Hedging costs and joint determinants of premiums and spreads in structured financial products

• Author: Georg Fischer,Oliver Entrop
• Published date: 01 July 2020
• Date: 01 July 2020
• DOI: http://doi.org/10.1002/fut.22109

J Futures Markets. 2020;40:1049–1071. wileyonlinelibrary.com/journal/fut

|

1049

Received: 21 October 2018

|

Accepted: 8 February 2020

DOI: 10.1002/fut.22109

RESEARCH ARTICLE

Hedging costs and joint determinants of premiums and

spreads in structured financial products

Oliver Entrop |Georg Fischer

Chair of Finance and Banking, University

of Passau, Passau, Germany

Correspondence

Oliver Entrop, Chair of Finance and

Banking, University of Passau, Innstr. 27,

94032 Passau, Germany.

Email: oliver.entrop@uni-passau.de

Abstract

Evaluating more than 317,000 discount certificates in the German secondary

market, we find that premiums and spreads are endogenous and negatively

related but depend on different key determinants. The fundamental determi-

nants of the premiums are mainly profit‐related, that is, dividends of the un-

derlying, issuers’credit risk, lifecycle effect, and competition, whereas hedging

costs are less important. However, initial hedging costs (IHC) are priced into

the premium in the case of large inventory changes. The spread is mostly

determined by hedging costs and risk components, such as IHCs, rebalancing

costs, volatility, scalper risk, and overnight gap risk—but also by dividends.

KEYWORDS

derivatives, discount certificates, hedging, market microstructure, pricing, trading costs

JEL CLASSIFICATION

D40; G12; G21

1|INTRODUCTION

A prominent and well‐studied phenomenon of listed retail (investment) structured products—also commonly referred

to as “(investment) certificates”—is their overpricing in the secondary market, that is both bid and ask quotes set by the

issuers (as well as traded prices) exceed their mathematically “fair”theoretical values.

1

This is reflected in a positive

premium defined as the relative difference between mid‐quote and “fair”value.

2

A variety of studies described below

have analyzed the determinants of this premium and acknowledge that the reported premiums do not represent issuers'

net earnings, as premiums have to cover issuers' costs. However, these studies do not explicitly incorporate the issuers'

cost side in their empirical analyses. Furthermore, the literature ignores the bid/ask spread, its potential determinants

and the potential link between premium and spread.

This paper aims at deepening our understanding of issuers' price‐setting behavior in the market for retail structured

products by closing these gaps. To the best of our knowledge we are the first to incorporate the issuers' cost side into the

-------------------------------------------------------------------------------------------------------------------------------------------

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any

medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.

© 2020 The Authors. The Journal of Futures Markets published by Wiley Periodicals, Inc.

1

Theoverpricing can be foundfor a variety of certificates andlocal markets, see,for example, the Germancertificatemarket (Baule, 2011;Baule&Tallau,2011; Bauleet al., 2008; Entrop, Fischer,McKenzie,

Wilkens,& Winkler, 2016;S chertler, 2016; Stoimenov and Wilkens, 2005; Wilkens et al., 2003), structuredequity productsin the United States (Benet,Giannetti & Pissaris,2006; Henderson & P earson, 2011),

the Netherlands(Szymanowska et al.,20 09), and Switzerland(Burth, Kraus & Wohlwend,2001;Wallmeier&Diethelm,2009).

2

The premium is also commonly referred to as “margin”or “mark‐up.”

empirical analysis of determinants of premiums and bid/ask spreads and also the first to consider the spread at all.

Briefly summarizing our key results before going into detail, we find that premiums and spreads are endogenous and

negatively related but depend on different key determinants. While hedging costs are generally of minor importance for

the premium (compared to standard determinants known from prior literature), initial hedging costs (IHC) are priced

into the premium only in the case of large inventory changes. The key determinants of the spread are hedging cost‐and

risk‐related—but dividends of the underlying are also relevant.

