-4 NNCINNC4NC4

-4NNm,.44fl>4 UIñLflo.-4 4C4NNX>4 0'mmmU,

z>4 00Om. NNN

N-4O4fl.-40N.4NO>4

m

NO N U, U, m -4 N N 4>4 a'a'a'OcN

X a'&flN,.4a'a'0(.>4 NNa'OOOO,,-4 'I-4.-4,-4.-4.-4NC4NNC4N

o a'NNfor-E NNNa'a'OOa',0%

o NONN,-4Z NNDN0%Oa'a'o'.0

o a'NNN,-4Ø%Lfla'z '.0N'.0N0%0%a'00U,

o '0ONz N0%O%0O

o N 0' N a' ma' -4 -4 a'zm

o U,OOr4NEN 444No mLnO4'.0a'Z mmNN.InNQ,-4 -4.-'.-4,-4,-i,-oEo

N In '.0 N 0% 0 -4 N-4-4-4

NU,a'-4

'.0004>4 -4r40In

>4 .-4C'40O,-4a'Ø'.OOeq

>4 NONO0,.4Na".In ..-l

a'a'0%NNO%,.N>4o.-4 NNNNe)NNmNNNN

a'InU,.q'IJ,mOO>4 N0NNNIn NC'4NNmNC4NNNN

>4 00%D%DN.4'mq,.-4NNNNmNNNNNNN

N0%Nmmo'.>4

m NNNNNNNNNNNN

>4 -4NNLflNq.,.40%a'N NN4Ne4>4 0NmNa".N'.,-4 NNNNNcqNN,-4,,.jo NOOm'.boa'OE ONInNN'.O'.0% NNN4NNNN,.4,.o mbo%oo,4'.cNa'NNO'-40mIt4NNON,g,'.0

o N'.0NNN0%mNq.a'a'z 0.40moN,g1It) C'4NNNN,-4o ON,-4'.0a'NoE O,-4Oma".mo NOInmNIna',..mN4' -4NNC'4C'1NN4o e4mNa'0%a'I,,E 0%-4ONN0..4N

o NN0%'.0m,-4.a'NN,E Oa'O0%'.00%NNa'.'Nr-4

o mNN'.U,'.a'E a'NN.0'.DmOo --40O.4.4N If'. '.0 N 0% 0 4 N

-4

mU,0%-4

0N

0'-4

In

..

m

N

>4

-4

a'

'.0

Lfl

N

-4

0

0U,a'.'-4

0% N In 0mme' N 0%'. 0%'. 01N CM C4 mm mm m mmNN Cl N N N N N N CM CM ('4

'.00'. CM N N 00'.0% a' 0'.00000 -4 ,-4 N 4

.-4 ,-4 ('1 N CM CM N CM N N ('4

'C NO In 0 '.0-4000%' NU, It'. LI '.0 U 'C N N N

,-I r4 ,.4 ,4 4 ,4 .4 4 .4 ,4 ,.4

In 'C 0' 4') 0%' ,4 .4 NOmm m In '.0 N N '.0-4 4 .-4 ,-4 _4 4

U) It'. N 4') .-4 CM 00%NN N mm mm U) '.0 '.0 U)-4 ,-4 _4 -4 ,-4 ,-4 _4 - 4 ,-4

'C NO '.0 U) 0 U) '.0 CM ('4-l .-4 N Cl CM 4') Nm In U, It)

U) N ,-4 It) 'C a' 0' '.04') (Ma' U).4 .1 CM N N N N 4') U-, In-4 ,-4 _4 -4 -4 -4 -4 -4 -4

'.1'. C') '.0 0 ma' CM 0%-4 .-4 ('4 (MN mm c". '0 In '0it)-4_I 4 4 ,-4 ,-4 ,-4 -4 4 _4 _4 ,-4

Ca' '0 NO .-4 C ('I U) 00 '-4- ,-4 N ('IC') mm It) I!) '.o

-4 ,-4 -4 ,-4 ,-4 -4 _.4 -4 -4 -4

'.00% '0 N .-4 '-4 N ('I In 0% 0,-'..1-4CM C m m '0 '0' In '.0-4-4-4-4-4-4 __4_ -4 -4

'C a' 0 N 00 N N 0 0' 0'-I '-4 N ('4 mm m '00% o- '.0

4') In 0% 0% '.0 .1 m '0 N N-l .-I ('4CM CM N 4') '00 '0 '0U'.-4 -4 _4 _4 -4 _4 .-4 1

4') '.00% N U) It) ,-4 NO N ma''-4 .-4 ..4 CM N CM mm o m '?0

NO ,—l m '.0 It) 0%' U',0-40 '-4 '-4 '-4 ,-4 ('IN.-'. 4", ('4

NO 0% NC') 0 '.000 m N 0'0%'. 0 0'. 0% 0% 0 .-l ON0.-400000,-4,-IQ ,-4 0

('44') '0 U) 'C N 0% 0 ,-4 CM-4

110

LACl

0C_I

0'-I

I/S

'0

m

N

'-4

as

50

115

'0

m

N

'-4

0

ONm'0NIA'0'ONN'OOS>4 WWWWWWOs0'soswosw >4IA . LAN C_INC_INNNC_ICINNC'IN N

0 LA_I '0—C N N 'OI US c'so>4 WWWWWWOS0'0'W0405 >4o .0Cl NNNNNNNNNCMNN N

>4 NWNWNWOSOsO'wo.osU) • I/S.-4 NNNNNNC'.INC_INNC_I '-4

moasuIN'0NInWoOSN>4 SON 50 N NW 0405 W 05 >4o • 04-I NNNNNNNNNC_INN _I

0%0SDNNU5N0%W01I5W X>4 N'O'OLALISSOWWNNWW >4IA C_INNNC_INNNNNNN IA

WOOS 40 IA W IA InC_I sOON Z>4 ,-4'0445'0'OUSNWNSOWW >4'0 NNNNNNNC'INCINN '0

II) _I04W 14)50 LAW _I04 c'SSD>4 omNmm-OSOSsOUSWW >4145 N N N N N N N N N C'4 N CM ('5

'OSOmuSWWO%soSOo_I'O Z>4 W_I_IN_INOsO'OU5W >4C_I _INCMNNNNNNNNN N

SN W LAN 5005.-lW _IOsC'S>4 '0WNO%W0s,-I'0Nm50N >4_I _I_I_I_I_I_ICMNNNNN _I

o IAOInUIIIS'0WC'5c'SI/SmW 0E mSSDWSW0mNNSO50 zOS _I_I_I.-4_I_INNCMNNN 0%

o N%O(MC'SN.-lsDC'SO%WIn_I 0E NLISIISSSONO%N_I,-ILA5O ZSO _I_I_I_I_I_I_IC.4NNC_Icq SO

o N_IN05NSON0%NUS.-I05 0,—IL15'O5OInSOQ4_I.-4_IUSU)

In #-I,-4,-4_I.-4C.JNNNN 1(5

o 44SLIIc'SL/SOONLASDmNSO 0E _I'O'0SOInSOW_I_I_I'0Lfl E'0 .-I_I.-4,--I_I_I,-ICMNC4NN '0

o wasome'souso.—im 0Z ommsO'0Inw_I.-s_I'0In Zm .-I_I_I_I_I_I,-4CMNNN(4 445

o N05LI5InNC')U1'0'0N_IO'. 0E oNmInm'Or—o.-som'0 ZN _I_I.-4_I.--4_I_INNNNN N

o 0 '0 m 0 LAN N I/S IA In_IN 0_I_ImIA_INSOo%_Io_I'0 Z

_I _I_I_I_I.-I_I_I_INNC_IC4 _I

o 0%_Im'O_I'OInmsOc'so'0 0,-4051'S'OQ%OUSW_IOWl'S E

0 ,-40.-I_I0_I.-4_INN.-IN 0_I Nm '0 IA 505W 050,-I N

4-4 -4_I

SoLA0''-I

US W NsQsD 445 sD_I_I 115(15W 04004040 CM 4 Nm'0

(IN N N ('5 44) (45 m m

'ON 0 IAN 'OW IAN SO In SDW 000505 ON_IN m IA(IC_I 44514) C_I C_I m c's c's mc's c's

I'S SD 'OLA 0 115 m N SO U5,-4'0w w 000504_I m N'050N C_I c'S m N C_I mm l's mm c's

_I m N500 N 04_I'ONO_IW 000%0S_I'Oc'S '01155

N N mm N N mm mm mc's

N%Osom 'ONN'0 N_I m _IN N O_I04 OS N In '0 I/S sO NNC_I c' C') c_I C_I mm mc's mm

W'0 USSO 1(51(5 N'O 11)050_I5ON0_I0% 05 N IA '0 In SOSc' C_I rn c's c_IN m c's c's mm m

c's_I m _Is0 N 5044505W ONSD NO N 0505CM 115.0 'ONS

N CM C ('S N N C C' 4 mc's l's

11550 W N i/S 11505050.05US 400% N 05W CM '0515 'OS N

(IC_I ('1445 CM N m c's c's mm c's

1(5.0145.0.005mm mm 050'OIA NO W 5005CM mm sO W

(INN m N CM N mm mmc's

.-405 N'O .-4 m 040 C'S '0,-ISO'O'0SOO%W SON_I CM N sON

N C_I (INN C_IC_I c'5 i-sm mc's

S.-IW N '01150 m _I0 '0 m145.0 '0 W bUS sO 0%_I_I '050

N N C_I N N C_IC4 C_I 445445 mm

sOW mW_INN 50 LII If) sO'Omm .055 In In W 90141 InN NC_INN CM NC-i mm mm

It) 'OW 'ON W 'OWO W SONc'sc's m bsD '0 'Oh 90.CM'0Cl N C_I N N N Nc-i m mm

(41 W NON In 1150% 1450% In 0%C's C_I 445 N.D '0141400' .-4 N

CM C'4 N N CM CM (1CM CM (4 mm

,-IW 50501410 'OW '050,-I LIIC's_IN SD ItS 'ON LAWN 0.-I(I N CM CM N CM NN44 c-imm

W Somr- m W 00S0,-IWN 05(440mm o 'ON US WON_INN C_INN C_IC_IC_I Nm

'00 m _Im 'OW NON c'50(ISO_ISO_IN W _I In CM In _I

N .-I (INN N_IC_IC_I N CM C'S

.-I N '0 In sON W 050.-IN-4_I_I

_IItSIAOm'0t—O'OOSN>4 NNN'0OIflIflSO50500usC'. mmmmmmc'smmmm

mS_I'0_Ioos_ISO(INc_I>4 ,-4NNm'0IA'OLASON50(-4oN mmmmmmmmmmmm

US_IN_Ic_IS115Nc'5.-4LA(->4 4 C'S (1.0115 SO SO SOW W 1.5 .1IA_I mmmmmmmmmmc'sc',

Z C_IsOC—mW0I05,-4_INIAus>4 CMc'SmIA40WWWOOSmOO_I mmmmmc'smm'0mmm

LflOSNSNNNNOIISC_IIg5>4 Nm'OLnsOwaswOos'0WIA c'smmmc'sme's.OC'smc-.

