AN EMPIRICAL ANALYSIS OF PRICING DUTCH REVERSE

CONVERTIBLE BONDS

Marta Szymanowska†, Jenke ter Horst‡, and Chris Veld‡‡,1

March, 2004

JEL classification: G10 and G19

Keywords: alternative investments, reverse convertible bonds, option
pricing, behavioral finance, framing

†
Corresponding author. Department of Finance and CentER, Tilburg University, P.O. Box 90153,
5000 LE Tilburg, the Netherlands, email: m.szymanowska@uvt.nl, phone: +31 13 4668705, fax: +31
13 466 2875.
‡
Department of Finance and CentER, Tilburg University, P.O. Box 90153, 5000 LE Tilburg, the
Netherlands, email: j.r.terhorst@uvt.nl, phone: +31 13 4668211, fax: +31 13 466 2875.
‡‡
Department of Finance and CentER, Tilburg University, the Netherlands, Tilburg University, the
Netherlands, P.O. Box 90153, 5000 LE Tilburg, the Netherlands, email: c.h.veld@uvt.nl, phone: +31
13 4663257, fax: +31 13 466 2875.
,1
The authors thank Jan Potters, conference participants at the ENTER Conference in Barcelona
(January 2004), and seminar and workshop participants at Tilburg University for helpful comments
and suggestions. The usual disclaimer applies.
 Abstract

The question regarding the deviation of the values of assets from their
fundamental levels has gained a lot of interest in the financial literature.
The recent increase in the global demand for derivatives has shifted the
attention from common stocks to derivatives. We focus on one type of
derivatives, namely reverse convertible bonds. These are bonds that give
right to a high coupon rate and at maturity the issuer has an option to either
redeem the bond at par in cash, or to deliver a pre-specified number of
common stocks. Therefore, they are in fact a combination of a bond and a
written put option. In this paper, we document a significant overvaluation,
of about 30%, of Dutch reverse convertible bonds in comparison to the
theoretical price. The documented overvaluation is robust with respect to
the estimation errors of the parameters and is persistent for approximately
one-fourth of the lifetime of the reverse convertibles. The results seem to
suggest that this overvaluation may not be driven by rational factors such as
transaction costs, but rather by behavioral reasons, such as the framing
effect.

2
1 Introduction

The question concerning the deviation from rational values of assets has
gained a lot of interest in the financial literature. Episodes of significant
overvaluation in the Treasury bond market were documented in e.g Cornell
and Shapiro (1989); Jordan and Jordan (1996). They found significant
overpricing over a certain period and potential arbitrage opportunities that
could not be attributed to changes in fundamentals.
Recently, the global equity derivatives market has become popular and is
growing rapidly. According to the Bank of International Settlements its
volume of trade has expanded by 52% in the period from June 1998 till June
2001 (Green, 2002). Among the derivatives, the convertible instruments
have reached their global issuance peak in 2001. This was driven by strong
demand from private customers as well as from hedge funds and
corporations (Green, 2002). As a consequence of such a shift in the
interests of investors, the market was flooded with a large number of
novelty derivative products. With time plain vanilla products became less
popular as investors were demanding more sophisticated capital protections,
and the derivative products started to include barrier clauses reducing
downside risk.
In this context reverse convertible bonds became very popular, especially in
countries such as Japan, UK, France, Italy, Germany and the Netherlands.
A reverse convertible bond (RC) is a bond that can be exchanged into shares
of common stock at the option of the issuer. In fact, it is a bond in
combination with a written put option. In order to compensate investors for
the possible loss due to written put option, the bonds carry very high coupon
rates. These rates vary according to the conversion conditions and to the
nature of the underlying shares; however, coupon rates as high as 20% have
been observed in this market. The market for the reverse convertibles
started originally with plain vanilla bonds. Lately, these bonds often include
a knock-in or knock-out clause. By appropriately specifying the barrier
level such clauses may reduce the downside risk of the investor’s.
The risk profile of plain vanilla reverse convertibles is similar to risk of
investing in shares although it is reduced by the high coupon interest. Since
at the maturity the issuer has the right to redeem either in cash or by
delivering a fixed number of shares, there is always a chance that the market
may collapse and that the bond will not be redeemed at par. This risk is
further reduced by introducing “protective barriers”. The knock-in reverse
convertible is initially the same as a “normal” bond. When the price of the
underlying security reaches the knock-in level this “normal” bond becomes
a reverse convertible. The risk of the investments is reduced; a knock-in

3
reverse convertible bond with a high barrier level has a high chance of
redemption in cash in a bearish market. Moreover, during the lifetime of the
bond the investor receives a relatively high coupon. The knock-out reverse
convertible works in the opposite direction. It starts as a plain vanilla
reverse convertible but it turns into a “normal” bond after the pre-specified
barrier has been reached.
However, some of the investors do not seem to be able to realize the true
riskiness of reverse convertibles. Ignoring the risk component has a
significant impact on the perception of these products. Given that investors
observe a high coupon rate, while they ignore the risk of a possible
redemption of the bond below its par value, it becomes less surprising that
reverse convertibles have gained a large popularity. These financial
instruments offer a high yield, which is desirable by investors especially in
economies with a low interest rate. Moreover, if the market is bullish - it is
less likely that the underlying value will fall below the exercise price of the
option - the investor receives a high coupon for the bond that most probably
will be redeemed at the full value. Such reasoning may be misleading as in
principal, the bond might be redeemed below its par value. The risk of
reverse convertibles is present in the option part: it is the issuer who can
decide at the maturity, whether to redeem the bond at or below par value.
Buying a reverse convertible issued on some common stock is equivalent to
buying a coupon-paying bond of the same issuer and selling the put option
on the same stock.
In line with the above argument the practical literature often suggests that
reverse convertibles are overpriced1. In this paper we study the pricing of
the Dutch reverse convertible bonds. In the Netherlands, the reverse
convertible bonds are very popular, especially with private persons2. This is
remarkable, since in the tax system that prevailed until 2001 they were taxed
very unfriendly. In that tax system, the whole coupon payment on the bond
was taken into account as interest. This interest was taxed at the same
progressive tax rate as, for example, income from employment3. This was
criticized by the issuing banks, which argued that potential capital losses,
because of redemption in stocks, were not tax deductible4. From January 1,
2001 the taxation of interest and dividend income was replaced in the

1
See for example: “Tante Agaath belegt in olies (transl. Aunt Agaath invests in stocks of Royal
Dutch)”, Het Financieele Dagblad, November 13, 1999.
2
See for example: “Aandelen kopen tegen heug en meug (transl. Buying stocks against your
wishes)”, Het Financieele Dagblad, October 5, 2001.
3
There was a tax exemption for interest income of 1,000 guilders per person.
4
See Het Financieele Dagblad of February 26, 1999: “Fiscus gaat rente op reverse convertible notes
belasten”. The taxation of reverse convertibles is discussed in detail by Bierlaagh (1999).

