Which private investors are willing to pay for sustainable investments? Empirical evidence from stated choice experiments

Journal of Banking and Finance 102 (2019) 193–214

Contents lists available at ScienceDirect

Journal of Banking and Finance journal homepage: www.elsevier.com/locate/jbf

Which private investors are willing to pay for sustainable investments? Empirical evidence from stated choice experiments Gunnar Gutsche, Andreas Ziegler∗ Institute of Economics, University of Kassel, Nora-Platiel-Str. 5, 34109 Kassel, Germany

a r t i c l e

i n f o

Article history: Received 6 January 2017 Accepted 9 March 2019 Available online 12 March 2019 JEL Classification: G11 Q56 M14 G02 A13 C25 Keywords: Sustainable investments Stated choice experiments Mixed logit models Latent class logit models Willingness to pay Psychological motives, values, and norms

a b s t r a c t Based on data from a representative survey among German private financial decision makers that comprised two stated choice experiments for fixed-interest investment products and equity funds, this paper empirically examines whether and which investors are willing to pay for sustainable investments. In fact, our econometric analyses with mixed and latent class logit models reveal strong stated preferences and a considerable willingness to pay (WTP) for sustainable investment products. Furthermore, our mixed logit model analysis implies that the mean WTP for certified sustainable investment products is strongly higher than the mean WTP for the uncertified counterparts. In addition, our estimation results suggest that investors with high feelings of warm glow from sustainable investments, an affinity to left-wing parties, and a strong environmental awareness have a clearly higher mean willingness to sacrifice returns for sustainable investment products than their counterparts. While risk perceptions seem to be additionally relevant for certified sustainable equity funds, they obviously play a negligible role for less risky sustainable fixed-interest investment products. © 2019 Elsevier B.V. All rights reserved.

1. Introduction Sustainable investments taking into account ecological, social, and/or ethical concerns play an increasing role on financial markets, for example, in the USA (e.g. US SIF, 2018) and Europe (e.g. Eurosif, 2018). The principle of these investments can be based on several approaches such as negative screens, whereby certain investments are avoided like in so-called sin stocks (especially alcohol, tobacco, weapons, gambling) (e.g. Barreda-Tarrazona et al., 2011), or best-in-class screens, whereby sustainability leaders from each sector are identified like in the Dow Jones Sustainability Indexes (e.g. Oberndorfer et al., 2013). According to traditional finance theory, sustainable investments would only be considered if they are at least as attractive as other investments in terms of risk and returns (e.g. Bauer and Smeets, 2015). While some studies (e.g. Derwall et al., 2005; Kempf and Osthoff, 2007; Edmans, 2011; Eccles et al., 2014) in fact show that ∗

Corresponding author. E-mail addresses: [email protected] (G. Gutsche), andreas.ziegler @uni-kassel.de (A. Ziegler). https://doi.org/10.1016/j.jbankfin.2019.03.007 0378-4266/© 2019 Elsevier B.V. All rights reserved.

sustainable investments are worthwhile, other studies find either that there is a financial price to be paid for these investments (e.g. Belghitar et al., 2014) or higher abnormal returns for sin stocks (e.g. Hong and Kacperczyk, 2009; Derwall et al., 2011) suggesting the relevance of non-financial motives for sustainable investments. This paper empirically examines whether private investors are willing to pay (i.e. willing to sacrifice returns) for sustainable investments, i.e. whether not only pecuniary motives, but also nonpecuniary motives play a role for these investments. By considering the relevance of individual characteristics and especially different psychological motives, values, and norms, we particularly analyze which investor groups are willing to pay for sustainable investment products. Our empirical analysis is based on (online) representative data for private financial decision makers in Germany. The representativeness of our sample (in terms of age, gender, and place of origin with respect to the whole German population) is ensured by recruiting respondents from an online panel of a German market research institute. More specifically, our econometric analysis is based on data from experiments on the stated choice (SC) among several fixed-interest investment products as well as equity funds