'OomSO'0bNoc'sr—us,->4 N'0'OIn50SWWOOS'0W'0

'0N'OsOO'O'OUO_IWW>4 N'O'0LISSDSWW_IO'ON445

SOUSIA'0sQ'0'OOs_IUSWN>4 N'O'0USuSSWO'NO'0NN mmmmmmmrn'O-om

0 NNSONOS,-IO_IW_IUSN>4 C_I'OC'S'O'OSW,-_IO'0N_I mmmmc'smm'0O.Oc'SN

o SO40Nm'0Ln'0sOo'OmWE N'0N445.0SONO,-IO.0r0. rnmmrnc'sc'sm'O'Omc'scI

o mc-4sO_Iw'OCMoInmbOSE N'O_IN441115s00'osWmr-so c'smmmmmmmmc'SmN

o _I05_ISsOO%NN05W'0OSE Nm.-I,-Im'0LAWWNmN'A c.Smmc'sc'5c'Smc'sc'sc'srsc-.i

o WUSNCMm'0_I,-I_INWO%Z _IC'50,-4m'0USN-WNNN'0 c'smc'sc'sc'sc-,mmmmc',c,.

o 'O04_INOSONWOUS0SWX ..-4Noomm'OU-ssso.-ISm mmmmmmmml-smmCM

o OWIACSs04000c'S'O_Ib_I_Io'OC_IC_Im'OlnInoN

c-I c'sc'SC_Imc',c'smmmmmc_I

o 50'OO%e'i.-IW%ONmSOc'smz OOWOC_IOO0CMc'5sON.-l (SNNm(4)c'Smmmc'sNC_I

o IA05_IIn'O_IsOOWOsDSOE omwo_IwsoLtss_IOsOO mNNC'sc'SCMNNNmNN

_I Nm '0115 %ONW 050_IN-4 _I _I

NIA0''—4

004_IW sO'O_IC'sm W SNW US sOIlS sO IA I/SIn US 1155050

N N N C'l (INN N (INN C_I

W N N 040055005_INN50 IA In '040 '0 '0 In '01154040

NN NN ('IN N N C_I NN N

WIn_I me NW '0L(5mW 04'04'S 115 '0 m'O'O'O'0'0

N N N N (INN C_I (IC4 C_I C_I

w Nc'SN C') IA O%_IW_I m 0OOS05NOW W05 05_I_INN_I_I _I N .4_I_I '-4 N N C_I

LAW U) 05145 '004_ION 05W05 N N IA W 5040 NW 05050_I_I_I_I_I_I_I_I_I_I .-I CM

W SO 'ON W _I'OW 'OW 05WS ItS In '01(5.00 '04055W-4_I_I_I_I_I_I_I_I -4 -4_I

m '0.0mw 415401/S In US 445 _Ius mm N N .-4 _IC'i'0 In In 50-4-4-4-4-4-4-4-4-4,-I .-I I

Os N '00440 In 0 '0050 In 0%_I_I_I0%05W 050 N N N CM.-4_I _I0000.-4.-4.--4,-4.-4

05005_INN '004CM W We.--I000SW NW 05_I_I_IN,-4 _I_I00000.-i_I_I_I

0,-I'Oc')O' W 04504450500000W ('-SO S 0400_I_I_I,-I_I00000_I.-4.'-4,-I

SO' N_I 40505 LAO SOWN0.040W NsONO%000000_I00000.-4_I ,.-4 _I

LAN_I W mc', In IA C— CM Lfl'O05050 N SSOSO% 05000OO_I000000,-I_I_INOO ItS 05_I m 50 'OWN_I04040 N 504050505050000_I0000000.-4_IN N 04-4500_ION 44) W W05 05 055 sO 50 NO 04 05 0% 0%

0000000-40000400505 NON W -40115 m In05W 045050 SO 50,-I 04 W 050%

0 C C

U) '005 N_IN '000CM 50CMO W 0450 us 50404'S 045W 05

_I000000_IO 000_I Nc') '01155Db W 050_IN-4_I_I

'0IA05_I

IALA05-4

11

P4NP4Di'0LflCIIAOODi>4U,N mmmmmmmmmmmm

>4 C"4N('4,-4.-4r4InNNID%Do('4 mmenmr',mmmmmmm

>4 ('4-4.-400.-4mNN'O.0U,.-4 mmmmmmmmmmme.,

>4o.-s mmNc'4pi('4mm

>4 IOIn'0mIAIDN%OIOIn NNNNPIC'4Ne.,mme.Im

G('4tOIANON'0ONO'0>4 NIn'0N.-4'0NIONIO10N'0

01IDNIAO%N'0'0NmInm>4

m

'0'0DIO%LDONLAONNIA>4

('4

'0O'OION0NIO'.ODIDI>4

'.4 Nmmmm0 LA'O0%N'0'.4O(NOmDIE 0%IO'0m,-4'.4C'm01(...oQ% 4-4.-4'.sc'4mcI4me.,

0 NNN.-4mInOmmNE s'0mN0101,-4IONN0%ID

0 '0.-4Ln'0mLAN'010IO.-4E IO'0NNOLA,-4N'.401Ill

0 IO'0'.4'00%OO%'0NN'0Z L11m,-4NNNO'0O10o'0

0 DIm'0Nm.-4InE '0N.4,-4'.DN0%r.In01Nm Q0OC.4NC14NCI

0 '0OLA'.DO%(NmONoNE '0(NO'.4'0NNIAmInION -I4,-4000(MN(4p4('40 IO'0OI%OLALAONIANONE '0.-40%.-4NNN0%cO01'0'.4 0.-I000.4'.s.-4.-4('4

0 mo'0O'0'0LAocoo O.-4000,-4.-4,-j0p4

'.4 N e.1 '0 In ION 010-4 N-4-4-4

,4.4Di.4LA.-4ID.-4P410m>4 OO01O0'.4O.-s,io.m >4IA U'SN m'0'0'0 N

.4IDLA'0NNDiIAm,>4 ODIDl0O.-4.-4.-4..-4,.-4N.0 >4o •N '0mm'0.,'0'0,O N

DINNmNN.DNNDINID>4IA . IA.-4 mmm'0'0'0'0'0'0...O -4

DiDIN0%'0P4.-4Oe1flN>4 01NDi'..-IP4.DtnU,U,IAN >4o . 01-4 mmm'0'0'0'0'0 .-4Dl'0O0N'OON'0.4 Z

>4 Dl N 01.-I P4 '0 In ION 'ON >4

US mmm'0'0'0.oo'0 IAIDIOIONN0ICNN,.4O,0>4 0%N0.-I.-4mIA'ON10N >1

'0 mmm.o'0.o'0'0'0.0 '0

0%-4Nmo'.4Dl,-4o'0,- X>4 N 010-4mm In N '0 >.I., C.,

m ON N N '0In DI 'ON In>4 NIODI'.4m'010m01 >4N (.4

OO.-410%OOIONIn'0N X>4 '0mIA'.oo%'0IADIIn0%O >4'.4 mmmmmm'0'0'0'0'0In .-4

o N0%OI0I.-LA'001Inm 0E (N (14 '0 '0N 0% (14 In 0'. LA 010 E

0'. mmmmmm'0.'0'0'0u.. 0'0 0"001NOLAO'.UION'0'0 0E O'-INmIONNmo%'0a'. z'.0 mmmmmmm'0'0'0'0'0 '.0

0 ON0%C4NNIONIONNE OOIOIANNmNIA mmmmmmm'0'0'0'0'0 In

0 01NIOONNOmmONN 0E 0'0.-4N'0m,-4LANDN Z'0 NNmmmmm'0'0.o'0.. '0

0 .Omo%'0mIO'0In.-SIONIO 0N0%O,-4O0'NOIAIn Zm N("JNc'4c)c)mm'0.o.o'0 C)

0 .-l'0(NONmIn'0O01mN 0Z 10IONa)01NN01N(NN E(N NNNP4NC.4p4mmm.o.0 ("4

o NN'0IAN0'0Nm 0E '0mLANNN'0'0IO'.IIAO E'.4 NP4(NNP4N(.4mmmm.o -4

o N'0'0OON'0m.o,..4N 0E mo'0r-No%..4'0.,0 Zo N('9('4(Nr-4mmp,u 0

.-4 (N t') '0111 ION 0%0.-4 ('4-4-4_I

0ID0''-4

112

'.4 '0111 0,-s '.4 '0 '0 NNO N .4 DIN DIm i. s', m m

DI m IANON 11) '.4'00i N .-4I)('JONP401 N

m 01001100.-s 10 ('I -40101'0ON .-4 '-4 010% N'0 '0 '0ir me., me., mm

In ('410 In10 0' '0 DIN N '.4ID'0O m .4 .-IN N DIN

ID DI11 0 In IA N '0 -4N In0 '04') 010111 IA ION'00'0'0 '004') m mmmm c.,

'00 1110 e1 10 '0 'ON DIN100% '0 m 0' '0'0IA'0 N0 m '0m mm mm mm

'.4NO U '0 .-4DI IAN .-4N In Dim ('4 N 4') (144') IA .-4

'0'0m'0'0mmmmmmm01 '0c') 0110 m '04') ON10 '0 N N .4 ID .-4 r-4 0.-s 4')