4
Netherlands by a wealth tax. This also solved the problem of the unfriendly
taxation of reverse convertible bonds5.
Financial products that combine elements from the spot and the futures
markets are often referred to as “structured products”. Despite the earlier
mentioned large extent of these markets, there has been very little empirical
research on this topic until now. Roberts et al. (2002) study stock-index
linked debt in France. They find that these bonds are significantly
overvalued by approximately 12%. Wilkens et al. (2003) study discount
certificates and reverse convertibles in Germany. Discount certificates very
much resemble reverse convertibles. With the purchase of a discount
certificate the holder acquires a bundle of shares at a “discount” compared
to the current market price. At the maturity date these shares are delivered
to the holder if the total value of the shares does not exceed a pre-specified
maximum repayment amount. Otherwise, the certificate pays this pre-
specified amount in cash. Wilkens et al. (2003) study daily quotes of
reverse convertibles and discount certificates in November 2001. They find
an average overpricing of 3.04% of reverse convertibles and 4.20% for
discount certificates over replication strategies that use options. Burth et al.
(2001) study the same financial instruments for Switzerland. They compare
issuance prices, derived from the issuers’ published term sheets, with
replication strategies that use options. They find an average overpricing for
reverse convertibles of 3.22% and for discount certificates of 1.40%.
In order to investigate the pricing of reverse convertibles, we compare the
market price with the theoretical price induced by the bond and long term
put option. Our approach differs from that of Burth et al. (2001) in two
important aspects. First we look at trading prices rather than the
information published in the issuers’ published term sheets. The most
important reason for this is that, at least in the Netherlands, the issuers do
not bind themselves with these prices. Second, we are able to use long term
options to price the option part in the reverse convertibles, thereby avoiding
pricing errors based on differences in the term structure of implied
volatilities. Such options do not exist in the Swiss market. We find a
significant overpricing of Dutch reverse convertibles issued from January 1,
1999 to December 31, 2002. On average they are overpriced by more than
23% with a median of around 19%. Plain vanilla reverse convertible bonds
are overpriced more than knock-in with averages of 29% and 19%, and
medians of 22% and 15% respectively. To calculate the overpricing we
used 2 different model specifications and we found that the difference
between these models is negligible. Moreover, the documented
overvaluation is robust with respect to the estimation errors of the

5
See Meussen (2000) for a detailed description of the income tax system that prevails in the
Netherlands from January 1, 2001.

5
parameters and is persistent for approximately one forth of the lifetime of
the reverse convertibles.
We investigate a broad set of possible explanations for the observed
mispricing. We first study rational explanations such as arbitrage
opportunities, transaction costs, and illiquid trade among others.
Furthermore, we also investigate other explanations that go beyond
rationality. Behavioral finance literature shows that investors may not be
“fully” rational in their decisions. They might rather be “under influence”
and this bias may create market inefficiency in the shape of mispricing
(Hirshleifer, 2001). Shefrin and Statman (1993) also argue that a different
representation of the same product might yield a different price. This
argument may also applied to the design of reverse convertibles. As a
reverse convertible is pronounced to be a bond, it is likely that investors
underestimate the risk attached to this investment. The fact that knock-in
reverse convertibles are less overpriced than plain vanilla reverse
convertibles supports these reasons for overpricing. The misperception of
true riskiness of knock-in reverse convertibles is smaller than for plain
vanilla reverse convertibles. The introduced protective barrier reduces the
risk of investment, narrowing the gap in risk between reverse convertibles
and low risk bonds. When investors place these contracts among other low
risk investments they apparently make smaller mistake than with plain
vanilla contracts.
The remainder of this paper is organized as follows. In Section 2 we present
the methodology used and we report some summary statistics of the data set.
In the following Sections 3 and 4, we present the results and a sensitivity
analysis. Section 5 presents a discussion of the results. Finally, Section 6
concludes.

2 Data and Methodology

In order to examine whether reverse convertible bonds are fairly priced we

compare the theoretical price with the observed market price. We define
overvaluation as the difference between full market price and the model
price relative to the model price:
Full _ market _ price − Model _ price
Overvaluation = ⋅ 100%
Model _ price
with both bond prices quoted in percentage. The full market price consists
of the price that is quoted at a certain date plus the amount of coupon
payments that occurred since the issuance of the reverse convertible. When
the result is negative we conclude that there is undervaluation.

6
2.1 Methodology

Reverse convertibles are convertible bonds that can be exchanged into

shares of common stock at the option of the issuer. Therefore, they are in
fact a combination of a bond and a written put option. The model price of a
reverse convertible has two components. The first component is an
otherwise identical bond issued by the same issuer, without the option for
the issuer to pay back the bonds by delivering shares. The second
component is the value of the embedded put options. The premium of this
written put option is reflected in the price of the reverse convertible by
reducing the value of the bond with the put value. The number of put
options is defined by the conversion rate.
Price(Reverse Convertible) = Price(Coupon Bond) - Price(Put Options)
We firstly discuss the bond valuation. Secondly, we show the valuation of
the put option.

2.1.1 Bond Valuation

In order to derive a theoretical price of the coupon-paying bond we use a
standard discounted cash flow approach. The price of a coupon-paying
bond can be expressed in the following way:
m
I F
P=∑ + , (1)
t =1 (1 + k ) t
(1 + k ) T −t
where
t = the time when the coupon is paid,
T-t = the maturity time,
F = is the Face Value or Nominal Value of the bond,
I = the amount of coupon paid,
k = the bond specific effective yield.
A fact that complicates the pricing of bonds is the impossibility of observing
directly in the market the effective yield. A reverse convertible is usually
issued by banks implying that a premium needs to be incorporated in order
to reflect the credit risk involved in issuing them. Such a premium is
specific for each institution and each financial instrument. We proxy this
rate by calculating for each issuer the average credit spread induced by 2-

7
year non-convertible bonds6. In that case the bond specific yield (k) is the
sum of the risk-free rate (rf) and the average credit spread (rc):
k = r f + rc . (2)

2.1.2 Option Valuation

In order to price the put option we use the Constant Elasticity of Variance
(CEV) models7. This class of models constitutes the relevant framework for
our research because it allows for testing several specifications of the
relation between the dynamics of option volatility and stock price dynamics.
The core of these models is the assumption that the underlying stock price
returns follow a constant elasticity of variance diffusion process:
dS = µSdt + σS β / 2 dW , (2)
where
S = the price of the underlying stock,
µ = the expected rate of return on the stock,
W = the Brownian Motion,
(β-2) = the elasticity of variance with respect to the price,
σ = the volatility of the stock price returns given by the following
equation,
σ ( S , t ) = δ ( S ( β −2) / 2 ) ).
We use two special cases of this model, i.e. the Black and Scholes (1973)
model and the Square Root model (Cox and Ross, 1976), both corrected for
continuous dividend payments (Merton, 1976).

When β in the CEV model is equal to two the volatility is constant and the
process of underlying stock price returns follows a lognormal diffusion
process. Hence, the standard Black-Scholes (BS) model is one of the
special cases of the CEV model. It is well-known that the world is not as
simple as the Black-Scholes model assumes. For instance, the volatility is
not constant over time and over different strike prices and maturities of the
options, the underlying security price dynamics cannot be represented by a
geometric Brownian motion (see e.g. Tucker et al., 1988). Therefore, we
require a different specification.

6
See Roberts et al (2002).
7
See Cox and Ross (1976).

8
When 0≤β<2 the volatility and the price are inversely related. This is in
line with several theoretical and empirical papers, see e.g. Beckers (1980)
and Schroder, (1989)8. We use one special case within this specification,
namely the Square Root (SR) model, where β equals one. In this model the
volatility is inversely related to the square root of the stock value. Since the
maturity of our sample reverse convertibles is almost always 2 years it is
likely that some of underlying stocks paid out dividends. Therefore, both
models are corrected for continuous dividend payments.