194

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

that are characterized by different attributes including uncertified and certified sustainability criteria. In contrast to our econometric approach with SC data, several previous empirical analyses are based on real investment and thus revealed preferences data. The main advantage of such data is that they are based on actual decisions and choices in real-world situations, i.e. market participants reveal their preferences through their real investment decisions. One direction of such studies examining the relevance of non-pecuniary motives refers to the socalled shunned-stock hypothesis. In the case of sustainable investments, this hypothesis suggests that unsustainable stocks such as sin stocks have higher expected returns since a group of investors (which must not be too small, e.g. Heinkel et al., 2001) shun these stocks due to personal values and/or social norms (e.g. Borgers et al., 2015). Therefore, the significantly positive abnormal returns for unsustainable stocks in, for example, Fabozzi et al. (2008), Hong and Kacperczyk (2009), or Derwall et al. (2011) support the validity of the shunned-stock hypothesis and thus the relevance of non-financial motives for sustainable investments. Another important direction of studies with revealed investment data refers to fund flows. For example, Renneboog et al. (2011) show that money flows of sustainable equity funds around the world are less related to past returns than conventional equity fund flows. Equity fund flow studies of Bollen (2007), Benson and Humphrey (2008), and Hartzmark and Sussman (2018) also suggest that sustainable investors in the USA are less concerned about returns than conventional investors so that non-financial motives obviously matter. This result is supported in Henke (2015), who analyzes US bond fund flows. He reports a considerable estimated willingness to pay (WTP) between 0.4 and 1.2 percentage points of the returns for sustainable fund attributes. While these studies examine real market and thus revealed preferences data, they do not directly analyze investments at the individual level and thus are hardly able to differentiate between several motives of institutional and private sustainable investors. An identification of specific single non-pecuniary motives is therefore not possible with such indirect approaches so that another strand of the literature directly examines the relevance of financial and non-financial motives for sustainable investments on the basis of individual level data. Most of these econometric studies (e.g. Nilsson, 2008, for Sweden; Bauer and Smeets, 2015, for the Netherlands; Wins and Zwergel, 2016, and Gutsche et al., 2019,1 for Germany) use survey data from investors and consider perceived risk and returns of sustainable investments besides several indicators for psychological motives, values, and norms. While survey data about actual investments are aimed at revealing real decisions and thus real market behavior, they have some well-known limitations (e.g. Graham and Harvey, 2001; Guiso et al., 2013; Riedl and Smeets, 2017). In particular, the respondents in surveys often have difficulties to report exact amounts of their investments in different funds, stocks, and bonds or alternatively might also overstate or overestimate their own sustainable investments. Furthermore, response biases are possible due to different framings of survey questions. Therefore, the use of administrative revealed individual investor data is certainly very attractive since it allows the analysis of real individual investment decisions. For example, Riedl and Smeets (2017) combine administrative data from a Dutch mutual fund provider with survey and (incentivized) experimental data for the corresponding investors. In line with the aforementioned studies, they also show that values and norms are more relevant 1 While the accompanying study of Gutsche et al. (2019) uses data from the survey of German investors that is also the basis of this paper, the main difference is that only the corresponding survey data about actual investments are considered in the previous study, but not the SC data, which are focused in this paper.

for sustainable investments than perceived risk and returns. In order to identify non-pecuniary motives for sustainable investments, Hartzmark and Sussman (2018) complement their fund flows analysis with an econometric analysis on the basis of survey and experimental data that were collected from students in Chicago and MTurk participants. In these experiments, the participants were asked to hypothetically allocate 10 0 0 US dollars between hypothetical funds with different sustainability ratings of the financial information service provider Morningstar and a savings account. However, in contrast to their analysis of fund flows, these hypothetical investments represent stated preferences instead of real decisions. Stated preferences data are generally collected in experiments or surveys and refer to situations where a decision or choice is made by considering hypothetical scenarios (e.g. Hensher et al., 2010). A general advantage of stated compared to revealed preferences data is that they are not limited to situations, products, and attributes of alternatives that currently exist or have existed in the past, but can also refer to new (e.g. investment) products, new attributes of (investment) products, or (investment) products with a low market penetration. In the example of Hartzmark and Sussman (2018), unavailable and thus hypothetical equity funds, which differ with respect to past performances, fund characteristics, and especially Morningstar sustainability ratings, are examined. In our empirical analysis, we also examine stated preferences data on hypothetical uncertified or certified sustainable investment products with transparency logos from different providers, which are not available on the capital market so far. Some further studies on the basis of stated preferences data, which directly asked for the WTP or willingness to sacrifice returns for sustainable investment products can, for example, be found in Dorfleitner and Utz (2014) or Borgers and Pownall (2014), who examine German and Dutch investors, respectively. In order to make hypothetical scenarios more realistic, Webley et al. (2001) use data from an experimental survey of 56 British investors, whereby the participants had to make decisions about their real investment portfolio in several scenarios that vary in the future financial performance of sustainable and conventional investments. Furthermore, Pasewark and Riley (2010) asked students to make a hypothetical investment of 10,0 0 0 US dollars either in a sustainable bond (i.e. a non-tobacco firm) or in an unsustainable bond (i.e. a tobacco firm) which differ in the fixed interest rates. Another recent stated preferences experiment can be found in Berry and Yeung (2013). They analyze data from a small sample of British ethical investors, where the participants had to choose one of five selected percentages between 0% and 15% from a hypothetical portfolio of 10 0,0 0 0 British pounds for ten investment opportunities with different levels of financial performance and sustainability. While some of these stated preferences experiments include choices between different investment products, none of them are common SC experiments (e.g. Hensher et al., 2010; Louviere et al., 2010). SC experiments are specific stated preferences methods, where respondents are asked to indicate their preference among two or more multi-attribute alternatives (e.g. investment products). Hensher et al. (2010) extensively discuss several problems of other stated preferences methods (e.g. directly asking for the WTP for some product attributes like sustainability criteria such as in Dorfleitner and Utz, 2014, or Borgers and Pownall, 2014), which provide at best only information on preferences. While almost no previous study has used this approach for the analysis of sustainable investment products,2 it is widespread in other economic sub-disciplines such as transport, energy, or environmental