' C') '04') mm mm N

.40(11 InN '.4 0 In 0N '010.-jO'. .40' C In'00 4') mm N N CII N 4') (11

LA N '.0 In 0% In 4') 100.4104') In 0% N Dl N N 01 In'0'0mmmrlc.4c.4cI4Ne,4N

'ON .-4 m In LAO N 010 N'0IA 4') '0111 '01010 10 N'0

04') C') mN N N ('4 N N N

Nb C') '04') (NO In '.0.-401,-4mm 4') IA LA In 100

'0'0mmmc%INC'4c"4Nc,4N

("C It) I') 'ON N N It) 0110mm (14 '.4 N 4') N N 4') In 4')

'.4 It) 0% 100.-I ,-4 C') 'ON-IN 00 '-4 N (1410 It) -4 '0 N'0'0mmc.,NNNNN(.4NN IA .-4 N N N (1410-44') 0%'.O001000% 0 4')'0N e. 4') .-4 ('4 N ('4.-I N ('4

'0.4 N 0104') N '0,-4 01 mIA '.001.010 0'. 4 10 N 0

00% ('IL!) In LA (14 '0 NNLA'0NsO'0'.O(.4mm N N .—4 ,-4,-4.-4 ('4 .-4.-4.-4

'.4 N 4') '01(1 ION 010-4 N'.4-4-4

>4 DiNN01ooDIDioInCII

Z>4 DINN0101o.0'DIDioo('.4

>4 NN0I0101ODIOIn'.4

>4 NNNDlDIOOo1-I mmmmmmmmmm'0'0

>4 IDIAIA'0U'CNN'ONIn mmmmmme.)e.,mmmrn

>4 IA'0'04')'0IDNNIDLAIDN'0

'0'00N0%'0ONIAN>4 mmNNm'0IAIDIn'0U,Ine'

'0LAmOIn10,..4'0N>4 '.-IOONNNC')e.'CNm'0N

>4 NDl0%0%00010'.1.4 ('4NNNNC.sNmmNmm

o '0Oc'4DlIAmN'0mIAE 'OD%NNNNNDiDiO0% NNNNNNNNNNNm

o N0'0O'00,-40NO'0E IA%D'0IOInInNN100''0 NNNNNNNNNC'.INN

0 NInNe.'CIA,-4O(NIn0%E '0NIAC'CIn'0'0%OIOInNIt) N(NNC4N(NN(NC'4NN(.,I

0 ODIOO10NNONmZ'0 NNNNNNNNNNNNo ONNNO,-4'0IO'0(NeInE m'O'0N'0mNmmm10Nm NNNNNNNNNNNN0 '0NO'0'0OOInUINm,-4InmN4')N.-4N,.'0ION NNNC'INNNC'.JNNNN

0 '00%m'0'0In0%IDE O'0.-40%NONODI01N'0'.4 NNN.-4N('4.-4(N,-4,.4P4(N

o NNNO0N'00%'00N0 '.INN.-4N.-4.-4'.4,-4'.4pj(N

.4 N'0 IA ION DI 0.-4 N'-4-4—

-4ID0%-I

'OONCO(NN(00NC.)e')(N>4 0-405050404_IO04osososIn ••••••.••••• In(N 'O'OCOCOC.e')'O'0COCOCOCO (N

z 40NU)'0N05NON>4 OO0405040%O0%0%040%o ...••....•.. 0(N •0'OmCOCOm'OCOmC.)COm (N

_IIn40IflOSNOSWNifl_I>4 _I.-40505040400%040%OsO%In •.••.......• In,-I 'O'OC.)COCOCO'0CO(OC.)COCO s-I

x NNN_I.-I%0'0In'0'0(.4Di _IO0%0404oososososo •..••.....•. 0.-i '0'0CO(0(0CO'OCOCOCOCOCO s-I

COCO0CO'OU)'040s0C.)N(NDi

In COC')COC.mC. I')

(0004CO0O'OCO,-IOsIneOD's 40'0In4040NInIn'0'0'0'0 COCOC'rnCOmmmm4')mc') '0

04D'OCN,-405'O_IcN(N'0D's N'0CO'0CO'OInCO(0COCOCOC') COCOCOCOCOCOC.)CO(0C')mCO C.)

N05'0In_INNInON_I>4 InN _IN'0_IO.-4,-4(N(N (0(0(0(0(0(0(0(0(0(0mm (N

p _i040%O'0CO40CO_I,-4D's (NO00%OCN_I0%00'-4 (0C.Nrne'sr.,C.,mcN(NmC.) _I

o 04NInCONO0504NInO%E 0O050_I00%040s04 COfO(NNNCOCOC.)N(N(N 05

o InCONN0%0%CNCOL0NOE 0%05N0%O00500'O40 CN(N(N(N(N(NCOCOCNNNC.) 40

o 0%0In400%COCO'0'400 0E z:U) N(N(NN(.N(NC.)C4(N(N(N(0 U)

o (N'0O(N'0NNO,.-4InO 3E 04N0%040%050%0 Z'0 (NNCN(N(N(N(NNCN(N(NC.) '0

o U)40COcNLflOInmInmN 0Z NNNN0%04N0%04 EC.) (NCNN(N(NCN(N(N(N(N(N(.4 (0

o %0U).-ICO40N_IN'0In40'0 040%0NN4005NN00% ZN (N(NNCNNC'JNN(N(N(N(N (N

o 0%CO0%InCO400%CON'0 0E '0'O4040'004404040U)400% Z,.4 (N(N(N(NN(N(N(N(N(N(NCN _I

o O'0'0NO4cNUE (N(N'0'0_I0\U)U)40mCO0%o (N (N C1 (N (N (N (N (N (N (N N (N 0

_i (N (0'0 4040 N 040_I (N,-4 ,-4 _I

(04005s-I

x0%00000000,-4_I_I >451)(.3

(0.-4NS0'0'ON_I40_IN04 Z0500000000.-i_i_i >4••••...•.••• 0m'0'0'0'O'o'0'0'o'0'O'0 N

0500000000_i-i_i >•••••.••...• 414CO'O'O'0'0'O'O'0'O'0'O'0 .4

044D05N040Ci0%N,-i400%05040'O%0O0.4,-4_I D's0'-4

(NN04OO0%N_Im0b40 X404040NNN0s0%O040 >4COCO(')COC')C')COCO(0'0CO'0 414

05OIn05O0(NInm(N(N Z'040U)In404004050 >4

C'C'COCOC.CO(')O '0

CO(0(0'0In40NNN0% D's

(0C.)C.)(0f0(0COC')4')(0(0(0 C.)

InInInN,-IcNCOIn400'_I Z_I_I_I(NCOCO4040%0NNOS D's

(.4

0%'00%_IcN4000%N'0_I Z040%040,4_I'040In0N >N(.4(NC')C')COCOC.)(')(flr')(0 '-4

NCO400_ION0%40CN 0040%00,-4,-I'0In4040NN ZC'(C'COCO(0COC.(4COCOC' 0%

—IU)(NCO(N(NCNP)NU)OsN 0O0%,-40--4_I'040404040N E40

(NNN(N(N(NOO40'0NO% 0004_IOr-4--4'04040,04040 ECOC4(0COCOC.)COCOC.)C')CO In

NN'00%0040N(N0'0'0 000400'_I'-iC''0In404040 Z'0

'0In'040In0%'ON'0N 00'0%0%OOCNO'0U)In40 E(NNCN(NCO(0e')C')CO(4)C')(0 CO

'0In040004(N4040Cfl 00%InO0_ICO'0'0'0Lfl Z(.4(N(.4(NCOCOmCO4')(0(4)CO (N

CO'00InInLt440'040N,.-4 0O%(N0%040r4'0')NIn Z4(N(.4C.)C0(0(0CO(0 _I

In(NCOCO_I0'mU_I(N 0040'050In0In E(NCN_INCN(.4(NC')COm(0(0 0

.-4 NCO'0Lfl 40 N 0% 0 _I (N-4 -4 -4

'0400%-4

404000 '0(0 40 0-4 -4 N (N_I_I_I_I .4-4N

N (0040 '004005 N 04_i-4 -4 (N (N ,-I .-4 _i N_I ,-I _i (N

ON In s-I 400 (.40 0% (.4(N (N (N (N_I_I (., (N (N -4 _4 N

04N 400 '00.-I 40 s-I.-I (N CNN_i -4 N (N ,-I s-I _i (N

'0.-I 0440 N 11) N 41440 U) (N0'-i,-i.-iOOOOOO,-s,-i'0'0'0'O'0'0'0'0'O'0'0'004 Nb CON 0504.-sO NO040.-is-400.0%000_i0(O'0 '0'0 C') CO'0'0'0'0'0

'ON '040 _I N (041) NC.)'00%0_I00%04 050405-40C '0 '0(0(0(0(04') CO'O'0

N 40 (.4(040(004 (N (0040%OOsO% N 04_IQ%

CO(') (0(0(4) (0(0(0(0'0(0

04_I,-4 (N In'004 N 4140N 05 N N N _I0(0 (4) C.) CO ('5 (0 (0 (0 COC.)'0 '0

(040404040_ICON 4011) 'ONN N N N N N _I0(0(0(0(0(0(0(0(0 COC.)'0 '0

NON 40 'ON 4004 05 0so N 40 tOSOsO 40 N bOOC') (0 C') C') C') C') C') CO (0(0 0 '0

'ON C.)(N 0 0 '0 '0414(0SO N N 4040404040 N N 90%(0(44') (0(0(0 (0COCOCO'O(0

ON N N 4040 '0OOSON'0sO N 40 In In InInsO 40 N 050%

C.) C. C' (0(0(0(0(0(0(0(0(0

4144000 (N'0 s-I 40(0 N 00%404040404040404040400%4' (4) (4 C,) 4' (0 (0 (0 (0(0 (4

04 (N (040 '040,-I N(0 _I0'04041)0 '0In '0404040(0(4; (4; (0(0(0(0(0(0(0(0(0

(N CO_I (N 0440 (N 04')U)(0sO'04040(0 N '0 rn '041) 040404') (4) (0mm (4) (0(0(0(0(0(0

U),-4N 0 N_I 00% N (0CO U) In_ION 9 (N II) CO C.) s-I

(04'; C,; (0(0(0(0(0(0(0(0(0

(NC.) '041440 N 040,-i (N.-4 -4,-I

>4 ss-IC'4,-4(.4(Ne')(0'044)U)(N 'O'O'0'0'O'0'O'O'0'O'0.