2.2 Data Description

We analyze a sample of reverse convertibles that were issued from January

1, 1999 to December 31, 2002 and that are listed on Euronext Amsterdam
(the Amsterdam Stock Exchange), and for which long-term call options are
outstanding. The reverse convertibles were identified from the financial
newspaper De Officiële Prijscourant van de Effectenbeurs (the official
newspaper of the stock exchange).
The sample consists of plain vanilla reverse convertibles (RCs) and several
variations, such as knock-in RC, and knock-out RC. A “knock-in” RC
initially starts as a “normal” bond and when the underlying hits a barrier it
becomes a reverse convertible. A “knock-out” RC works in the opposite
direction. It starts as a plain vanilla reverse convertible but it turns into a
“normal” bond after the pre-specified barrier has been reached. The risk of
a payoff in shares only arises if the price crosses (“knock-in”) or does not
cross (“knock-out”) the barrier. Consequently, this risk is lower compared
to a plain vanilla reverse convertible and therefore we expect that the price
will also be lower. The idea of pricing these products is equivalent to the
plain vanilla reverse convertibles. However, we use barrier options instead
of simple put options.
The sample consists of 108 reverse convertibles. All reverse convertibles in
our sample have a maturity of two years. The sample can be sub-divided
into three groups: plain vanilla reverse convertibles, “knock-in” reverse
convertibles and “knock-out” RC. More details on the descriptive statistics
of our sample are given in Table 1.

8
Beckers (1980) mentions two arguments for the use of a model in which the volatility is inversely
related to the stock value, i.e. operating and financial leverage. The former is based on the fact that an
increase in the stock price may reduce the variance of the stock's returns through the reduction of the
debt to equity ratio even when a firm has almost no debt. The latter argument stems from the fact that
since every firm faces fixed costs a decrease in income will decrease the value of the firm and
increase its riskiness.

9
 [Please insert Table 1 here]
There are four banks issuing the reverse convertibles in our sample. These
are ABN AMRO, Fortis Bank, ING Bank, and Rabo Securities. There are
47 plain vanilla convertibles, the majority of which is issued by either ABN
AMRO (21) or ING Bank (16). Rabo Securities has issued 29 knock-in
reverse convertibles and Fortis Bank has issued 14 out of the 54 knock-in
bonds. Apparently, the issuers in our sample are specializing either in the
plain vanilla or in the knock-in reverse convertibles.
Information on the expiration date, coupon rate, conversion price (defined
as the nominal value divided by the number of shares that can be paid at the
redemption), barrier levels, basic characteristics of the underlying assets
were derived from the prospectuses. Market prices and trading volume for
all reverse convertibles and underlying stocks as well as all interest rates are
retrieved from Datastream. In cases when we had information on options
but prices and trading volumes of reverse convertibles were missing we
gathered this information from De Officiële Prijscourant van de
Effectenbeurs (the official newspaper of the stock exchange).
We did not treat the issue price given in the prospectuses as the market
value of the considered reverse convertible for several reasons. First, not all
prospectuses mention such a price. Second, in case it is mentioned, the
issuers do not consider it to be binding. Finally, some of the RCs are not
actively traded directly after the issuance. Therefore, we took the first
trading price at which Datastream reports a positive volume as an issue
price. Moreover, we took a window of five days at which Datastream
reports a positive trading volume in order to avoid the possibility that the
results will be driven by nonsynchronous trade of reverse convertible, stock
or option.

2.3 Estimation Procedure

2.3.1 Pricing bond component

In order to price the bond part of the RC we used the following variables:
(1) the amount of the coupon paid (I), (2) the dates when the coupon is paid
(t), (3) face value of the bond (F), (4) time to maturity (T-t), and (5) the
bond specific effective yield (k).
The first four variables could be observed directly. The last variable, i.e. the
effective yield, had to be estimated based on the average credit spread.
The average credit spread for each issuer was based on all 2-year non-
convertible bonds that were available in Datastream during period January

10
1999 and December 2002. The spread for each bond was calculated as the
difference between the yield on the bond and the appropriate risk free rate
proxied by the yield on government zero-coupon bonds. The horizon of the
bond and risk free rate was matched on a monthly basis 9.

2.3.2 Pricing option component

To price the option part of a reverse convertible we used the following
variables: (1) price of the underlying stock (S), (2) strike price (X), (3) time
to maturity (T-t), (4) dividend yield (q), (5) the risk free interest rate (rf),
and (6) the volatility of the returns on the underlying stocks (σ).
The first four variables were retrieved directly from Datastream. The fifth
variable, the risk free rate, was approximated as the average yield on
government bonds with a maturity of 2 years10. The final variable, the
volatility, had to be estimated. There are two ways of estimating the
volatility. The first is based on the realized, historical volatility of the stock
returns. The second, is based on the volatility implied by observed option
prices. A large body of the literature showed that the implied volatility is
superior to the historical, data-based forecasts and that it can be seen as the
“market’s volatility forecast” (e.g., Schwert, 1991; Beckers, 1981; Amin
and Ng, 1997). In line with this claim, we follow a number of other option
pricing studies and use the implied volatility, see e.g. ter Horst and Veld
(2002).
To price the option part of the reverse convertible we had to calculate the
volatility implied by long-term options. There are two disadvantages of
using long term put options: (1) The long-term put options traded on the
exchange in Amsterdam are of the American type, but the put that we had to
value is of the European type; (2) More importantly long-term put options
can be very illiquid, implying that the implied volatility is unreliable. Long-
term call options do not suffer from these drawbacks as early exercise is less
likely and they are often more liquid11. In order to obtain the most accurate
estimate of the implied volatility we restricted our sample to reverse
convertibles written on underlying stocks, for which long-term call option
were traded.

9
For example, if the bond matures in January 2002 we calculate the spread in January 2000 as the
difference between the yield on this bond and the 2 years risk free rate, in February 2000 we use the 1
year and 11 month risk free rate, etc.
10
See e.g. Hull (2003, pg 247), Roberts et al. (2002).
11
For stocks in our sample, on average, trading volume of call option was higher then of put options.

11
We retrieved the information on the long-term call options from Datastream.
A problem that arises is that usually the maturity and the strike price of the
reverse convertible are different from the maturity and the strike price of the
long-term call options. This can lead to a “volatility smile” (implied
volatility as a function of its strike price) and “volatility term structure”
(implied volatility as a function of its time to maturity), see e.g. Hull (2003,
pg 334-336). In order to correct for these, we used a weighted average of
implied volatilities with respect to time to maturity and strike price, see e.g
ter Horst and Veld (2002). We used two different estimates for the implied
volatility. The first is based on the Black-Scholes model, while the second
one is based on the Square Root version of CEV model.
Once we estimated all six necessary parameters, where we used two
different estimates for the implied volatility, the put option was priced using
the same two models, i.e. Black-Scholes and Square Root version of CEV
models.