2 One recent exception is the study of Nakai et al. (2018), which examines 46 Japanese undergraduate students.

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

195

economics to examine, for example, the WTP for fuel availability in the choice among vehicles (e.g. Achtnicht et al., 2012), for renewable and nuclear energy in the choice among electricity mixes (e.g. Murakami et al., 2015), or for reducing the risks of microbial and cancer diseases in the choice among public drinking water programs (e.g. Adamowicz et al., 2011). However, an important problem of empirical analyses with stated preferences data including data from SC experiments is their hypothetical character, which can lead to hypothetical biases and thus a restricted external validity of the estimation results. As discussed above, the avoidance of hypothetical biases is the main advantage of revealed preferences data. However, reliable analyses with a high external validity require real market data. In our empirical analysis, we do not only consider uncertified and certified sustainable investment products with transparency logos from different providers as aforementioned, but especially focus on sustainable fixed-interest investment products, which were not available in the past and are currently still scarce on the capital market (e.g. Henke, 2015). Future field experiments with financial providers would therefore certainly be attractive in this respect. However, it can be argued that they are also restricted if, for example, the considered sample is not representative for the whole capital market, as discussed in Hartzmark and Sussman (2018). Against this background, our empirical analysis is based on data from two SC experiments for fixed-interest investment products and equity funds. Our econometric analysis implies strong stated preferences and a considerable WTP for sustainable fixed-interest investment products and equity funds. This result is robust after several techniques to mitigate possible hypothetical biases. Furthermore, the estimated mean WTP for certified sustainable investment products is strongly higher (e.g. more than twice as high in the case of fixed-interest investment products) than the estimated mean WTP for the uncertified counterparts. In addition, our estimation results suggest that investors with high feelings of warm glow from sustainable investments, an affinity to left-wing parties, and a strong environmental awareness have a clearly higher mean WTP for sustainable investment products. While risk perceptions seem to be additionally relevant for certified sustainable equity funds, they obviously play a negligible role for sustainable fixedinterest investment products. Therefore, our SC study complements and contributes to the previous literature in several directions. It especially contributes to the literature on pecuniary and non-pecuniary motives for sustainable investments. For example, our empirical analysis does not only examine equity funds, but (in line with Henke, 2015) especially fixed-interest investment products, which allow a cleaner setting for the estimation of WTP compared to the consideration of past return profiles of equity funds. Furthermore, our SC experiments also consider sustainability certificates. Therefore, our empirical analysis is able to provide an even broader picture of the stated preferences for several variants of sustainable investment products. In contrast to many previous studies, our empirical analysis is based on data from a representative sample of (private) financial decision makers. According to Hartzmark and Sussman (2018), exclusively considering sustainable investments or investors provide results that are not representative for all investments or investors market-wide. In addition, we argue that also empirical analyses with data from a specific population need not necessarily lead to the identification of representative preferences of all investors, either.3

Another main contribution of our study refers to the literature on psychological motives, values, and norms in the individual investment behavior. The combination of our SC data with such variables allows us to estimate the WTP for sustainable investment products for different investor groups (i.e. comparative static effects of individual characteristics) and thus to complement and expand previous studies (e.g. Nilsson, 2008; Bauer and Smeets, 2015; Wins and Zwergel, 2016; Riedl and Smeets, 2017; Gutsche et al., 2019). Our econometric analysis is not only based on the application of mixed logit models, but also of latent class logit models (e.g. Greene and Hensher, 2003), whereby investors are probabilistically assigned to different classes, which depend on several individual characteristics. As a consequence, the WTP for sustainable investment products can be estimated for different investor classes. The remainder of the paper is structured as follows: Section 2 describes the data from the survey including the two SC experiments. Section 3 discusses the estimation results on the basis of our mixed and latent class logit model analyses. Finally, Section 4 concludes.

3 For example, Riedl and Smeets (2017) examine customers of one of the largest mutual fund providers in the Netherlands, as discussed above, Wins and Zwergel (2016) use data from a survey that was distributed via several German investment fund-related fora, websites, and foundations, as well as among other researchers, and Nilsson (2008) generated a sample of Swedish investors from an ex-

isting customer database of a European mutual fund provider that offers SRI profiled mutual funds. 4 The respondents were asked to assume that all investment product alternatives are completely identical besides these four attributes (e.g. in terms of investment type or deposit guarantees).