0000_iOCNInCOIninIn>4 (N(NN(N(NNCN(NCOC')'0Ino(N '0'0'0'O'0'O'0'0'0'O

O(N'-i04')_ICNNNOON>4 (N(NN(.4CNNc'ie's(0'OInsDIns-I '0'0'O'0'O'0'O'0'0'0'0.

NO0C'404NNO40'0(0>4 _INN_ICN_I(N(N'O'0InNo,-4 '0'0'0'O'0'0'0'0'0'O'O'O

05(N'0'O'00Nm40'040N>4 0_I.-I_I,-s_I_INCO'0Ino,U) 'O'O'0'0'0'O'0'o'''

400N(N_IN1140'0eiin'0>4 O_i,-4_I.-40,-4c'sCOq,s.o'0

CN0'0-4_INm0'O04>4 000.-400.-4_I(')'0&flO(44 'O'O'O'O'0'0'O'0'0'OQ)

04NNN40IrICO'00N>4 0400000%0_ICOC')tnON CO'0'O'0'0CO'O'0'0'0'0l4)

(0'ObNCO0O40In'040Di 0_i090040.-4C.)C')'004s-I 'O'O'O'0'0CO'0'0'O'O'O.

o 40InO40'O04eim(0(Nc',z 0_I_I000'0%_I(0m'O0405 '0'O'0'0'04')40'0'O'0'O.0

o 40ONNCO(')N00%NN0s-IOO00500(N(Nm4° '0Cfl'q'0'O'0.o N'-I'O40ONs0'0'OsO(N40E 0-4O000040c'iNmNII) 'O'O'0'O'OCOC0'0'O'0'0'O

o NN_icW05CO(N4040E 04O000%0%0%_I(N(N40'0 (0'0'0'OCOC,)('4(0'O'O'0',

o (N'O0%IflN04CO,-4CONE 0%00%0%050s_i_I,-i40(4) (0'0CO(0mC.)COC.)'O'0'0'O

o 040'O'0(')'0'O-4N0%(NE 00%040%00OInN (0'OC')COC')COCOC44'0'0'O'O

o U)N(.44005_I,-IInc's(N'0N0%040%'0,-I (04C')C.)(0CO(0(0CO(0m'0

o _I_I0(N'0.-4_I0COI1)N_iS0%ONNN0%40(0o

s-I (NC.) '04040 N 0.0_I (N'-4_i_i

13)4004'-I

N4004-4

113

>4 InN%D%0NN40inN '0'0'0'0000ifl0000O.-Ii0000U)oiInin

>4o('4 0000000In0000>4 4004N0Qmoa,,-.I40in

'00 '000 U) Inin mO in '0

>4 N0%O%,-4NiO,-l,-4NNo.-4 0in000inifliflininin'0

>4 ON04O4N'0O'0m4.,in ininininin,n40u)u4in'0

>4

'0 Lflin'OtninInU)iOU)U)in'0

>4 ONOONm'0NiOU)in0%4') inininlninu,u)iOI.ninU)'0

'0.-I'0inin00%0,-4c'i>4 0,-iooc.im'0N40400N iflIninLflLfliO,nin,flin

0mr-U)InN.-I'0,-4,-im>4

— 'Oin'0ll)UininULfliflo N'0mN'0NNmminz0% OinU)'OinInLflinLflin

o '040tflNinN0%Lnm0%,-

40

o N0N400%,-440'0NNNE 0%NWN0LnN40moin '00'00'0LflLflinu)InLflo 0%N0'0LflQ%00,-IQ%9E'0

o N'0.4O0%N.-4NLfl'0E NN40S4040,-4'0mNa,4')

o .-1u)ma,inO40u)a,inNE %0iOiniO4040NOm,-iN "0'0'0'0'0'0LflU)infl'0o NO0m0OWO%QinNOE U)in'0%0inInInNN40N— 000000U)'0'O.Oo 000%'0.-4r40,,-400400E inN.-4'0NNONONiOiO0 00000000In'0mo

'0ifl 40 N 04O,-4N,-4 ,-4 -4

iD40Oi-4

.-I NO m ui c 4') N ('1 40>4 U)N40.-4.-INNIn 40inN '000 O'0in in in in in in

mNNONmNmP1In'0>4oN 00000Ininiflinininin

NmN'0inN40.-I.-4mO>4 in400%04NNm'040r-in.-I 00000ininininu)InU)

>4o,-I 00000InU)iflu4u)inU)

0.-I0O,-4O'0ONN0>4 iOONN'0N'0In40N

00%U)('4C.4N,-40U)'04')'0>4

'0

O400%40minNoNb>4

m

>4

N '0'0'0"0ininLfliflinifl,fl

>4

.-4 00000ininiflLflU)inifl

o m'0'0a,u)NinoLrl'0oa,E0% '0'0'Or'm'0Lnu)intninln

o m%D404')O'0NIniONONE 4040.-40%InN0%,-4N,oN40 000mm000inininino m'0N'0Lt)0NNmNZ 40io'-NmioOO'0Inill 000mm000inininino NN.-INNm,-u.)oE 4040,-4O.-m40N'0'0

o .-I04%DN04Lfl0%04m040xm

o .-40%00m400400in40'0

('9 000mm4',000000o 0N.-4'0'0In0%minr-mZ inmou)mNNa,N.-IN'0— '0llm4')'000oo NN,-4,--I.-I('40%0400NOZ0 '0mmmmmmm'0mmm

.-4 Nm '0in 40 N 0i0.-IN-4-4-4

NiD0%'-4

04 UImN0%ON40>4 '0in .N IA in in i in in In in in

0Inin'00inONN,lN>4 iO4004NN40'0'0iøO'0oN InInU)InU)InInInInIn40%0

X>4 404000%iOinin.Oin.-I ininiOInininininInIAiOiO

>4 biOa,a,a,F-inin40p.a,'0o.-4 tflInU)InU)iflu)U)IninIniO

NO04')iflplNONu)'0'0>4 U)inUi,>4

'0 tflinin40ininuin.ninin,c>4 LflN04OOLfl'0'04040inC')

Z '0%D.-4N'0Or-u44minQi>4

N U)U)U)40%DtflU)IflU)inin40

X -4'0N'040..4..-4040N>4

,-4 U)ifliflinhlhl)U)inifliflifliO

o m000inNiOm'0oN4',E m'0400%,-4'0'0'0N400% LfltflIniniOu)ininU)U)Lfl%D

o N'0N00%m,-0N40E .-4m'0o40'0'0'040bin40 40U)inU)U)U)in40

O ifl.-4040040mZU) IninInU)iOLflininu)inin,0

o 0.-4ON,-I00%m'0Na,X oNm40a%inmNm4040m'0 U)ininU)LflU)Li)LflLflininiO

o InN 4'0Nm40O%o,-4E 0%,4N40'0NNNin40mI')

O '0.-IX,-I4000%0040N'0Z 0.-IU)Nm,-4c'4N'0'0N('9 '0iflU)U)ininifiiflLJ)u)iniO

o '0O%O C') Na, 0U) 0 ONOZ NNO%C')U)N..40N4')0,.4.-4 O'0OU)Ininininu1inino '0NNN0NN0%mU)NZ iomLnoc'9No%in,..IC.4ma,o

Nm '041)40 N 0% 0.-IN-4 .-I -4

114

air->4 P1.-IInN 4040

0%Dini0m'040440inin>4 in'0'0N40Ome440iJia,oN iDiOiDi0404040i040404040

>4 '00'0m40in'0'040ai..-lin-4

>4 mC')'0miniOiniOmso'0o-4

X iniO0%O%'000Qr.4NDi

In

>4 NinN'0N.-40,-INNOa,'0 404040%D4ONNNNNI-r.

NNNN,440N-4O40in>4

C')

xDi

N 4040%O%0i0r-NNNNN

o o,-Imo40'0m.-.4o4fl'0>4 '0inmN4o4o40in'0a,— 4O%0%O%OiONNNNNNN

o I4000u1,-4m,-NN'0Z InU)mNu)4o40inN4no,.0% 40%040NNNNNo .-IinNON0400%'0,-IinNZ U)inN.-40N'0mU)inONiD %O%Di0,O4ONNNNN

o 0%N0040inLflNE '00.-I '-4004 C') NO00% NIn 4040'D%D4O40NNNNN

o ,-40% N .-4 N N N C') N in

'0

o 0N.-400%0%N0%in40m

C')

o 0NQ%'0in0r-moa,E .-I-400%0O0.-iQ0,a,N 40%D%0LI)40%0N('-N4ONN

o in0%'0.4N40ininma,a,E O'%0-40%NU)00%ON.-4 InifliniOU)4040NN40,QN

o N'040-4ma,'0'0oNooE0 U)40In411In%DN%D%DU)

.4 N '0 in 40 N 0% 0.-IN'-4-4-4

0%400'.-4

'44' 4'U)>4 ,-4 ('IC')0N '.O'.D N ID%0

SOCV)0>4 Ol0U)U)

5101,05 1010

x UIN4'NNN4'CV)0%'.0>4 UIN'.DUIUIU)el_CCV)No,-4 N'.DIOSNSNNNNID'0

mNa.c'4O,-IIQCV)4'mN.0>4

U) NNNNNNNNN'.D'.041)