2.4 Summary of the sample

We restricted our original sample of reverse convertibles listed on Euronext
Amsterdam and issued between January 1, 1999 and December 31, 2002
due to several reasons. The details regarding the sample selection procedure
are given in Table 2.
[Please insert Table 2 here]
In order to price reverse convertibles; first, the main characteristics of these
products were taken from their prospectuses. 18 reverse convertibles were
excluded because it was not possible to find information on prices and
trading volume for both reverse convertible and the underlying option. In
case the information on the option was available and it was not possible to
find in Datastream information on the reverse convertibles the dataset was
filled up with quotes published in the official newspaper of the stock
exchange (De Officiële Prijscourant van de Effectenbeurs). Finally, 15
observations were excluded because we could not identify 5 days at which a
price with a positive trading volume would be quoted. In total we were able
to price 75 reverse convertibles (70% of the original sample) where 32 were
plain vanilla (68%) and 43 are knock-in (80%).
The summary statistics for the final sample are presented in Table 3.
[Please insert Table 3 here]
Panel A gives a general description while panels B and C gives more
detailed statistics; per issuer and per underlying stock. The whole sample
consists of two groups of reverse convertibles; plain vanilla RC and knock-
in RC. The average price for the plain vanillas is between 82% and 94%
while for the knock-in it varies between 71% and 79%. Both groups of

12
contracts are actively traded; the average daily trading volume for plain
vanilla reverse convertible is approximately 250 millions reverse
convertibles and for the knock-in reverse convertible it is approximately 75
millions. Plain vanilla contracts seem to be more expensive and more
traded than knock-in reverse convertibles. The average conversion ratio is
94.99 and 78.59 shares per contract for plain vanilla RC and knock-in RC
respectively, which means that on average a plain vanilla can deliver more
shares. The average barrier level for knock-in reverse convertibles is 80%
of the closing price of the underlying stock at the pricing date. This means
that on average when the price of the underlying stock drops by 20% a
normal bond is converted into a reverse convertible bond.

3 Results

In order to examine whether reverse convertible bonds are fairly priced we

compare the theoretical price with the observed market price. We define
overvaluation as the difference between full market price and the model
price relative to the model price:
Full _ market _ price − Model _ price
Overvaluation = ⋅ 100%
Model _ price
with both bond prices quoted in percentage. The full market price consists
of the price that is quoted at a certain date plus the amount of coupon
payments that occurred since the issuance of the reverse convertible. When
the result is negative we conclude that there is undervaluation.
Table 4 presents the results calculated with Black-Scholes model (1973)
corrected for continuous dividends.
[Please insert Table 4 here]
Panel A gives a description for the whole sample and two groups of reverse
convertibles: plain vanilla and knock-in. From this panel it appears that on
average reverse convertibles are overpriced by 23.03% with a median of
19.06%. Plain vanilla reverse convertible seem to be more overpriced than
knock-in reverse convertibles. The average overpricing for the plain vanilla
is 28.75% with a median of 22.47%, while knock-in are overpriced on
average by 18.78% with median equal to 15.44%. All means and medians
in this panel are significant at the 1% level.
Panel B reports a more detailed description. The overvaluation is calculated
for each issuer in the whole sample and in two sub-groups. From this panel
it can be concluded that the overvaluation varies between issuers. In the
whole sample ABN Amro, ING Bank, and Rabo Securities value reverse

13
convertibles more than our theoretical price by 23-25%. When we consider
medians for these three issuers (16.89%, 22.35%, 30.55% respectively) we
can see that for ING Bank the mean and the median are close to each other
while for ABN Amro and Rabo Securities the median differs substantially
from the mean. Apparently, it is the case that for ING Bank the
overvaluation seems to be consistent over all issued reverse convertibles,
and for ABN Amro and Rabo Securities it can be driven by outliers. The
overvaluation of reverse convertibles issued by Fortis Bank is on average
equal to 19.55% with a median of 15.22%. All means and medians in this
panel, except one plain vanilla RC issued by Fortis, are significant at the 1%
level.
Panel C presents the overvaluation per underlying stock in the whole
sample. The most overvalued appeared to be reverse convertibles issued on
TPG stocks (average of 53.19% at 1% significance level). The least
overpriced or not overpriced at all are reverse convertibles issued on ASML,
DSM, Elsevier, Numico, and Unilever (the average overvaluation is not
significantly different from zero).
Apart from the presented Black-Scholes model with implied volatility based
on Black-Scholes model, we performed some robustness checks. We
calculated the overvaluation for the Square Root model with implied
volatility based on Square Root model, also corrected for continuous
dividend payments. Table 5 gives the overview of the results over all model
specifications used.
[Please insert Table 5 here]
It appears that on average reverse convertibles are overpriced by more than
23% with a median of around 19%. All means and medians in the Table are
significant at 1% significance level. Due to the fact that there is no closed
form solution for the price of the barrier options using Square Root version
of CEV model, we could only price plain vanilla reverse convertibles with
both models. However, for these contracts, the difference between the
overvaluation estimated with both models is negligible; on average plain
vanillas are overpriced by more than 28% with a median of more than 22%.
The fact that the overvaluation is consistent through different models
suggests that this overvaluation is unlikely to be explained by model
misspecification.

4 Sensitivity analysis

14
In the Section 3 we concluded that reverse convertibles are overvalued in
comparison to their theoretical prices. In order to assess the robustness of
this procedure we investigate these results in more detail.
The difference between the observed price and the model price can be due
to the model that we used to value the reverse convertibles. We used two
models that were shown to work superior in different settings. Black-Sholes
model was shown to price “approximately” correctly at-the-money options
(Emanuel and MacBeth, 1982). The Square Root version of CEV model
captures the empirically observed inverse relationship between volatility
and stock price and thus fits the market data better than the Black-Scholes
model. Hauser and Lauterbach (1997) showed that this model generated the
lowest average absolute pricing error out of 5 models tested in their study.
From Table 6 it can be seen that in this study overvaluation does not change
much across different model specifications. This may indicate that the
results are not simply due to model imperfection. Moreover, the magnitude
of the overvaluation (approximately 23%) may also suggest that it is less
likely to find a model that will fully explain this overvaluation.
Apart from the model specification, misspricing can stem from the
estimation error. We performed a sensitivity analysis for parameters that
were either estimated or we used proxies for. We tried several values of
implied volatility, effective yield, and risk free rate. We considered values
larger and smaller by 10% from the initial parameter value. We tried
different values separately for each parameter as well as in combinations.
The results do not differ much across different specification and are
available from authors upon request.

5 Discussion

5.1. Introduction
In this paper we document a significant overvaluation in the Dutch reverse
convertible market, on average 28.75% for plain vanilla reverse convertibles
and 18.78% for barrier reverse convertibles.
5.2. Regression analysis
To test formally the possible reasons for overpricing we perform a
regression analysis of the documented overpricing. As the explanatory
variables we use price, trading volume, and conversion ratio of reverse
convertibles; indicators for each issuer; an indicator for plain vanilla reverse
convertibles; and the barrier level defined as the percentage of the stock
price for the knock-in reverse convertibles.