2. Data and variables 2.1. Stated choice experiments Our empirical analysis is based on data from a computer-based survey that was carried out in cooperation with the German market research institute GfK SE, which drew an online representative sample (in terms of age, gender, and place of origin) from its internal online panel during December 2013 and January 2014. The population of the survey consists of financial decision makers in Germany, who are defined as persons who are at least 18 years old and mainly or equally responsible for financial decisions in the household. To ensure that the respondents have a minimum of investment experiences, we further required the interviewees to have at least a savings account. Overall, 1173 respondents participated in the survey. Using a quality saving system that was provided by GfK SE, 172 respondents were excluded from the original sample due to qualitatively insufficient response behavior (e.g. in terms of duration of their responses). Therefore, 1001 respondents are the basis of our empirical analysis. The survey comprised several parts. One part referred to general investment decisions, specifically to sustainable investments, to other pro-environmental and pro-social attitudes and behaviors, to several values and norms, as well as to socio-demographic variables. The main part referred to two SC experiments which comprised choices among several investment products. The first SC experiment referred to fixed-interest investment products. It started with a detailed description of the choice situation (see Appendix A). The 1001 participants were asked to choose among four alternative fixed-interest investment products with an investment horizon of three years. The SC experiment was based on six choice sets, i.e. each respondent made six decisions. The participants were informed that some of the displayed investment products are currently not provided by banks, but they were asked to imagine that these products can in fact be purchased. The four fixed-interest investment product alternatives were described by four attributes, respectively:4 • Provider • Annual nominal interest rate

196

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

Table 1 Attributes and attribute levels in the SC experiments. First SC experiment on three-year fixed-interest investment products Attributes

Attribute levels

Provider Annual nominal interest rate Sustainability criteria Transparency logo

Big bank, municipal savings bank, co-operative bank, direct bank, sustainability bank 1.30%, 1.50%, 1.70%, 1.90%, 2.10% No consideration, consideration without sustainability certificate, consideration with sustainability certificate No transparency logo, transparency logo issued by an NGO, transparency logo issued by the state

Second SC experiment on equity funds Attributes

Attribute levels

Value of the subscription fee Net return in the last year Average annual net return in the last five years Sustainability criteria Transparency logo

3.0 0%, 4.0 0%, 5.0 0% 4.0 0%, 5.0 0%, 6.0 0%, 7.0 0%, 8.0 0% 3.0 0%, 5.0 0%, 6.0 0%, 7.0 0%, 9.0 0% No consideration, consideration without sustainability certificate, consideration with sustainability certificate No transparency logo, transparency logo issued by an NGO, transparency logo issued by the state

Table 2 Exemplary choice sets in the SC experiments. First SC experiment on three-year fixed-interest investment products Please indicate which of the following four investment products appears so attractive for you that you would most likely purchase it. Attribute

Three-year fixed-interest investment product A

Three-year fixed-interest investment product B

Three-year fixed-interest investment product C

Provider Annual nominal interest rate Sustainability criteria

Direct bank 1.30% No consideration

Transparency logo

No transparency logo

Direct bank 1.70% Consideration with sustainability certificate Transparency logo issued by an NGO 

Big bank 1.50% Consideration without sustainability certificate Transparency logo issued by the state 



Three-year fixed-interest investment product D Municipal savings bank 1.30% No consideration No transparency logo 

Second SC experiment on equity funds Please indicate which of the following four equity funds appears so attractive for you that you would most likely purchase it. Attribute

Equity fund A

Equity fund B

Equity fund C

Equity fund D

Value of the subscription fee Net return in the last year Average annual net return in the last five years Sustainability criteria

3.00% 4.00% 3.00%

5.00% 8.00% 7.00%

5.00% 7.00% 5.00%

4.00% 5.00% 9.00%

Consideration without sustainability certificate Transparency logo issued by the state 

No consideration

Consideration with sustainability certificate Transparency logo issued by an NGO 

No consideration

Transparency logo

No transparency logo 

• Consideration of sustainability criteria • Transparency logo The upper part of Table 1 summarizes these attributes and the corresponding attribute levels in the SC experiment. It reveals that five different providers are included, namely big banks, municipal savings banks, co-operative banks, direct banks, and sustainability banks. With respect to annual nominal interest rates as the only pecuniary attribute, five levels between 1.30% and 2.10% are considered. On the basis of these two attributes, it is, for example, possible to estimate the mean WTP for sustainability banks. However, our main interest refers to the WTP for the consideration of sustainability criteria in the investment products. In the description of the SC experiment, we explained that our concept of sustainable investments refers to investments which generally take ecological, social, and/or ethical criteria into account. We consider three levels for this attribute, namely “no consideration”, “consideration without sustainability certificate”, and “consideration with sustainability certificate”. A certificate means that the consideration of sustainability criteria was tested and confirmed by an independent organization. Finally, transparency logos are considered as fourth attribute with three levels, namely “no transparency logo”, “transparency logo issued by an NGO”, and “transparency logo issued by the state”. A transparency logo means that the investment products