>4 0.00IO4'CV)0%4' NNNNNNNS'.Q10UII.fl

10NNOCV)a.,-40.QN>4 01,-40%NN'.O4'N'.0'.DNCV) NN'.ONNSNNI0'.DUIUI

NN4'0N4NCV)N,-410>4 0.0NNU)N,-4.0UI('1'qC'l NNIDNNNNNID'.OU)U)

'.0a.UIN4'_C0%O4',o0.>4 a.IDin4'cwS'.0CV,-la..4 NI0'.DNNN'.D'.D'.D'.,DU)'4.

o 0%Or-44'_CUI100%0%Oa.0%a.a.IO4'CV)0N1flCV)00% N'.DIONNNID%0%D'.OUI'4.

o %00%N001IONCV)C'aLII,-4Z Oa.UICV)NN10'.DUINOa.'.0 0'0NN10IDIO10IOU)4'

o S00%N0'.O'.ONN'.0CVE OOUIN,4I0U)UI4'_C00%UI N10SNIO1010'.010UI4'

o U)NN0%%DN4'N0%NE OOUI,-I,-4UICV)4'CV')OOo%4' N%DNN10IO'.0'.0'.DUI'4'

o ,-4.-4CV)(V)NCV))ION0z O04'00Lf)4'CV)00%O0%CV) N1DNNIO'.0'.0IOL1IU')4'

o ('14'OlN4'OCV),-4.0f--4'E 0%0%C'1N4'UICVNN0%N NI0IO'.O'.D'.DID%OU1LCI4'4'

o N(V)N.0NN4".DUIUINE O'.D4'UIUI_C0%NUI4'NUI.4 N'.OID'.D'.D'.DLfl%DU)UI4'4'

o '.t)O4'a.'.Or4NNUIOE a._C'.DNN4'CV0%O04'0%o 'D%O'.D'.0%DUI4'IDUILCI4'CV)

.4 N ('14' UI ION 0.0.-C N-I_C .-l

S0>40('.4 UI%OIfl

411)4'>4 0,-IN 0U) •'.,-* '.0.011) '.0

Z 0.NN,-4CV_CCVNUI,-4NN>4 Oa.'.0_C4'0%el0'.o'.o.4 '.0U)UIID10%0'.O10UIUIUIUI

a..-Ca.0NNON,-44'>4 N4'O00NO%0%NN4'UI UIU)UI'.D'.,DIO%OLflUIU)u')U)

I04'CV)UICV)NCV)04'IOIOU)>4 U)0%N4'UIN4' IIIU)4'IflUIIO'.OU')UIUILnU)

0.4'O0UI4'4'0.N.0el>4 -CNUI'0N10N'0'.0,-4N0(I) 11)4'U)UI10%DU)UCUI41')UI

N.-4UINN4'O4".,0N>4 SC'4_C,-44'4'4'LIlUIOON 4'4'4'UIUIIO'.OUIlnt1)Lfl4'

4'4',-CCV)'.DUI%D.0SOInO'.>4 NSIO0%OOC'CN%DNN.-I CV')CV)4'0.0UIUI4'4'4'

o 0.0.4'0UIN,-4.4.00'._CZ -C%0NUINO,-4,-44".D,-C0'. mCV)4'4'UIIOUIUI4"4..4'

o U)_C0%('1.-I.0NelO%O')E N%D%DCV')Lfl4'0%4'UIO10

o UI100.4'4'01CV4'a.NUIZ NLn'.0el4'N.D'.o4'4'0.II) CV)CV)'4'4'UIUI4'4'4'4'CV)

o N00%CV)OUI('4N('10%NE NUIIO,-44'r-44'4'N4'm4' m'4'UIUI'4.4'4'4'cV)o NN4'4'N.-4OUINUIOE _C4'10OCV).-4C'I4".OCVICVNCV) 4'CV)CV)4'4'UIU')4'4'4'4'CVI

o UI4'4'NUIC4CVINC'40%WOE ,-4CV)UI0%CV),-.4CV)4'UINNUIN U)UI4'4'4"4.(Io _CUI'.O_COUI,-C'.DN%O'.DCV)E ,-4N4'NO%N4'4'0OC.Ir.4 4'CV)CV)C'I4'4'UI4'4'4'4'CV)

o 0%N'.D0%NNO%0%UIION4'E 0'..-44'4'0%UIN4'UIIJ)IDC.IO CV)C'I(')Ilr'14'4'4'4'(,')CV')CV)

,-C N CV) 4' UI .05 010,-IN-4-4—

W UIN CV) '.00% N 4' CV) 5.0('4.4.4.40-4 ('14' Cl) CV) ('4 CV)

'01010.0.0.010.0.0.010.0

O CV) 00'. .4 CV) ON NOSIDON%O0.0',-4000'._CUI UI '.0 UI UI UI U') 1010.0 UI'.Q

N 11)_C CV) 0 CV) 00 N,-C 10UI 4' 01.0 UI 000010U) U1 U') UI U) UI U) '.0 10 '.0 U) 4,0

'.0010 N .4 IDOlS ID '.010eV)("INN UI CV) '.010 0.0% 0UI U) UI UI UI UI UI U) UI UI UI ID

CV)NOC%I N LI) .0 UI 0 U) 0%II) .40.4' m '.0 10

4' 4' UI U) 4' UI UI It) UI UI U) UI

CV) 00110.05.0 CV) 5.0 0'..4 N 04' UI ,-4 CV) '.04' CV) UI

4'4' UI 4' 4' UI 4' Cl) UI UI U') UI

0% UI N CV) CV) ,- CV) 5 ('1.010 n0'.ON CV) 4' 05 N UI CV) CV) U)

('14' 4' 4' 4' UI 4' UI U) UI U') U)

N ID N .-4 0'. CV) N 4 '.0N 4' ,-4 ('4104' 0 CV) .-I ('44'CV) CV) '4.4' 4' 4.4' UI UI UI UI U)

N CV) .40% .-4 UI 4' N 0% ION'.05 CV) 00 U') ('4 .40.44'CV) CV) 4' 4' 4' 4' U') UI UI U)

('10.010 UI ,-44' If') N4' 4'U) ID .-4 0% CV)0.00%0'.OmCV) CV) C) CV) 4' 4'UI U')

4'NN ,-4 .4'.D_C4' N N ('.41') 4' 0.10 '-4 UI N 0% NCV) CV') CVI 1") 4' ('14' 4'4'4' U)

014' 4' N ,-4 ON ('4 II) C')N N 04' '.00% N 4' UI ID 0('1 CV) C') CV) C') C) 4' '4. '4. 4' U)

N U') N ON 10 U') 4' 0'. CV)0 N4' N UI '.0 UI ('14' UI ('-01

.4.0 NOC'4 0%_CS 4' 0% N N4' CV) CV) CV) CV) CV) 4' N 04' N

_CN CV) 4' It) ON 0.0.-IN.-4 ,-C,-4

U) N'.OUINCV4O.0'.,.-45 Q4CV)..40'.NON'.0 ID N S ID NNN

50100 N CV) N '04'S CV)UI U) '.010 CV) '-4 .4040'.00 .01010.0 NNIDN.ON

N .00 100_CS U') bID4' U) '.0 UI N 4' ON SN'.0 '.0 '.0 ID '.0 IONS 40 '.0 .0 '.0

CV) N 5,-C N 4' .4 UI UI U)CV)I0'.0IDl0N.0,,'.0 '.010

IO'.D N 5101010.0

4' 4' 0. ('44' N UI a. UI,-I,-4CV) .0.0.010 N,-4I00%510'.0 '.0 '.0 10 '.0 .0 NN'.D 10.0 10

4'UI N N 100% ,-4 UI_C CV) 4' NCV) '.05.0.0 0'. C') NOIS NU;10.010.0.0 N 51010.010

1000% U') _C'.D 0% 4' '.0 CV) UI 4'("4 ION10 0.-ISa. WSNID ID '.0 '.0 40 NN 10 ID '.0 10

100% '.0 N UI 0105 014' ('4O CV) 0.5010.04' UI 014' .4ID '.0 '.0 '.0 N N N ION N

N 0% N 0%0%NSN II) 04'0 N 0%NONN'.O N CV) ('I

I0.0O.4UI4'N 0I10O'.0'. .40% '.0-4 ID N N '.00 UI

10'.ON('-NNNCV') .-4 N CV) 011) .-40N CV) N CV')01.-I 10,-I N UI N 10100.0UIIO'.O'.ONN N N NN CV) 0% UI N Na. U) CV)

010 UI 0%D CV) '.04' 4' .0Ln'.o'.o%oNNa)NNNN

NO ('101,-IN '.0 UI_CNNNO'. UI CV) 010 CV) ION 4' '.0 UI

UIUI%O'.0NNN5NNN CVI CV)N '.0100.04' '.0 CV) UIION C') ,-4 UI ('150 CV)4' 4'U) U) .0 '.0 '.0 N SNN N

4' 0% .4 CV) 11)014' CV) UI 010%4' UI00 UI 4' N '.0.4.44' NUIUI'.O '.0.05 NNN N

('40% N 1(10010.4014' 141 CV)('44' N 0.0 ('4014' UI NUI U) UI UI ID N N N IOSN

('4 CV) 4' LII ID N 010,-IN.4-4-I

InN

>40Cl

>4UI-I

0-4

>4

U-C

x>4

4'

>4

CV)

x>4

('4

>4

'-C

a'.

'.0

UI

4'

CV)

C")

-I

0

NN0%.4

0N

0'-I

UI

4'

CV)

N

-4

0%

'.0

UI

4'

CV)

N

.4

0

4')N0%

0N0'..4

-IC—a'.-I

115

0C.'.