15
Given that we define the overpricing as the difference between the observed
full market price and the theoretical price, we expect that the higher the
observed full market price the higher the overpricing. In contrast, we expect
the volume of trade of reverse convertibles to have a negative impact on the
overpricing since larger deals may be sold at lower prices. A barrier level is
defined as the percentage of the stock price; when the barrier level is high
the stock price has to drop less in order for a normal bond to be turned into a
reverse convertible bond. This leads to a higher risk since there is a higher
chance that the bond will be redeemed in shares, below its par value. Since
investors seem to underestimate the true risk of reverse convertibles, we
expect that a higher risk of reverse convertibles will lead to a higher
overpricing. Hence, the barrier level should have a positive impact on the
overpricing. The results from the regression analysis are presented in Table
6.
[Please insert Table 6 here]
Table 6 reveals that, as expected, trading volume has a significant and
negative impact on the overpricing. This holds for the whole sample (5%
significance level) and for the plain vanilla reverse convertibles (1%
significance level). Moreover, the plain vanilla reverse convertibles are
significantly more overpriced than knock-in reverse convertibles (1%
significance level). In line with our expectation, the coefficient for the
barrier level has a positive sign. The increase in the barrier level
significantly increases the overpricing through the increase of risk of the
investment (5% significance level).
5.3. Rational explanations
One of the possible explanations of such high and significant overpricing
can stem from the absence of arbitrage possibilities. We based our
valuation on a non-convertible bond with the same characteristics as the
reverse convertible. In principal, the non-convertible bonds do not carry
such high coupons as reverse convertibles. Thus, investors are not able to
exploit observed overvaluation and profit from arbitrage opportunities.
This, however, does not explain the existence of overvaluation. The fact
that arbitrage is not possible should not stop other market forces from
correcting the prices, for instance supply-demand force should drive prices
around their fundamental levels.
Figure 1 shows the relation between the average overpricing of reverse
convertibles and their remaining time to maturity. It indicates that the
reported overpricing is persistent for almost a year after the issuance of the
reverse convertible. The formal test of the significance of the overpricing
revealed that after approximately half a year the reported averages are not
significantly different from zero at 10% significance level. Since the
averages are calculated across reverse convertibles that were issued between

16
January 1999 and December 2002, a pattern observed in Figure 1 cannot be
directly linked to the performance of the market.
[Please insert Figure 1 here]

Frictions that might be present on the market and that our models fail to take
into account may influence the documented in this study overpricing. For
instance transaction costs and/or illiquid trade may result in a large spread
between bid and ask price. We tried several specifications in a sensitivity
analysis and we found that it is unlikely that these frictions could explain
such high overpricing12. Additionally, transaction costs might play an
important role since we compare a strategy that buys one instrument
(reverse convertible) to the strategy that involves two instruments (bond and
option). In principal, the transaction costs associated with the second
strategy will be higher. However, in the regression analysis (see Section
5.2) we were not able to detect a significant relationship between the full
market price of RC and its overpricing. Thus, it is implausible that
transaction costs would account for the overpricing in the magnitude of
30%.
5.4. Behavioral explanations
Behavioral finance literature may supply other, possible reasons for the
documented overpricing. There exist a wide body of literature within this
field that looked at the effects of frames on choices13. Framing is a part of a
decision process undertaken by a prospect theory investor (Kahneman and
Tversky, 1979). In this theory, a decision is described as the evaluation of
the potential gains and losses relative to some reference point. They
distinguished between two sequential operations that lead to such a
decision: an editing stage and an evaluation stage. The former phase
consists of preliminary analysis and simplification of the choice problem.
This can lead to Thaler’s mental accounts (Thaler, 1985), where gains and
losses are kept in separate mental accounts. Creation of such accounts can
be influenced by the way the choice problem is presented, for instance,
different reference points for comparing the outcomes, different definition
of the choice problem that are associated with different emotions. Tversky
and Kahneman (1981) and many others14, showed that the presentation of
logically identical decision problems had large framing effect on choices.
The latter stage of decision problem involves the application of decision

12
See details in Section 4.
13
See, for instance, Kahneman and Tversky, 1979; Elliott and Archibald, 1988; Shefrin and Statman,
1993; Loke 1989; Andreoni 1995.
14
op.cit.

17
rules to the framed accounts. The choices are evaluated and the one of
highest value is selected. Shefrin and Statman (1993) showed that the
framing effect, among other behavioral finance reasons, have led to the
success of covered calls.
We believe that this framing effect is also present in RCs. The strategy of
buying a bond and writing a put option is now presented as a strategy of
buying a high-yield bond. The financial newspaper Het Financieele
Dagblad of February 1, 2003 gives the following quote from a press release
of the ING bank of a reverse convertible bond on Ahold: “A two-year bond
with a coupon interest of no less than 15% per year with the additional
prospect of receiving a solid package with shares of Ahold”. Investors often
do not see that a RC is a combination of a bond with a written put option. In
an article in the financial newspaper Het Financieele Dagblad it is argued
by a derivatives specialist: “A lot of investors actually do not see that a
reverse convertible consists of two products. They often think that it is a
bond with a high coupon interest and that they can decide themselves what
will happen at the end of the maturity”15. The same specialist also argues:
“Options only deter (..). However, if you combine it with a stock or a bond,
or if you give it a beautiful name, people are often willing to buy it”16. This
strengthens the idea that framing matters. In the last mentioned article the
specialist also argues that RCs are mainly bought by investors who are not
aware of options. In addition another derivatives specialist argues that RCs
are mostly bought by “bond investors” who normally do not trade on the
stock exchange17,18. Kim and Zumbansen (2002) describe a court case in
Germany where a plaintiff claimed compensation from a bank for the losses
incurred in buying reverse convertible bonds. The plaintiff argued that he
was not properly informed about the nature and risk of the transaction. This
also illustrates that not all investors are aware of the fact that a reverse
convertible bond is a combination of a bond and a written put option.
Another behavioral element in the design of financial products is the
existence of cognitive errors. In this context Shefrin and Statman (1993)
refer to the market of LYONs that existed in the United States in the
beginning of the 1980s. LYONs are zero-coupon, convertible, callable and

15
See Het Financieele Dagblad of August 1, 2000: “Baan geeft knock-out aan reverse convertible
ING (transl. Baan gives knock-out to reverse convertible ING)”.
16
See Het Financieele Dagblad of February 12, 1999: “Aandelen duur? Koop ze met korting (transl.
Are stocks expensive? Buy them at a discount)”.
17
See Het Financieele Dagblad of February 1, 2003: “Risicovolle obligaties (transl. Risky bonds)”.
18
Another party that is present on this market are institutions that are legally restricted from investing
in shares. These parties use RCs as a means to circumvent regulation. See for example Het
Financieele Dagblad of January 16, 2002: “Ik heb geen inzicht in rendementen (transl. I do not have
any insight in returns)”.

18
puttable bonds. These bonds were very popular in times of high interest
rates. However, when interest rates started to fall, the issuers started to use
the call features of the bonds. This led to a quick disappearance of the
market for LYONs. According to Shefrin and Statman (1993), some of the
investors who bought LYONs overestimated the probability that they would
not be called, perhaps because they were not called in the recent past. The
authors refer to this as cognitive errors. The same pattern occurred in case of
RCs in the Netherlands. According to the earlier quoted derivatives
specialist most investors stopped buying RCs after the first RCs were
redeemed in shares rather than in cash.
The fact that knock-in reverse convertibles are less overpriced than plain
vanilla ones may further support the behavioral reasons. The misperception
of true riskiness of the knock-in reverse convertibles is smaller. The
introduced protective barrier reduces the risk of the investment, narrowing
the gap in risk between knock-in reverse convertibles and low risk bonds.
When investors place these contracts in one mental account together with
other low risk investments they make a smaller mistake than with plain
vanilla contracts.
Knock-in reverse convertibles seem to be traded less frequently. A bit
counterintuitive is the fact that from one pool of available investments
classified in one mental account, plain vanilla RC and knock-in RC,
investors seem to choose more overpriced products. This could further
support the view that investors are not aware of the true riskiness and thus
true value of reverse convertibles.

