EFM Assignment 16-17 by A Bochkov

1.
(a) Default risk free government bond means that there is no risk of default on all the payments and
redemption value meaning investor requires no risk premium for this bond. Therefore, the YTM is the
only constant nominal discount rate which equates the PV of the bond’s payments to it’s nominal
value.

(b)

(c) Consul is a bond with no redemption date, so Present Value of it can be calculated via a perpetuity
formula: P=C/r, 98=1/r, r=1/98 r=0.0102
OR
by plugging in C=1 into above formula, which yields following results:

2.
(a) Shiller’s idea was that stock price movements excess the underlying fundamental of that stock or
new information becoming available. Hence, questioning efficient market hypothesis. His idea bases
on the fact that real prices should reflect fundamental price based on dividends and a small forecast
error. If this holds then we can write p*=u-p and hence var(p*)>var(u)+var(p) as cov(u,p) must be 0
due to optimal forecast of p*. In reality, he found that this variance inequality is violated.

In our test for excess volatility for the period 1981-2010 we need to assume a constant return rate for
the stock as it’s one of the assumption for the model. Shiller in his paper used expected one period
holding return for each t, Et(H)=r, which with this data I found to be 3.74% using Ht=((Pt+1-Pt)+Dt)/Pt
and then taking an average for all 30 periods. However, he also found r to be the mead of dividends
divided by mean of prices in this way r turns out to be 5%, which is the value we used in our analysis.

To estimate perfect foresight price (p*) we used the formula which relates prices now to dividends in
future p*=sum (d/1+r) as the terminal price tends to 0 an n tends to infinity. However, as we are using
the data of only 30 periods we have p*T=pT (actual price at Terminal date) and use the formula p*t=
(d/1+r)n + pT/(1+r)T. As we have data in actual prices and dividends we estimate the p* using actual
dividends and compare it to actual price p observed in the data set. After acquiring p* for each 30 t’s
we can compare the variances between p* and p. In our data var(p*) is 5556 and var(p) is 54148,
which completely violates the inequality above.

To estimate p* I have used mat Lab with following script. And the values I got from it, were pasted
into excel for variance calculations:
(b) As we have 29 observations then it’s OK to use normal distribution t-statistics instead of the
student t-stat. We calculate the t-stat for the coefficient as 1-0.9/0.07=1.42 meaning that it’s not
significantly different from 1 at the 95% confidence level meaning there is a high probability of a unit
root in prices. Therefore, the whole excess volatility test made above is wrong as unit root implies
non-stationarity and hence the standard errors are wrong and the results are sensitive to stationarity
assumption.

3.
(a)

(b)

Where bt is not stationary as the parameter 1/delta is larger than 1.

(c) As P* contains bubble term then it doesn’t reflect the NPV of Dividends. In addition, the model
based on only two periods, which makes current price related to expected future price and not only
dividends. If the model was to infinity, then P term would go to zero and P* would be the NPV of all
future dividends.
 (d) You could separate the equitation in (a) by taking out bubble term and testing the coefficient on
the bubble term and if it’s not significant then no bubble exists in the price of a stock.

4. Demand for real money balances is given by L(i1,tYt) where i1,t is nominal yield on a short dated bond
and supply of real money balances is given by Mt/Pt and in equilibrium Mt/Pt=L(i1,tY). From Fisher
Equation we get i1t=r1t+Ep1t, where r1t is real yield on short-dated bond, as we have sticky price then
Ep1t is set to 0. Sticky prices mean that in the short-term prices don’t alter themselves and only Mt
causes change in money supply. In response to monetary contraction, you would expect the nominal
yields on short-term dated bonds to rise, while yields on other bonds may rise or fall depending on
the maturity of the bond and the extend of nominal price rigidity. Under PEH the long-dated yields at
a point in time are the average of current and future short-dated yields. The yield on long-dated bonds
are expected to fall, as with decrease of money supply price of bonds will fall which increases the
yield. As medium-term yields won’t rise as much as short-term, we would expect the yield curve to be
less responsive to the change in monetary policy but twist in response to permanent changes.

It

In the long run we would expect the government to increase the money supply by buying back bonds
in order to stimulate the economy by lowering the interest rates, stimulating spending and thereby
increasing output. In this instance, the buying of bonds will increase the price of bonds and thereby
lowering yield to maturity. If, this is known by the public the YTM curve might not change while the
prices are sticky. In addition, with passage of time employment will return back to normal and interest
rates will rise therefore the further away from shock we get, the less effect will be on the long-dated
bonds.

5.
(a) -

(b) You could run a regression of st+1=a0+a1+ftt+1+ut+1 where if EMH is true a0 is insignificant and a1
should not be statistically different from 1. In addition, ut should be a white noise process which is
uncorrelated with ft and Durbin-Watson should not find autocorrelation. In older studies it has been
found that the EMH is held in earlier periods, now the hypothesis is largely unsupported which may
be due to excess FOREX volatility. Coefficients were found to be significantly less and negative from 1
– forward premium puzzle.

6.
(a) PEH states that the expected excess returns on the long bond, over the life of a short bond are zero
which implies that the coefficient of unity on the spread. As we have two spreads in our case the two
deltas have to be 0.5 for the average to work out. This has been tested in Bulkley and Giordano paper.

(b)-

7.
(a) As we are asked about exact inflation expectation we use the formula for expected inflation over
the holding period as follows:

,
where i is nominal yield and r is a real yield.
By using it we get the values of ((1.0525)/(1.0278))5=12.6%; ((1.0511)/(1.0254))10=28.08%
((1.0498)/(1.0278))20=68.72% for 5,10,20 years respectively.

(b) In order to get constant rate of annual inflation we use n-th root of total rate of expected price
growth which yields following results:
2.4% 2.5% 2.64% for constant rate of inflation for 5,10,20 periods respectively.

(c) In order to infer results above for the inflation expectation between two periods we need inflation
to be lognormal and to be serially independent. Also there should be no risk premium and the rates
to reflect only inflation risk.

(d) By using the formula for total expected inflation between two periods,

((1.024)5*(1.025)5)/(1.024)5=1.131, then take 5th root to get 2.5% inflation between year 5 and 10.
(1.025)10*(1.0264)10/1.02510=1.297, then taking 10th root to get 2.64% inflation between 10 and 20
years.

(e) Approximation which can be used to back out annual average expected rates between periods can
be as follows:

Under PEH, nominal minus real yields are approximately equal to the average expected inflation rates
over the time period. We calculate total inflation over 5 and 10 year periods by multiplying inflation
values from the table, then take the difference and divide by 5 to get 2.67% inflation between year 5
and 10, similarly for years 10-20 we yield 2.85%.