In

0'-I

In

N

'-I

0

iO

in

0E

N

1

0

00 in 4 '0 ..I N 0 N 'C'C C' C' 0 N N C' a

N N N N N N N N S

sO N 00'00 _I m N 0% 0.-4 N in C' Os 0 N 0 N NSN

so — m in Os m 0' in 50 N050 in 0% 0% 0 m N 'CSON N N N N N N N N

N in 50 _I l' 0 in N 0_IN in 00% .-4 5' C' in_INNNNNNso N N m N inN 'C _I 0% ('4 0' NNNNNNNsO0 ('4 ON in iD _I 0 (N S_I 0 ('4 in 0NNNN

N .-'I 00 NO in N NN N f' ON 0% N N.-I'C 'C N N N S N

m 00 0 m N 'C N m00 C' C' N 'C 'C N N in_IS N 0% N N N

'00% sO 'C'0 ,-I 0_I N'00% 'CN _I '0 ON SD 'C 0' NO.' 'C _I

V

N 'C 0%_I 0 iD ('4 ('4 (''0'0 Ui 0' in C'4 N S sO 0S N C.' N N N

'0 N V'i NC' 00%_I sOin Ui sf10.' 000 Ui m 0N N 0.' N N N

C" 'C _I _I in N sO m 0' 5 m sosO in O 0 '0

NN0%bN0.'sCNr-.'C C' '0_IC' SO '00% ('4 C')'C its 'C 0CM 5 N N C') 0 sO _IN N C.' N N C' sO N N

N U) its 0 in 0% '05') NON 0SO 'C sO 0% 0% 'C S ('405 its 0

N N SN N 0% sO N S N

N _I (' SO U) Its 0 'C N its NU) Its in N 'C sO in (MO_I '05

NNN5N0.'sDr-NsO'C _I _I in in '05 sO S '0_IC') _I 0 Its in 'CO_I N (' '0iflNNNNN0.'sOsCV'-s_I CM C') '0 in sO S 0% 0_IcM

_I ,-I _I

N 0 C'C' in> 5 0 _IN'0 Nin •N SC'sDmsOin50_I'05.O

>4 inin_I_IOC'_IN'0ON0%0N NN55OC'UI_INONSONmN

>4 in'0O_IOOsO_I'0C'NIn.-4

NC'NNOSD'0'0m_I>4o.-4 NNNNNØNN

OOIflNON_IOinOSS>4 NOmC'insOO%OC'4itsNmIts NNNNNNNN55

o_IInsoOr-mNmm>4 _I_ImsO0%c'4msoe4'0 N5DNSSNNNNNN

LflCCNSDSDLA.-455_I50_I>4 C'C'S_I'0_IN'0OC') 5D%O%OSNNNNNNN

C'if5inNO'0C'_I'0_I'C>4 %DN5OinC.45OsDOC'N50('4 %D%Ds0NsDNNNsCN50

inNsDLnsD'0irsln_IC'50C'>4 _IC'_I0N'0sONits0_I D%ONNNsOSOsD

o Ifl'00N'0N'C0%O0in_Im'o0mas

o NOONC.'e'4'Cu)mE ONNsOmON,-o50 IhSCIflSOSONNU)5OIj)

o OmN_Iou)'0mo.'ONC.,E ONOinNSC'OOinin 0s0Sins0u)o '0b'CC'OD'0_ImsD_Iz 0%'CNN'0_IiflNC.'s

o '0iflsO0'05')'00C%c')C'EC') Lt0LIIUo NC'S_I'0e')m'0N0.'sDso_Ir4'0'0'0_ICM in'Cjf1U)U)o NU1'0NO_I'CNSOE U)C.'cM_IO."C0C.'e')m_Io_I

o NsDSsDO.'Smininso.'ONsOitINNCMsOQ.'O itsininin'Cin.

_I N 5')'0 in 'C N 0' 0 _IC'.l— _I .-4

_I '00% 4'S CM 00500% In '0C' Os S 0050% N sO SD C'S _I

SNNNNNNNNNNC' N '0 sO in '0 sOON SD Il) 0

N 50% N SD in NON S S N NNNS N N N N

0 ('1 ('4.-I '04') in SO SD '0111 NS sQ 'C N Nit) 0 N

0% 0' '04'S 50500% sONC') C'S N N sO '04') DOs sO 0% 0

SO'0 SN 'C N SN sO N,-4 N_I 0 in C'S _I N '05NSN NNN N 'C sO sO U) in

5050 in In 1') '0 '0 in N In C's'000% '0 s-I 0% 'C U) N SO SD

S '05'S 505'S ON_I 1(1_IC') S 'C Ill NC' SO '0NO C') 45)

50 sC sO 'C U) IS)

11)05'S ,-4 U) III SN C') In '0in_I00 'CC') 0 N '0

C'S 0% 0 C'S 0% '0 5(5 U) 0% ON N(5) N 45) _I X 'C U) C'S 55

Ill in sO sO in in in U)'0'0

0_ION_I C') 11)0 sO 0 C')_I 50 II) '0$ 0% sO U) '0CM 5050

U) U) U) U 50 its U) IS) in its 0'0

CM SD SO 'CO 11)0 in 45)0%0 5') 4') 0% in '05') ,-4 50 IS)

it its U) 11 in in in ui U) II)'0'0

NO.' ('4 SN (5) .-4 'C C')0% N N_IN 50 0CM ('40 U) inOinLIIIinLrinit'0% 0% U) 0%_INNS 0 U)N000% 'C 0 ('4 s-I 4 0% in C

U) its '0 U) IS U) Li in '0'0'0

I') 0% 0% so 'CO.' 0% in 50'0SO N '05') s-I 00 5')

'00 '0 in U in itS in '0 '0'0

('4 S 0 '000 Sin_I 'Cin 5055 ('44') _I 0% OS '0CM

in'0in '0 '0 '0

CM SO N so 'C 'CO.' 05') 0% NC'.U)'0 N NO 5Dm N'0'0 '0 '011) Lii '0in '0'0'0

_I CM 5')'0 in SON 0.' 0_IN5-4 _I _I

('4504') 0 in S N Ill N>4 N 'C sO SO In Ifl N SInN NNNN S NNSr-X>4 In%DSDSOI11'0In'0InN(..0C'. NNNSSNNNNNSN

>4 '0Ininin'0p)'0m'0SDSDNIAs-C SNNNr-NNSNNSN

N'0_ImNOSN.-4NN>4 NmmmNsm_Ic'.itSOS-i NSNSNNNSSNNN

>4 C'N'COs0'0N0In 'CSDOSO%O50S0%DNNNN

>4 'CN%D'C5DISSN0%C.SC.'Im'0 SO%DSD%DSDSOSQ'CSDNNN

mNinNsoN,.4m'0>4

C'S 'C%DsO'C50'CsQsOsOSNN

USO5int155')'0OsQ'0OsO>4 _IOOSO_IO'0'0Nr.IOOC". S0'0in5O%OSOSOSO%DNNN

'0_IOO0m'Csoo'C_I>4 Inin'0inN50OC'4It5OC'_I IISUuIflit)IflLfl5Q'CSDSSØ'C

0 '00NOInNNNNIflOz mm_Im1nLnC'_I'0NOs U)inU)ininin1n'C'C'CsO

0 '0N_ISISN0%_IN(5)_INmZ _I0C%0mm0m-'C inin'0ininitsU)'C'CsO'C'C

0 '0C'minC'Nc'i'0ing.'E O0%O%NmNosN5o'0I4sin U)'0'0'0U)iniflin'C'C50sp

0 inoin'0_ImONW'0sOUiE C'C'NNC'4s0_Iinm'0'0 '0'0'0'0inifliniflSD'C'C50

0 _IC'inCM_ICMu)m5osNE SSON,-I_I'0s00m_ImC'S

0 'CUiin0_I'C0.'sfs'CNmE 'C'Cinin0Om_Io.'_I('1 '0'0'0'0inininil)u)5DIfl.0

0 NCM_I'0O'Cifl'0_INCME U)inin'0C'0ScMmN.-4sQ— '0'0'0'0'0I.flIf1U)'CIf1IS)

o _I'0mooininoL0 '0'0'0'0'0'0tnu)ig)sCItsLfl

_I cM '011)505 0% 0_ICMs-Is-i_I

NNOs

>'4In('4

0N

>4IS,_I

>40_I

>4

US

>4

'0

i's

>4

N

>4

_I

0'

'C

U)

'0

m

('4

-4

0

S SCl 05

S05

116

Ui O N 0% %0 .40 %0 %0C4 0 U0% '0 N 0.0OO%O-4

0.0% 0 Ui 0 Ui 0% N N Ui00. 0 N 0% 0-40' 0'0 C 0' 00' 0

-4.-4 Ui 000.0N S NONN 0 0% .40% D 0% 4 e 00

0%000—-4—

00 C% 0 N .-4 N 0 .-I 0N 00% .4 0 N Ui .-I 0

0% 0' 0.0000-4 -4 -4

00 N 0 N N Ui 0.0.N rN 00% .-4 Ui N .-l N N -4

0' 0% 0.0% 000-4 - -4

Ui 0CM C' N .-5 N 'C .-4 .-4N 00%.-4 Lfl N N 50 5 N0. 0% 0' 0. .100_l $ lUi0 0% '00% Ui 4 40 .-4'0a. CM .-l NO' Ui 5') Ui 00

,-4 .-4 _I

'00000% e 00% N o o 0'0Ui.0 N NO Ui .0 0%

0% 0% 0% 05-4 0 0-4.4 _4

ON .4 N - 0% 4 Ui Ui 00'NO' '0-4 'ON 00 m o

_I 4-4-4.4Ui0.00% CM i 0' N ('0

0 0% N CM '0 N Ui .0 N0.00% 0% 0.0% 0' 00 N ,-4 .-4

-4 -4-4-4-4m 05'i Ui N 0% N N CM 0%N 0% 0% N N .0-4 Ui NON

0% 0% 0% 0 0% 00CM-4.4II -4N 000% 00.0 N 00

0% N CM Ui .-l 'ON 0% 40% 0% 0% 0% 0% 0% 0 0 CM .-4 N

_4 — -4 .-4 -4

% m.c.o0 '% ('i C') C') '0 C') 0'CS N N Ui 00.0 C')