Transparency logo issued by an NGO 

publicly provide detailed information about the investment strategy. The upper part of Table 2 reports an exemplary choice set and Fig. 1 presents the corresponding original screenshot (in German) for this SC experiment. The second SC experiment referred to the choice among four different equity funds. Since a reliable assessment of several equity funds requires a certain amount of knowledge and experience of this type of investment products, not all respondents were allowed to participate in this SC experiment. Only participants who indicated that they have already invested in or are sufficiently informed about equity funds, stocks, or other complex investment products (e.g. bond funds, mixed funds, open property funds) were included. Thus, only 801 of the overall 1001 respondents took part in this second SC experiment. Again, it started with a detailed description of the choice situation (see Appendix B) and was based on eight choice sets. Once more, the participants were informed that some of the displayed funds are currently not provided on the capital market, but they were asked to imagine that these funds can in fact be purchased. The four equity funds alternatives were described by five attributes, respectively:5

5 Once more, the respondents were asked to assume that all equity fund alternatives are completely identical besides these five attributes.

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

197

Fig. 1. Original screenshot of an exemplary choice set in the first SC experiment on three-year fixed-interest investment products.

• • • • •

Value of the subscription fee Net return in the last year Average annual net return in the last five years Sustainability criteria Transparency logo

The lower part of Table 1 summarizes these attributes and the corresponding attribute levels in the SC experiment. In contrast to the first SC experiment, this second SC experiment thus comprised three pecuniary attributes. While the values of the subscription fee are fixed amounts that vary between 3.0 0%, 4.0 0%, and 5.0 0%, the five values of the net returns in the last year between 4.00% and 8.00% and the five values of the average annual net return in the last five years between 3.00% and 9.00% are past returns. Since it is not clear whether these returns persist in the future, the underlying assumption in this SC experiment is that the past returns reflect the average returns of the corresponding equity funds and that the choice of the respondents is based on the perception and belief of these returns. Due to the uncertainty that this assumption is valid, the inclusion of the latter two pecuniary variables as basis for the estimation of WTP should be treated with more caution. In contrast, the consideration of fixed interest rates in the first SC experiment provides a cleaner setting in this respect so that especially the data from this SC experiment are considered in this paper.6 The two remaining non-financial attributes in the second SC experiment, i.e. sustainability criteria and transparency logo, are completely identical to the first SC experiment and thus comprise the same three levels, respectively. The lower part of Table 2 reports an exemplary choice set for this second SC experiment. The exemplary choice sets in Table 2 as well as the screenshot in Fig. 1 reveal that both SC experiments are unlabeled as 6 As a consequence, the latent class logit model analysis only refers to the choice among the fixed-interest investment products, but not to the choice among the equity funds, as discussed below.

it is common in many empirical studies (e.g. Goett et al., 20 0 0; Adamowicz et al., 2011; Murakami et al., 2015). The consideration of the four attributes in the first SC experiment and the five attributes in the second SC experiment was based on consultations with practitioners on the capital market before the survey, which increases the practical relevance of the SC experiments and thus the validity of the empirical results.7 This also includes the choice and description of the attribute levels, especially with respect to the consideration of sustainability criteria.8 Furthermore, the ranges of the attribute levels for the annual nominal interest rates in the first SC experiment and for the three pecuniary attributes in the second SC experiment were based on usual values on the capital market during the time of the survey at the end of 2013. The experimental design was generated by GfK SE with the Sawtooth Software. In order to keep both the statistical efficiency as well as the precision of estimated interaction terms at an acceptable level, a “Balanced Overlap” design approach was applied (e.g. Chrzan and Orme, 20 0 0). In total 50 different versions of randomized choice sets were created for each SC experiment and assigned to the respondents.

7 This means that while it would have been technically possible to only include the sustainability and financial attributes as basis for our empirical analysis, the additional inclusion of other attributes (here with respect to transparency logos and the provider) is common in SC experiments and increases the relation to reality and thus the reliability of the estimation results for the consideration of sustainability criteria. 8 In line with the consultations and due to our focus on uncertified or certified sustainable investment products, we can only consider a very general concept of sustainable investments. While a separate analysis of several subgroups of the very broad concept (e.g. based on negative or best-in-class screens, as discussed in the introduction) would also be very interesting, this would have required the inclusion of several additional attribute levels. However, more complex SC experiments can decrease the clarity for the respondents and also lead to a lower statistical efficiency in the experimental design, which decreases the reliability of the corresponding estimation results.