6 Summary and Conclusions

The question regarding the deviation of the values of assets from their
fundamental levels has gained a lot of interest in finance. Recent increase in
the global demand for derivatives (Green, 2002) has shifted the attention
from common stocks to derivatives. In this paper we focus on one type of
derivatives, namely reverse convertible bonds. These are bonds that give
right to a high coupon rate and at maturity the issuer has an option to either
redeem the bond at par in cash or to deliver a pre-specified number of
common stocks. Therefore, they are in fact a combination of a bond and a
written put option.
We studied the reverse convertibles issued on Amsterdam Stock Exchange
between January 1, 1999 and December 31, 2002 for which also long-term
call options were outstanding. We choose the Dutch market because on this
market the reverse convertibles and the long-term options are popular with

19
the investors. This allows us to price the reverse convertibles as the
combination of a bond and a put option.
The results indicate a significant overvaluation of reverse convertible bonds
in comparison to the theoretical price based on Black-Scholes (1973) and
Square Root version of CEV model (Cox and Ross, 1976). The documented
overvaluation is robust with respect to the estimation errors of the
parameters and is persistent for approximately one forth of the lifetime of
the reverse convertibles. Moreover, the results seem to suggest that this
overvaluation may not be driven by rational factors. Frictions present on the
market, such as transaction costs and illiquid trade, are unlikely to explain
the overvaluation of magnitude between 20 and 30%. The lack of arbitrage
possibilities shows that the investors are not able to exploit observed
overvaluation and profit from arbitrage opportunities. This, however,
should not stop other market forces from correcting the prices. Our results
suggest that the behavioral reasons, such as the framing effect and mental
accounts, may be useful in explaining the documented overpricing.
Whether the data will corroborate this hypothesis is left for further research.

20
References
Amin, K.I., and Ng, V.K.: 1997, Inferring future volatility from the
information in implied volatility in Eurodollar options: a new approach, The
Review of Financial Studies 10(2), 333-367.
Andreoni, J.: 1995, Warm-glow versus cold-prickle: the effects of positive
and negative framing on cooperation in experiments, Quarterly Journal of
Economics, 1-21.
Beckers, S.: 1980, The Constant Elasticity of variance model and its
implications for option pricing, The Journal of Finance 35 (3), 661-673.
Beckers, S.: Standard deviations implied in option prices as predictors of
future stock price variability, Journal of Banking and Finance, 363-381.
Bierlaagh, H.M.M.: 1999, EC-update – What’s going on in the Netherlands:
tax authorities state position on tax treatment of reverse convertible notes,
European Taxation, 332-333.
Black, F., and Scholes, M.: 1973, The pricing of options and corporate
liabilities, Journal of Political Economy 81, 637-659.
Burth, S., Kraus, T., and Wohlwend, H.: 2001, The pricing of structured
products in the Swiss market, The Journal of Derivatives, 30-40.
Cornell, B., and Shapiro, A.C.: 1989, The mispricing of U.S. Treasury
bonds: a case study, The Review of Financial Studies 2(3), 297-310.
Cox, J.C., and Ross, S.A.: 1976, The valuation of options for alternative
stochastic processes, Journal of Financial Economics, 145-166.
Elliott, C.S., and Archibald, R.B.: 1989, Subjective framing and attitudes
towards risk, Journal of Economic Psychology, 321-328.
Emanuel, D.C., and MacBeth, J.D.: 1982, Further results on Constant
Elasticity of Variance call option pricing model, Journal of Financial and
Quantitative Studies, 533-554.
Green, I.: 2002, Pushing the equities button, Risk March 2002, 69-70.
Hauser, S., and Lauterbach, B.: 1997, The relative performance of five
alternative warrant pricing models, Financial Analysts Journal, 55-61.
Hirshleifer, D.: 2001, Investor psychology and asset pricing, The Journal of
Finance, 1533-1597.
Hull, J.C.: 2003, Options, futures, and other derivatives, Fifth Edition,
Prentice-Hall International Inc., Upper Saddle River, New Jersey, 2003.
Jordan, B.D., and Jordan, S.D.: 1996, Salomon brothers and the May 1991
Treasury auctions: analysis of a market corner, Journal of Banking and
Finance 20, 25-40.

21
Kahneman, D., Tversky, A.: 1979, Prospect Theory: an analysis of decision
under risk, Econometrica, 263-292.
Kim, S.S., and Zumbansen, P: 2002, The risk of reverse convertible bonds:
German capital market law and investor protection, German Law Journal,
1-12.
Litterman, R., and Iben, T.: 1991, Corporate bond valuation and the term
structure of credit spreads, The Journal of Portfolio Management 17 (3), 52-
64.
Loke, W.H.: 1989, The effects of framing and incomplete information on
judgments, Journal of Economic Psychology, 329-341.
Merton, R.C.: 1973, Theory of rational option pricing, Bell Journal of
Economics and Management Science, 141-173.
Meussen, G.T.K.: 2000, Netherlands: Income tax act 2001, European
Taxation, 490-498.
Roberts, G.S., Vijayraghavan, V., and Aintablian, S.: 2002, Stock index-
linked debt and shareholder value: evidence from the Paris Bourse,
European Financial Management, 339-356.
Schroder, M.: 1989, Computing the Constant Elasticity of variance option
pricing formula, The Journal of Finance 44 (1), 211-219.
Shefrin, H., and Statman, M.: 1993, Behavioral aspects of the design and
marketing of financial products, Financial Management, 123-134.
Schwert, W.: 1990, Stock volatility and the Crash of ’87, Review of
Financial Studies 3, 77-102.
Ter Horst, J., and Veld, C.: 2002, Behavioral preferences for individual
securities: the case for call warrants and call options, Working Paper,
Tilburg University.
Thaler, R.: 1985, Mental accounting and consumer choice, Marketing
Science 4, 199-214.
Tucker, A.L., Peterson, D.R. and Scott, E.: 1988, Tests of the Black-Scholes
and Constant Elasticity of variance currency call option valuation models,
The Journal of Financial Research 11 (3), 201-213.
Tversky, A., Kahneman, D.: 1986, Rational choice and the framing of
decisions, Journal of Business, 251-278.
Wilkens, S., Erner, C., Lamprecht, D., and Röder, K.: 2003, The pricing of
structured products – an empirical investigation of the German market,
Working Paper, University of Münster (Germany).

22
Table 1: Summary statistics. This table presents a summary of the sample reverse convertibles
(RCs) issued on Amsterdam Stock Exchange between January 1, 1999 and December 31, 2002. A
plain vanilla reverse convertible is a bond that at the maturity can be either redeemed in cash or by
delivery a pre-specified amount of common stocks. A “knock-in” RC initially starts as a “normal”
bond and when the underlying assets hits the barrier it becomes a reverse convertible. A “knock-out”
RC works in the opposite direction. It starts as a plain vanilla reverse convertible but it turns into a
“normal” bond after the pre-specified barrier has been reached. Panel A presents the overview per
issuer and panel B gives the description per underlying stock.
Number of Reverse Convertibles
Panel A
Issuer
Plain Vanilla Knock-in Knock-out Total
ABN AMRO 21 7 5 33
Fortis Bank 3 14 17
ING Bank 16 4 20
Rabo Securities 7 29 2 38