0%0%0%OON.4r1,-4 .4 .-l .-I .-4

'CS '-4 0.0.05') O N NUi '0 N N N .4 00 ON N0% 0% 0% 0% 0% 0% 0% 0 0CM r-ICM

-4.4-4 .-4 -40% U) Ui N C') (MO 0 Ui 0% NO .0.0.0 N -I CM .-4 C') .-4'0

0.0% 00% 0.0% 0.00CM .-4 .4.-4 ,-4 ,-4 .-. -4

5') 0% 0 .-4 CM C') C') NUI '0Ui LI) N 0% .4.4CM N CM CM

0%0'0%0%0%0%OO.-4,-4O.-4 -4 -4 .-4-4

O 0005') CM 0% ON N'CLI) LI) Ui SrI 0% 0% CM CM .4

0%0%0%0%0'0%,-4ON-4 -4

.4 N C') 0 U).ON 0.0,-I CM—4-4.4

Lfl .-4UIO '0 0. 0% LI)CM O.-4,-IOO0'O..-4,-4...4,..4_s

_4.-4-4 4.-4_4.-4I>4 ,-4'-4NN0%NN00.4oCM 0NNOO0%04CMC'-4

.4.-4,.l.-4.-4 l..-4.-4,.4I>4 tNONCM%O.-45')OIll'.4 0NN000O.-I,-4(4NCM

-4.4 4 .-4.-l .4 .-4 -4.4.-I .-4 .-4I ON'CIi'IInUINUiNNOC')>4O .'-I ONN000%0.-4CMC'IN-4.4-4,-s_s -4.4.4.4-4-4I 0..-ILflUiN'0UiO.-4N,-4>4 C'-0.O00-4UI 0c')NO0.G0-4,-4c%aCMCM

,-4_4_s_s .4.4-4-4,-s_sI>4 0%OONO%0.DIflON0 OC')e'IO0%0%0.C4C')c..4-4.4-4 -4-4-4-4-4I>4 0%S00,-4ONNN'0CMC')

-4-4.4-4 -4-4-4-4-4I UiOUiN'-IC')C')LnOrION>4 I')UiNUiO0.'0C-INNUICM .-I000C4C'ICM-4-4.4-4 -4-4-4-4-4I Ui0-4.0e.sc')c.i0.-4NUi>4

-4 .-4000.-4.4rn'0e').4-4-4-4 -4-4-4-4-4

o '0.-ILflN,-4C'I,-40%IflNN.DE 0O.0LI1C'O0.N0%0% N0Ui0'Q.-4C')0C').4-4-4-4 -4-4-4-4-4oE 0N0'OC'),-40%N0%N'O'0 O')O0.4-4-4.4 -4.4-4-4.4o Ui0C'%0%OUi

0UIUi0%N.-40%%DCMNLE)U) N00.4(')'O'O

_4_4-4_4 -4-4-4-4-4o N0'NO.-4N0.IC')N0Z .-I,-4'OS-4NN0

_s_s_s.s -4-4-4-4-440CM 0 N .-4 LI) 0 Ui Na. Ui 0C') C') N '00%,-4 CM '00' SNO U) ON 0.4NO 0-4-4-4.4 -4-4-4-4.4

U) 0 CM %0'O N Ui CM N00%CMNOU) 50% Ui 0 UiCM 0 U) ONS 0%_s CM 'p.4444SCM N0% 51) CC') 00 LI) 0.0W 00 C') tO CM CM0 C') 0%

.4 C') LI) ON 050%_s_sON-4-4-4-4 -4-4-4-4C') 0.0% CM OhS C') C')O CM CM CM 00CM 0. C4'0

'-45') C') 0 0 tD 00 C') 0-4.4-4-4 .4-4-4-4.4 N C") 0 U') 'ON 0.0.4 N

-4 -4 -4

I 'C0'O0UiCM'00'ONa.>4II)CM .4NC'4C")NNC')00C')Ncfl_4-4-4-4-4-4 .4-4 _4 -4 -4_I OIflOUiC')UIN00bN.0>4 .4SUim.-4S%0,.4ONUiON CMNNC')NC')C')'0IflONC.5

.4-4-4-4-4-4-4-4-4-4-4-4I '0SO0.NO'O0%0'C')0%>4 N0'NUiOC')ONC'4N5'0In.4 NNCMC')mC')ooU)'ON

-4.4-4-4-4.4.4-4.4.4.4-4I 05.ONNO'%O0%UiO0NN>4 NON%D0O.-40%C')CM'C0.4 CMC')N4C')C')O'0UiOr1

-4-4-4-4-4-4.4-4-4-4-4-4I>4

Ii') CMUILICOCMS")-4-4-4-4-4-4-4-4-4-4-4-4I N.DUic-4S.0,-4enr-D.s O00%O.O0.0'.0q')Ui'0O NOC')C')'0t1)%fl'OC'C')-4-4-4-4-4-4-4-4-4.4-4.4I NO,-4N0'N0Uir'INsn>4 IflUI0%.-40%,-4%0N'qUiC')

-4-4-4-4-4.4.4.4.4-4-4-4I N%0,-4C')O.4t0N0'O0.>4

N CMC')NOOU').C.,O0(MC").-4-4_4 -4-4-4-4-4-4.4-4.4

I 0.CM.O0'0C')UiUiOUi0'>4 40ODNOSC'0.0IflN.4 C')ON000II).CLE')O,.4C')-4-4-4-4-4-4-4-4-4-44-4o NCMOSNNSNO%OONz .4C'INU).O0',4C')00% ON0'0Lfl'0UiO.-4C.,-4-4-4-4-4-4-4-4-4-4-4-4o II)tOU).0S.OQ%OC')U)c.,E UiUI.O00'.ONLI).O.0OUitO OON000Ui.0UiC'),-4CM

-4-4-4-4-4-4-4-4-4-4-4-4o 0%NNOLI).DNC'),-40

%0.D.000.0.OC')mO0.CMII) NLflUiOUi'CLflC")OCM-4-4-4-4-4-4-4-4.4-4-4-4o '0N0N0Nuie.').-4t-.00 OONUiUiOUi%OLAC')O

-4-4-4-4-4-4-4-4-4-4.4-4o N0%N0Ui00LI)CMN0E 0t0CMtO%DO0N.-4U)U)C') IICM0U').0'OmO,4-4-4-4-4-4-4-4.4.4-4-4.4o tfl%OO.-4,-4N'O50.NO%E 0It)0%0%0t0N5')0C')NCM UiONO.DOt$')U')'0C")OO-4-4-4-4-4-4-4-4-4.4-4.4o N0.-4050,-%z NN0.-sCM0%.O.O0%NN.4 O'OC')O%000tI)C')CMO0%-4-4-4-4-4-4-4-4-4-4-4o NOONNmOO%C')UiS.pz 0%.O,-4UiNOUim.DN.4CM0 C')C')C')N%D0OUiCMNQ

.-4 CM C') 0 U) tON 0% 0.4CM-4 -4 -4

-s0%-4

I ,-SONCMCIOO%.OCIOUi'O>4 0NC')0C')N0NUi5It%t_4

I>4 0%0.4CMC')'OOCMONUioCl

>4 0'0—4C'%OOCMON.0U)-4

I>40.4 NNNI>4 UiN0%.4C')C4CM4')%D0IA SNC'-N0%I>4 UitON0C')C'4CMr40%N.-4'0 NNtN0%I>4

m

I>4 N0UiN0%CMNCMOCMCMCl CNNC"N0%O.a.I ..4UiCM,-sNO.0%'CUiUi,-4>4 .)UiN,-4,-4r40ONC').4 NNNNN0%0%0

-4o c's.om-4o%c')-4E 0o-,.o00%-4.oNNN0% F-SNNNNO%O%0

-4

o LtI.-4t0U)0.-4LflUi0%Ui£'C '.O%CONNNbN0%0%0%

o N.-4SC')0C')C')OO%OUiNx N0%N'COUiC')ONU) .O%0.0..DNNC'-N0%Q%0%

o NUiONNC')C')N0%OOz %0.OSNOONSOOC')N'0 %0t0'0%0NNSN0%0%0%

o mInNO.-4000CM'ONUiUikfl0UiCM0NC')0%.-4U)

C') .SO00o Ui.-I.4UI0'0.D.CLfl0tO0(')0UiCM.00'NN.-'I0CMCM %O%0%0'D%0N0%0%

o SC'I.OmOIUiSOUiO%C')Z 0NUI00t0U)O.O0%.O'4

o O%'O%0C'IO000%OUim

0 UiU)%0UI'0t0%Oa)SN

.-4CM C') 0 Ui.ON 0% 0.-SCM-4 -4 -4

N0'-4

0CM

0-4

U)

0

C')

CM

'-4

0%

tO

LI)

0

C')

CM

-4

0

0N0%-4

C')

N

-4

0

00%-4

117

mtNDC'S mNCl,.400O.4— - ,-4 .-.S .-l — .-4 .-4 .-4 ,4

>4 NC.4IAa'0%DoCl

.-4 .4 ,-4 .-4 .-l ,-S .4 l -4———

UC'bNON.-4.-4>4In,-I C'IN,-l0OO

,4 ,4 - .4 1 -I .4.4.4.4 .4 .-4ClOa'OIflC4

>4 NC')mOb0%Iflo.4.4 ,-4 - .4 .-4 ,-4 .4.4.4.4 ,-4 .4om'0oo%oo'In

>4In COoc.4-4 ,-4 ,-4 .-4 .4 .-4 .-4 ,-4 .4.4 -

>4 NNLflIfl-4m-ooc,4 .-4 .4.4-4.4 ..4 ,-4 .-4 4.4 ,-40N%DO'OUIONb

>4 OIim mc,-iooo.4IS .-4 .4.4 ..-4 - .-4 ,-4 .4

4'INON0000%C'SN>4 N4')0%,-l0%UN .-40%O%a'