198

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

2.2. Variables in the econometric analysis The dependent variables in our econometric analysis refer to the choice among the four three-year fixed-interest investment products and among the four equity funds. The explanatory variables are based on the four attributes in the first SC experiment and the five attributes in the second SC experiment as discussed above. While the financial attributes are treated as continuous variables and thus can be directly included, the other attributes are discrete so that dummy variables are defined for each level. In the case of our variable of main interest, i.e. the consideration of sustainability criteria, the dummy variable for the level “no consideration of sustainability criteria” is used as base category. While one model specification includes the remaining two dummy variables “consideration of sustainability criteria without certificate” and “consideration of sustainability criteria with certificate”, another specification only includes the dummy variable “consideration of sustainability criteria”, which summarizes the previous two dummy variables. Similarly, the dummy variable for the level “no transparency logo” is used as base category. While one model specification includes the remaining two dummy variables “transparency logo issued by an NGO” and “transparency logo issued by the state”, the other specification only includes the dummy variable “transparency logo”, which summarizes the previous two dummy variables. With respect to the providers in the first SC experiment, the dummy variable “big bank” is used as base category so that the dummy variables “municipal savings bank”, “cooperative bank”, “direct bank”, and “sustainability bank” are included in the econometric analysis. While the inclusion of these attributes is sufficient to estimate mean WTP for uncertified and/or certified sustainable investment products, we especially focus on the effect of individual characteristics on the stated preferences for sustainable investments. Pecuniary motives for the valuation of sustainable investments can refer to higher expected returns or to lower expected risk (e.g. Nilsson, 2008; Bauer and Smeets, 2015; Wins and Zwergel, 2016; Riedl and Smeets, 2017; Hartzmark and Sussman, 2018; Gutsche et al., 2019). However, expected returns of sustainable investments do not play a role in our empirical methodology since the SC experiments explicitly include returns (i.e. fixed interest rates in the first SC experiment and past returns in the second SC experiment) as attributes. Therefore, the econometric analysis of sustainable investment products directly controls for returns. In contrast, a possible WTP for sustainable investment products might be influenced by a lower expected risk. Therefore, the respondents were asked to assess the average risk level of sustainable investments compared to conventional investments on a symmetric scale with five ordered response categories, i.e. “much lower”, “rather lower”, “neither higher nor lower”, “rather higher”, or “much higher”, respectively.9 On this basis, we construct the dummy variable “high perceived risk” that takes the value one if a respondent indicated one of the two highest categories (i.e. “rather higher” or “much higher”). However, our empirical analysis focuses on specific nonpecuniary psychological motives, values, and norms. One motive that is often important for sustainable behavior is warm glow, which can be described as a good feeling through the act of giving (e.g. Andreoni, 1990). Such feelings can lead to psychological benefits and thus higher utility levels from sustainable investments. The dummy variable “warm glow” takes the value one if a respondent agreed rather strongly or totally to the statement “it makes me feel good to make sustainable investments” or to the state-

9 The exact wording of the questions and the corresponding response categories for all variables can be found in Appendix C.

ment “I feel responsible for a sustainable development and want to contribute by making sustainable investments”.10 In line with the results of Bauer and Smeets (2015) and Gutsche et al. (2019), we expect that feelings of warm glow lead to a higher WTP for sustainable investment products. Preferences and the WTP for sustainable investment products can also be affected by social norms or social pressure. In order to avoid social sanctions, individuals often adjust their behavior by complying with the norms of the social environment (e.g. Akerlof and Kranton, 20 0 0; Nyborg and Rege, 2003). Therefore, we expect that such social norms lead to a higher WTP for sustainable investment products. We consider the dummy variable “expectation social environment” that takes the value one if a respondent agreed rather strongly or totally with the statement “my social environment (e.g. family, friends, colleagues) expects me to make sustainable investments”.11 In addition to these two variables, we examine indicators for environmental awareness and political identification. With respect to the former indicator, the dummy variable “membership environmental organization” takes the value one if a respondent is member of a group or organization engaged in the conservation and protection of the environment and nature. Since ecological investments as a main component of sustainable investments are one dimension of pro-environmental behavior, we expect that environmental values do not only affect, for example, behavior like the payment of a price premium for electricity to finance a wind turbine or the consumption of less electricity (e.g. Kotchen and Moore, 2008), but also lead to a higher WTP for sustainable investment products. For the analysis of political orientation, the participants were asked with which political party they are most likely affiliated. In order to examine the relevance of a left-wing political identification, we consider the dummy variable “affinity left-wing parties” that takes the value one if a respondent is mainly affiliated with the Social Democrats (SPD), the Green Party (Bündnis 90 / Die Grünen), or the Left Party (Die Linke).12 In line with previous studies on the effect of a left-wing identification on pro-environmental behavior like climate protection activities (e.g. Schwirplies and Ziegler, 2016) and especially on sustainable investments or socially controversial investing (e.g. Hong and Kostovetsky, 2012; Hood et al., 2014), we expect a higher WTP for sustainable investment products in the population group with this orientation. We also examine the relevance of five socio-demographic variables. The dummy variable “female” takes the value one if a respondent is a woman, while “age” is the age of a participant in years. The dummy variable “high education” takes the value one if the highest level of education is at least an advanced technical college certificate or a high school graduation. The dummy variable “living together or married” takes the value one for these two marital statuses, while the dummy variable “Western Germany” takes the value one if a respondent lives in one of the West German federal states excluding Berlin. Table 3 reports the numbers of observations and some descriptive statistics for the perceived risk of sustainable investments, the psychological motives, values, and norms, as well as the socio-demographic variables.