Total 47 54 7 108

Panel B
Underlying Stocks
Plain Vanilla Knock-in Knock-out Total
ABN AMRO 5 6 3 14
Aegon 7 9 16
Ahold 3 9 1 13
Akzo Nobel 2 1 3
ASML 2 2
DSM 2 2
Fortis 4 3 1 8
Getronics 1 1
Heineken 2 2
ING Groep 7 4 2 13
KPN 4 4 8
Numico 1 1 2
Philips Electronics 3 3 6
Elsevier 1 2 3
Royal Dutch 5 6 11
TPG 2 2
Unilever 1 1 2

Total 47 54 7 108

23
 Table 2. Sample Selection. Column (1) presents all reverse convertibles that were listed on
Euronext Amsterdam from January 1, 1999 to December 31, 2002 for which long-term call options
were also outstanding. Column (2) gives these reverse convertibles for which it is not possible to find
at least five days with quoted prices and positive volume for the reverse convertible itself and for the
option (whenever the information on reverse convertibles were missing but the information on options
were available we have filled up our database with quotes from De Officiële Prijscourant van de
Effectenbeur). Column (3) gives these reverse convertibles for which there were less than 5 quoted
prices with positive volume either in Datastream or in De Officiële Prijscourant van de Effectenbeurs.
Columns (4) and (5) give the number of reverse convertibles in the final sample and the percentage of
the original sample respectively.
Number of Reverse Convertibles
Original No price Less than 5 quoted Final Sample Percentage
Sample information in prices with postive (4) priced
(1) Datastream trading volume (5)
(2) (3)
Plain Vanilla 47 13 2 32 68%
ABN AMRO 21 5 16
Fortis Bank 3 2 1
ING 16 3 1 12
Rabo
Securities 7 3 1 3
Knock-in 54 5 6 43 80%
ABN AMRO 7 1 6
Fortis Bank 14 1 13
ING 4 4
Rabo
Securities 29 4 5 20
Knock-out 7 7 0%
ABN AMRO 5 5
Fortis Bank
ING
Rabo
Securities 2 2

Whole sample 108 18 15 75 70%

24
Table 3. Descriptive statistics. The sample contains reverse convertibles that were listed on Euronext Amsterdam from January 1, 1999 to December 31, 2002 for
which also long-term call options were outstanding. Column (1) gives the size of the sample. Columns (2), (3) and (4) give average, median and standard deviation of the
price of reverse convertible per 100 face value. Columns (5) and (6) give average and median of the trading volume of the reverse convertibles. Columns (7) and (8) give
average and median of the conversion ratio, which determines the number of shares that may be delivered at redemption of the reverse convertible. Columns (9) and (10)
give average and median of the barrier level as the percentage of the stock price for the knock-in reverse convertibles. Panel A gives an overview of the whole sample. Panel
B gives the statistics per issuer in the whole sample, and separately for plain vanilla and for knock-in reverse convertibles. Finally, Panel C gives an overview per underlying
stock for the whole sample.
Avg Price Med Price Std dev of Avg TV Med TV Avg CR Med CR Avg Barrier Med Barrier
Size in % in % Price in thousand in thousands as % of stock as % of stock
s price price
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Panel A
Plain Vanilla 47 85.12 86.03 14.72 250784 182252 94.99 75.99
Knock in 54 76.25 75.86 13.82 74794 29865 78.59 60.38 80 75
The whole sample 101 80.37 80.59 14.24 158542 102336 86.22 70.55 80 75
Panel B
Whole sample
ABN AMRO 28 80.69 81.03 15.87 295309 236504 76.30 51.02 77 79
ING 17 81.07 82.18 16.01 103136 48842 79.69 72.15 78 79
Rabo Securities 20 81.65 81.88 12.85 99101 48425 93.96 75.99 83 75
Fortis Bank 36 76.35 75.36 12.62 119055 54595 95.86 63.98 78 75
Plain Vanilla
ABN AMRO 21 82.05 82.80 15.87 363868 296338 76.66 61.80
ING 3 83.55 84.98 16.06 127427 61333 79.82 71.35
Rabo Securities 16 93.89 94.68 11.19 171888 105055 134.54 105.99
Fortis Bank 7 93.90 93.67 8.26 260072 168373 211.94 197.86
Knock in
ABN AMRO 7 75.91 74.83 15.85 55350 27083 67.73 43.48 77 79
ING 14 71.78 71.71 15.84 12045 2000 76.90 78.44 78 79
Rabo Securities 4 78.70 78.79 13.25 80904 34268 85.12 70.92 83 75
Fortis Bank 29 72.59 71.44 13.55 88837 30214 70.99 55.46 78 75
Table 3 continued.
Size Avg Price Med Price Std dev of Avg TV Med TV Avg CR Med CR Avg Barrier Med Barrier
in % in % Price in thousands in thousands as % of stock as % of stock
price price
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Panel C
Underlying Stocks
ABN AMRO 11 86.27 85.45 10.13 178064 80383 124.22 81.53 74 75
Aegon 16 76.12 76.55 18.40 415266 370188 71.28 61.40 92 84
Ahold 12 76.84 79.77 19.01 93818 42083 96.51 89.83 80 85
KPN 8 61.02 54.42 16.89 148722 54593 129.42 124.20 81 81
Akzo Nobel 3 90.76 89.33 8.20 33735 11697 92.98 102.12
ASML Holding N.V. 2 67.34 68.75 13.70 33709 14000 13.22 13.22 75 75
DSM 2 100.26 100.65 2.70 40000 20000 49.38 49.38 74 74
Elsevier 3 92.53 93.92 7.30 231671 141667 242.85 322.98 74 74
Fortis 7 88.12 91.75 14.65 103089 69400 45.18 70.55 75 75
Getronics 1 84.61 83.95 3.04 38000 38000 20.16 20.16
Heineken 2 91.35 95.28 15.55 39282 22000 60.26 60.26
ING 11 77.09 77.31 16.81 108243 38624 46.69 52.08 79 79
Numico 2 66.09 71.25 17.56 20393 6750 79.52 79.52
Philips Electronics 6 70.86 69.54 15.23 70125 35833 59.17 52.88 73 75
Royal Dutch 11 92.28 92.62 10.10 101507 48808 72.59 52.26 82 84
TPG 2 93.81 92.63 6.16 32299 5060 197.36 197.36
Unilever 2 99.53 99.78 4.28 69051 24500 24.86 24.86 75 75