As already mentioned, several studies have analyzed the determinants of the premium, whose key results can be

summarized as follows: The premium diminishes over a certificate's lifetime; this is commonly referred to as the “life

cycle hypothesis”(e.g., Entrop, Scholz, & Wilkens, 2009; Stoimenov & Wilkens, 2005; Wilkens & Stoimenov, 2007;

Wilkens, Erner, & Röder, 2003). Issuers adjust the premium in accordance with the order flow they expect, that is they

increase (decrease) the premium in phases of positive (negative) net expected investors' buying pressure (“order flow

hypothesis”; Baule, 2011; Wilkens et al., 2003). Furthermore, the premium decreases with higher competition between

issuers (e.g., Baule, 2011; Entrop, Fischer, et al., 2016; Schertler, 2016), and increases with a higher issuer's credit risk

(e.g., Baule, Entrop, & Wilkens, 2008; Entrop, Fischer, et al., 2016; Schertler, 2016), higher dividend yields of the

underlying (e.g., Entrop, Fischer, et al., 2016), a higher volatility of the underlying (e.g., Entrop, Fischer, et al., 2016;

Szymanowska, Horst, & Veld, 2009) and higher unhedgeable risk (e.g., Baller, Entrop, McKenzie, & Wilkens, 2016).

We already noted that this literature ignores the issuers' costs side and does not consider the spread at all. While the

bid/ask spread usually serves as a market maker's compensation for costs (e.g., transaction costs, hedging costs) and

risks (e.g., informed traders, illiquidity) in standard markets such as the equity, bond, or options markets, it can

additionally serve as a further source of profit for the issuer in the market for structured products. Given its market

design, described in Section 2 in more detail, the theoretical model by Baller et al. (2016) shows that there should be a

substitution effect between premium and bid/ask spread.

3

This implies that each determinant of the premium and the

bid/ask spread, respectively, is a potential determinant of the other one.

Our analysis builds on a large quote and trade data set with 396,249 discount certificates on DAX stocks that were

tradable on the Euwax between January 2006 and December 2013. From our tick data we form five time bars each day to

keep the analysis numerically manageable, which still results in more than 80 million observations while most studies only

observe daily quotes or quotes at issuance. For calculating the premium we use a structural model by Baule et al. (2008)that

relaxes the standard Hull and White (1995) assumption of independence between the market risk of the underlying of

discount certificate and issuer's credit risk. The relative bid/ask spread can be observed directly from the quotes.

4

Our econometric design is a 2‐equation system where premium and spread are the dependent variables and the

potential determinants are the independent variables. We also add the spread as an explanatory variable to the

premium‐equation and vice versa. To account for the resulting endogeneity, we use GMM‐2SLS as well as GMM‐3SLS

approaches to estimate our equations.

We group potential determinants into three categories. The first one is hedging costs. We borrow from the literature

on options (see e.g., Boyle & Vorst, 1992; Huh, Lin, & Mello, 2015; Leland, 1985; Wu, Liu, Lee, & Fok, 2014) and

warrants (see e.g., Petrella, 2006; Petrella & Segara, 2013), and split hedging costs into initial hedging and rebalancing

costs.

5

IHC are the costs associated with setting up and liquidating a delta‐neutral position. Rebalancing costs (RC)

represent the costs of rebalancing the position to keep it delta‐neutral throughout the certificate's lifetime.

As delta‐hedging strategies are usually carried out in discrete time for reasons of transaction costs, they are not

perfect and issuers have to bear the remaining risk. Additionally, there is risk such as jump risk that cannot be hedged

or only with high costs. These remaining risks also affect issuers’cost side as they result in opportunity costs. Therefore,

our second category is risks. Here we subsume overnight gap risk and downside jump risk. We also add volatility as a

broad measure for uncertainty. Additionally, we consider scalper risk, which is the risk of informed trading against the

market maker. Our third category, other variables, captures the commonly analyzed determinants from the structured

product literature, that is issuer's credit risk, order flow, competition, lifecycle effect and dividends. We also add perfect

hedge opportunities via the option market.

3

In contrast, Baller et al. (2016) find a positive influence of the spread on the premium in their empirical analysis. They consider highly speculative short‐term knock‐out products, however, and do not

consider spread determinants.

4

For simplicity, we use the term “spread”in the following instead of “relative bid/ask spread,”if not stated otherwise.

5

Issuers' price‐setting behavior for warrants also exhibits overpricing, that is quoted prices above the theoretical fair value or option quotes (e.g., Bartram et al., 2008; Baule and Blonski, 2015; Horst

and Veld, 2008; Li & Zhang, 2011).

1050

|

ENTROP AND FISCHER