10 In the survey the participants were asked how strongly they agree with the two statements on a symmetric scale with five ordered response categories, i.e. “totally disagree”, “rather weakly agree”, “neither strongly nor weakly agree”, “rather strongly agree”, and “very strongly agree”. 11 Once more, the participants had to choose among the five ordered response categories “totally disagree”, “rather weakly agree”, “neither strongly nor weakly agree”, “rather strongly agree”, and “very strongly agree”. 12 The questionnaire comprised four further dominating political parties in Germany during the time of the survey at the end of 2013 besides these three parties and “another party”, namely the Christian Democrats (CDU/CSU), the Liberals (FDP), the main right-wing party (AfD), and the Pirate Party (Piratenpartei).

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

199

Table 3 Descriptive statistics of individual characteristics. Number of observations

Mean

Standard deviation

Minimum

Maximum

782

0.35

0.48

0

1

936 955 942 778

0.46 0.10 0.10 0.49

0.50 0.30 0.30 0.50

0 0 0 0

1 1 1 1

1001 1001 997 995 1001

0.49 43.91 0.61 0.67 0.82

0.50 12.97 0.49 0.47 0.38

0 18 0 0 0

1 78 1 1 1

Perceived risk of sustainable investments High perceived risk Psychological motives, values, and norms Warm glow Expectation social environment Membership environmental organization Affinity left-wing parties Socio-demographic variables Female Age High education Living together or married Western Germany

3. Econometric analysis 3.1. Mixed logit model analysis 3.1.1. Econometric approach As discussed above, our empirical analysis is based on data from two SC experiments whereby a respondent i chose M = 6 times among J = 4 different fixed-interest investment products with an investment horizon of three years and M = 8 times among J = 4 different equity funds. In the first SC experiment, the alternatives were described by the four attributes provider, annual nominal interest rate, sustainability criteria, and transparency logo, respectively, while in the second SC experiment, the alternative equity funds were described by the five attributes value of the subscription fee, net return in the last year, average annual net return in the last five years, sustainability criteria, and transparency logo, respectively. A corresponding econometric analysis of the relevance of these attributes for the choice among four mutually exclusive alternatives must therefore be based on multinomial discrete choice models, which assume utility functions for each choice alternative. In our case, the utility of respondent i (i = 1,…, N) for investment product j (j = 1,…, 4) in choice set m (m = 1,…, 6 or m = 1,…, 8) is:

Ui jm = βi ’xi jm + εi jm The utility Uijm thus depends on the vector xijm = (xijm1 ,…, xijmK )’ of explanatory variables that are based on the attributes and individual characteristics as well as on the corresponding unknown parameter vectors β i = (β i1 ,…, β iK )’. As discussed above, one model specification includes the aggregated dummy variables for the consideration of sustainability criteria and a transparency logo besides the other attributes so that K = 7 in the case of the first SC experiment and K = 5 in the case of the second SC experiment.13 For the other model specification, K = 9 in the case of the first SC experiment and K = 7 explanatory variables in the case of the second SC experiment are considered due to the inclusion of the two disaggregated dummy variables for the consideration of sustainability criteria and a transparency logo instead of the aggregated variables. In order to analyze the relevance of perceived risk of sustainable investments and psychological motives, values, and norms as well as to control for socio-demographic variables in our first model specification for the first SC experiment, we have to construct ten interaction terms between the ten individual characteristics and the aggregated dummy variable for the consideration 13 We do not include alternative-specific constants as it is common in econometric analyses with data from unlabeled SC experiments (e.g. Goett et al., 20 0 0; Hensher et al., 2005).