26
 Table 4. Results for the Black-Scholes model with implied volatility based on Black-
Scholes model. The Table presents overpricing of reverse convertibles that were listed on Euronext
Amsterdam from January 1, 1999 to December 31, 2002 for which also long-term call options were outstanding.
The overpricing for each reverse convertible is calculated as the average of the first five trading days. Column (1)
gives a size of a sample for each group. Column (2) gives the value of overpricing averaged over reverse
convertibles. If the value is positive it means that reverse convertible is overpriced. Next column, column (3)
gives the standard deviation of the overpricing. Columns (4) and (5) give minimum and maximum value in the
considered sample. Column (6) presents the median value of overpricing. Finally column (7) gives a number of
reverse convertibles that were overpriced. Significance of the average and median are calculated using t-test
appropriate for small sample. *** indicates the significance at 1%, ** the significance at 5% and * the
significance at 10%.
Size Avg St. dev Min Max Median Number of
Overpricing in % Overpriced
in %
(1) (2) (3) (4) (5) (6) (7)
Panel A
Plain Vanilla 32 28.75*** 14.80 12.34 57.36 22.47*** 32
Knock in 43 18.78*** 13.73 -1.79 49.87 15.44*** 42
Whole sample 75 23.03*** 14.95 -1.79 57.36 19.06*** 74
Panel B
Whole sample
ABN AMRO 22 24.85*** 17.10 6.36 57.36 16.89*** 22
ING 7 22.76*** 6.73 11.57 39.08 22.35*** 7
Rabo Securities 32 23.60*** 17.44 -1.79 50.42 30.55*** 31
Fortis Bank 14 19.55*** 14.59 3.69 55.84 15.22*** 14
Plain Vanilla
ABN AMRO 16 29.02*** 17.36 12.34 57.36 18.58*** 16
ING 3 24.50*** 6.56 16.97 39.08 22.59*** 3
Rabo Securities 12 35.27*** 18.91 14.07 50.42 41.32*** 12
Fortis Bank 1 55.84 1
Knock in
ABN AMRO 6 13.75** 10.97 6.36 35.70 10.42** 6
ING 4 17.53*** 4.50 11.57 21.57 18.50*** 4
Rabo Securities 20 21.85*** 17.02 -1.79 49.87 30.25*** 19
Fortis Bank 13 16.76*** 10.60 3.69 39.25 15.00*** 13
Panel C
Plain Knock-
Underlying Stocks Vanilla in
ABN AMRO 11 23.16*** 14.52 4.34 50.42 22.32*** 5 6
Aegon 12 25.01*** 12.99 8.01 51.34 20.31*** 4 8
Ahold 12 21.84*** 15.59 1.61 47.66 15.81*** 3 9
KPN 3 26.95** 13.19 13.01 39.25 28.57** 1 2
Akzo Nobel 2 27.70* 19.26 14.07 41.32 27.70* 2
ASML 1 7.89 1
DSM 1 15.44 1
Elsevier 2 21.46 22.12 5.81 37.10 21.46 2
Fortis 5 13.63* 14.04 -1.79 29.95 17.56** 2 2
ING 7 19.88*** 11.50 5.47 41.88 20.09*** 5 2
Numico 1 23.12 1
Philips Electronics 6 29.39*** 19.37 10.18 57.36 21.38** 3 3
Royal Dutch 8 21.76*** 15.90 1.15 52.47 17.33** 3 5
TPG 2 53.19*** 3.75 50.54 55.84 53.19*** 2
Unilever 2 10.33 9.39 3.69 16.97 10.33 1 1
Table 5. Results over all model specifications. This Table presents overpricing of reverse
convertibles that were listed on Euronext Amsterdam from January 1, 1999 to December 31, 2002 for
which also long-term call options were outstanding. The overpricing for each reverse convertible is
calculated as the average of the first five trading days. Each panel presents the result for a different
model specification. Column (1) gives a size of a sample for each group. Column (2) gives the value of
overpricing averaged over reverse convertibles. If the value is positive it means that the reverse convertible
is overpriced. The next column, column (3) gives the standard deviation of the overpricing. Columns
(4) and (5) report minimum and maximum value in the sample considered. Column (6) presents the
median value of overpricing. Finally, column (7) reports a number of reverse convertibles that were
overpriced. Significance of the average and median are calculated using a t-test. *** indicates the
significance at 1%, ** the significance at 5% and * the significance at 10%.
Size Avg St. dev Min Max Median Number of
Overpricing in % Overpriced
in %
(1) (2) (3) (4) (5) (6) (7)
Panel A: Black-Scholes model
Plain Vanilla 32 28.75*** 14.80 12.34 57.36 22.47*** 32
Knock in 43 18.78*** 13.73 -1.79 49.87 15.44*** 42
Whole sample 75 23.03*** 14.95 -1.79 57.36 19.06*** 74
Panel B: Square Root version of CEV model
Plain Vanilla 32 28.24*** 14.57 12.36 56.43 22.57*** 32
Knock in 43 - - - - - 42
Whole sample 32 28.24*** 14.57 12.36 56.43 22.57*** 32

28
Table 6. Regression results. The Table presents the regression analysis of the overpricing of the
reverse convertibles that were listed on Euronext Amsterdam from January 1, 1999 to December 31,
2002 for which also long-term call options were outstanding. The overpricing is calculated with the
Black-Scholes model and as the average of the first five trading days of each reverse convertible. Panel
A reports the results for the whole sample and panels B and C for the plain vanilla and knock-in
subsamples respectively. Conversion ratio is defined as the number of shares that might be delivered at
maturity. Barrier level is expressed in % of the stock price. The reported standard errors are
Heteroscedasticity consistent (White). Significance of the average and median are calculated using a t-
test. *** indicates the significance at 1%, ** the significance at 5% and * the significance at 10%.
Trading
Price Convertion Issuer Issuer Issuer Volume Dummy Barrier
Intercept of RC Ratio ABN ING RABO of RC PV level R-sq
Panel A: Whole sample (69 obs)
Coef -0.10 0.22 0.00
Std. Error 82.30 0.82
t-Statistic 0.00 0.27
Coef 9.34*** 0.17*** 0.55
Std. Error 1.90 0.02
t-Statistic 4.93 6.86
Coef 19.55*** 5.30 3.21 4.05 0.02
Std. Error 3.86 5.32 4.21 5.32
t-Statistic 5.06 1.00 0.76 0.76
Coef 39.08*** -1.55** 0.08
Std. Error 7.75 0.70
t-Statistic 5.04 -2.22
Coef 18.78*** 9.97*** 0.11
Std. Error 2.10 3.35
t-Statistic 8.95 2.98
Panel B: Plain Vanilla RC (30 obs)
Coef -21.77 0.50 0.01
Std. Error 87.19 0.86
t-Statistic -0.25 0.58
Coef 12.39*** 0.19*** 0.65
Std. Error 1.79 0.02
t-Statistic 6.92 9.37
Coef 55.84*** -26.83*** -31.35*** -20.57** 0.16
Std. Error 0.00 4.49 1.94 9.53
t-Statistic 71309037 -5.97 -16.16 -2.16
Coef 52.96*** -2.26*** 0.24
Std. Error 7.26 0.74
t-Statistic 7.29 -3.06

29
Table 6 ctd.
Trading
Price Convertion Issuer Issuer Issuer Volume Dummy Barrier
Intercept of RC Ratio ABN ING RABO of RC PV level R-sq
Panel C: Knock-in RC (38 obs)
Coef 15.64 0.01 0.00
Std. Error 103.80 1.03
t-Statistic 0.15 0.01
Coef 6.82*** 0.15*** 0.59
Std. Error 2.03 0.02
t-Statistic 3.36 5.95
Coef 16.76*** -3.01 0.78 5.10 0.05
Std. Error 2.96 5.22 3.60 4.90
t-Statistic 5.65 -0.58 0.22 1.04
Coef 20.37** -0.30 0.00
Std. Error 9.73 0.88
t-Statistic 2.09 -0.34
Coef 5.48 16.61** 0.03
Std. Error 7.33 8.00
t-Statistic 0.75 2.07

30
Figure 1. Average Overpricing. The Figure presents the average overpricing of the reverse
convertibles that were listed on Euronext Amsterdam from January 1, 1999 to December 31, 2002 for
which also long-term call options were outstanding. The overpricing is calculated with the Black-
Scholes. The average is calclulated at each day between the issuance and maturity of all priced bonds.

Average overpricing of the Dutch reverse

convertibles

40
overpricing in %

20

-20

-40
0 365 730
time after the issue (in days)

31