.-4 .-4 .-4 ,-4 - .-4 .-4 .4

z>4 bN4''IN-4 O0%a'O

.4 .-4 -4.4 .-4 .4.4 .-4 —

o m4'O%04IlLflE '0 ON 0%0%a' 4'cooa-4.4-4 ,-4 _4_ -4.6

o 0ON.4'N4'E'0 0%

_.4 -, _.4 .-4 .-l —

o mNN,-4,-4zin

-4.4SI .4 .-4 .4o ONN'0UIN.-4'i'tnE -4'-4N0%OODN

-4 -4 ,-4 ,-4 .4

o 0%('lN.I0NEI'l

-4 - .4 ,-.4 -4.4 -o .-4 N N '0C'4 '0 N mE UNNN-4,-4N NCN0%NNN

,-. ,.4 ,-4 ,-I .6 .-4

o C'i'Lfl14'C'4NNZ 0Nr-4UI0%NO.4 N-4O%NDNN.4-4-4-4.4

04' U) N .-4 '0UI 0% '0 ON U)N U) m I1 O '0 UI O O•.4.4 '0 ('4 .-40 N O N N.4-4-4-4-4-4.4 Cl ('1 '0 UI'O N 0% 0.4 r'S

.4-4—

OIn00%O.4b,-40,-grn'0>4 '0C404ObNU)mflU,C'S .4000004.-4,-4.4-4 .-4 .4-4-4-4-4-4-4-4-4

>4 NNN'0OQN(')W'0WoC'S .4000-6,-S,-4,-4.4-4-4-4-4-4-4-4-4-4-4')U5WNO%C'404>4 OWrn0,ONNC')U)'0UI

U).4 .-40000.4,4,-4.4,-4_S

>4 a'NInNNO%tDbNUIC')U)o.4 000000.-4,-4.-4,-4,-4.-4

>4

U) 0G0CO04.4.4.4,-4,-4.4 .4 .4-4-4-4-4-4.4.4

>4

'0 00400O0.-4,-4O,40.-4-4 .4 .4.4-4.4-4-4-4-44a'OW0r-

>4 0a'UI,-4m.-4C'a'0bsrn oco,-ccco.4 .4.4.4-4-4.4.4.4b_S.-4b.-4LflNCl'0O

>4

N 0%O'a'0%a'0000000.4.4-4-4-4-4.4Ii) 0 '0 .-4 ,-4 N C') U) 0t .00% 40 Nb 4') 44)1') 4') tO tO N

0. a' 0% 00 ot a' a' a'.4-4N tOot '0 4') ('S N N N 040 NO'0.4 1') 0 '-4'0 ('I UI N

0% 0% 0' 000' 0% 0% 0%.4-4'0C'4 .4 UI 1') 0.-S '0 Cl NO

U) ("SO '0ON 0% N .4 ') UI

tDrnOO%ONClOUIN 00NOrnO.4b.-4ON'0 Eoa'a'a'0a'a'0a'N UI U) 4') a' 0 UI '0 Na' N.40,4') a' 0 '0 tOO .4 C')

40 C') 4') '0 Cl N Na'('SO C'S 0% '0'0 0% NO .-4ors;oa'

0 Na' NUImNN.40.5,-S N C'S C'S UI 0%

N

.-4 N C') '0%fl tO N 0% 0 '4N-4.4-4

>4 .4m(',mf',Ii)C'S ClNmClC'4.-4,-4,4,-4.4-4.4-4.4-4-4-4.4-4-4-4

O'0040%a'id'lN.-4C'I4'l..4>4 In MN O'O'0'0oN NNNC%SN.-4,-4,.4

x>4 '0NmO'0mrnU)'-4 ,-4.-4NNC'5m4ClC'i.-4,-4..4.-4.4.4-4.4.4,-4.4.4,-4.4.-4

>4 MN.-4'0UI1'04'5s',o.4

>4 O'0O%C%IMN'0'0.-SM,-SO%LI) _S.-4.-4C'44')4')C'44'4C44,'-4.-4O,.4 .4.1.4.4.4-4.4-4-4-4-4

'0C'SO%DC'i0O4.0b,-15115>4

0 0,-4.-4t'Sme'5C4C'4cl,-lQO.4-4-4-4.4.4 .-4.-4.-I,-4 .4.1

NC'lW0O-4C%i4O>4

C') O04,-SS')MNN..4,.400

>4 MbC'S6n%0NCl.-6bc'sN 00.-4,.-4NNC'4N,.-400Q%.4-4-4.4.4.4 .4 .-4.-4,-I,-4z O0%a'UI.SiflU)m,-4..s>4

'-I 0la'O0.-4.-4,-4,-4.-400.a'————.4-4-4-4

o '0C'4.4N.40.'0MOZa' OO,4.4,4,-40a'O%-4.4-4.4-4-4-4o N%OM0%%0ClN40'0'0x M4DNMa'a'O.-4b400%4')tO a'a'OOOO.4,-4O0%

UI 0.4 '0 N.4 Na' N UIN '0 N Nb N 040 tO UI N040.0 O,-COtQ.4-4.4.4.4.4

o ONht4UItOtCNO'0tOa'E '0Na'tO'0N.-4'0 a'0%OOOOO0Qa'

———.4———o Ob'0'0'0ONO'0NClZ a'a'O,-4U1U)NtO,-4e' a'Cy00OcOOOOt-4-4-4.4.4O Nba'MCStO'0E O.-4NNtOMf-a'mC'S a'0%0%0%a'Q%OO0N

'-4.4-4

'0Ot-4

-4 N C') '044) 406'- 0' 0.4 N—.4—

OO0,NIfl%00NNtoIn>4II)C'S .-4'-4,-4,-4000000tØsa'.4.4.4-4.4-4-4-4.4

N .40 ("4 Na' -4 ('44', NW0 UI '05 U) tON 4')OO0O0.4.4.4.4-4.4.4.4.4.4ON '0 US a' '004040' N II) N.4 II) C') '04') tOO U) Nb N

'00 C') ON .4 tO LI) Ma' U) N0. N '0(4 -4.-S'0.4('4 '0,- .4.400000040%

IntO '040 II) .4 '0 N.4CMtOM .-4 Nb NOW N 0.4'0

040.040040.4-4.4.4 .4C', C') Na' 0 U) 00% U) '0044')'0.4 a' '061) '0 '0'0.4 4')

040% 0 0% 0. 0.4.4_S 4U) 40,-SO Ii) 04 '061)0% 4') ON.4 tO ('1 ('40 '0 N ,-4 0. '0.-SOO0,a'o,o4-4-4-4.4N .-4 U) '0004') tO0'0 04WN '0C'Sb %0a' N N LI) ('4

.4-4NM o me '0 N 0.-S '00%a' 40 '0040 NO 0800'NbNNNN UI M,-SN '04') 41) a' b'0.4N C') .4W 40 '0 5 tO '0 4000'a'NNNNNNNN4') rn UI t0'0 '0 .-4400%'0.-S 4') '0 NW UI '0 4) '0(40%NNNNNNNN'00NO.6') 00 tON 0.-S U)C') ON ('4 '0 C'StO'0 C'S'0 '0 C'S

0%NN(-NNNNNb.4 U) tO C') '-4.0.-S O 06')("Ia' W.4M.-S UI MC') C')

N 555 N bbSa' 0.-S N MOb LI) UI %O.4 N U) a' 1 0 '0 ('S NM 1') .-l

N N N N N Sb N N

4') NNW 00 UI ('4080W 0'06')C') N ('40' ('4 ('4.44') ,—4

N NW N N N N N tO

o Nb'0NONW'0,.4tØa"0NWOUIONON4,-linCl.4 NNNNtONNNNW0o a'Nb'0O%O('SUINUItOØ8Z NNtONtDUIMOa'tO,.40 NWNNtD4O%ONN%OUIU)

U)

0.'-4

('SM '0 U) 'ON a'0.4N.4_I 4

0('4

In-I

0.6

U)

'0

CS

.1

a'

40

U)

'0

C')

N

'-4

a'

t0

U)

'0

rn

o N-4WOWNUI.40,m QO0%UI.4UIUIa'ob'0U)W a'tn'0UImON'0InOCl.-4 '-4 a'0O'a'0OOWN

.4.4-4o .4OtO.'SLflM'0b40 N 0 NU)NU)UIMr-Nr-boZ oa'moMCl'ONN'0.4.-S Z '0NNtOtONU)O'04')o 0 0,0%00400bW-4-I

N0.'-4

118

LA Q .-4 0 C'l N 4 .-4 0 N O>4 4N r-,-loLom >In . LA(4 GNNNbNN C4 SNLA ONNO'.-s>4 m NDO OO%O, >40 .o'I C NNNNsr- ci SN

0LAqe,IN 0.->4 >4 'oinSn . U.,0% N N N SN S N N .-S N NNS 0>4 oem Ln..4r-..-4N >4o . 0,-$ 0NNNNNNSNN .-4 SNSNN>4 ,-4u1,-4,O-. >4 SNIn NNNNNNr-.. sn

OS N — Os C LA LA D m 0 LA 0 .-4>4 m N0055fls >4 %%0NNNNNNsoNssDso

LAspbN%D>4 ,-SNOON c'NCmsD >4m SN S SN D sO sO sO sO 50 5050

OsNU0NCNIfl>4 >4 CThC4 NNsOsON lAsOsOsOsO 5.050

mNassOoN0%,.I>4 >4

o oOSmmOsor- 0z NC40m%DLnS0 £Os 505.05050 sO 5.1 LA LA IA LA 05 LA LA

o so N so m 5050 0E mnnr-50 NN50 t%IflULfl 50 lAin

o .i LA N I' LA 50 0 .-4E Cmusr-LA N LAUILALflLALJ IA II;In

o 05 os me c.s m c .-4 0 0E .-4 i c'a LA .-i as es mm LA LA

LALAULLAUI LAIn

o soosOme4oIno 0E C4 r4 ('4 mm E N IASN sO 5050 sO IA IA IA LA LA LA LA LA

o IDS-SOSO 05OLnS0.NN 0e .-i .-m ON .-4 ('4 m E 50 LA('4 NN5D505050LflUUIflifLA ('4 LALA

o 005m,-sosOom.oo 0LAOLf%.,OO Es-I sO IACItLAlflLA s-S LAIn

o r4 ('4.-4 50 5 LA N 0 mE LAN s-s as em - 000 E eso 5050 sO LA sO LA LAO IA In 0 IA IA

m5nsOr-oso..4c'4 s-I('4s-S s-S ..-I

5.0 N0% 0%

s-I

119