of sustainability criteria (e.g. an interaction term between “warm glow” and “consideration of sustainability criteria”). This leads to overall K = 17 explanatory variables in this case. For the second model specification in the case of the first SC experiment, 20 interaction terms are included so that the number of explanatory variables increases to K = 29. The corresponding numbers for the second SC experiment are K = 15 in the first model specification and K = 27 in the second model specification. The values of Uijm cannot be observed and depend on the error terms ε ijm , which summarize all unobserved factors for the choice of an investment product. According to the random utility maximization theory (e.g. McFadden, 1973), it is generally assumed that a respondent chooses an investment product in a specific choice set if the utility for this alternative is the largest among the utilities for all four alternatives. With β i = β (i = 1,…, N) the assumption of independently and standard (type 1) extreme value distributed error terms ε ijm leads to the common multinomial logit model. This model approach is characterized by the very restrictive independence of irrelevant alternatives (IIA) property, which implies that the choice probabilities between two alternatives are independent of the existence of further alternatives. However, if the IIA assumption is not correct, the parameter estimates are inconsistent. In fact, this IIA assumption is mostly not adequate (e.g. Hoyos, 2010). Furthermore, multinomial logit models cannot capture unobserved taste heterogeneity and correlations due to the panel nature of our data since each respondent was asked over several choice sets (e.g. Adamowicz et al., 2011). Therefore, we consider much more flexible mixed logit models (e.g. McFadden and Train, 20 0 0; Hensher and Greene, 2003). While these models still assume independently and standard (type 1) extreme value distributed error terms ε ijm , they are not based on the restrictive IIA assumption, but allow for taste heterogeneity among the participants and thus are able to incorporate correlations between the choice alternatives. Mixed logit models (i.e. random parameters logit models as specific variants) specifically assume that the parameters β ik (i = 1,…, N) of the non-financial explanatory variables are continuously distributed across i (e.g. Greene, 2012):

βik = βk + σk uik The uik capture the individual specific heterogeneity and are (in our case) independently normally distributed with mean zero and standard deviation one. Furthermore, σ k is the standard deviation of the distribution of β ik around the mean β k . In contrast, the parameters of all interaction terms as well as the parameters of all financial attributes, i.e. the annual nominal interest rate for the first SC experiment and the value of the subscription fee, the net return in the last year, and the average annual net return in the last five

200

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

years for the second SC experiment are specified to be fixed as it is common practice (e.g. Goett et al., 20 0 0; Hensher et al., 2005). However, in contrast to multinomial logit models, the maximum likelihood (ML) estimation of mixed logit models is not feasible since the probabilities for the choice of the four alternatives (which are components of the loglikelihood function) are characterized by multiple integrals, which cannot be computed with deterministic numerical integration methods. Instead, the probabilities can be approximated by simulation methods. The inclusion of these simulated probabilities leads to the simulated maximum likelihood (SML) estimation (e.g. Revelt and Train, 1998; Train, 2009). For our SML estimation of mixed logit models we used the Stata command “mixlogit”, which was written by Hole (2007), and used R = 10 0 0 Halton draws.14 In contrast to the application of multinomial logit models, where one parameter per explanatory variable is estimated, this approach leads to the estimation of the mean and the additional estimation of the standard deviation of the random parameters (i.e. for the variables of the consideration of sustainability criteria and a transparency logo for both SC experiments and additionally of the providers for the first SC experiment), whereas fixed parameters for the interaction terms as well as for the annual nominal interest rate for the first SC experiment and for the value of the subscription fee, the net return in the last year, and the average annual net return in the last five years for the second SC experiment are estimated. On the basis of these estimated parameters, the WTP for each non-financial attribute (i.e. for the consideration of sustainability criteria and a transparency logo in the case of both SC experiments and additionally for the providers in the case of the first SC experiment) and for all interaction terms is estimated. While the WTP for a specific attribute is commonly based on a price or cost variable (e.g. in the case of the WTP for renewable and nuclear energy in electricity mixes), in our case of investment products, we consider the annual nominal interest rate for the first SC experiment and the subscription fee, the net return in the last year, and the average annual net return in the last five years for the second SC experiment. The basis for the estimation of a WTP is the utility function as discussed above. The mean WTP is the change of one financial variable that keeps the utility constant for a marginal change of the explanatory variable of interest. It can be derived by setting the total derivative of the utility function with respect to the explanatory variable (e.g. the consideration of sustainability criteria) and the financial variable (e.g. the annual nominal interest rate) to zero (assuming that all other explanatory variables are held fixed). Mathematically, this value can be estimated by the ratio between the negative value15 of the estimated parameter for the explanatory variable of interest and the estimated parameter of the financial variable. This procedure refers to the standard approach with fixed parameters such as in the case of multinomial logit models. The mean WTP in mixed logit models for attributes with assumed random parameters is estimated by the ratio between the negative

14 Also all other estimations and statistical analyses for this paper were conducted with Stata. 15 In common SC experiments with only cost or price attributes, which generally have negative effects on the choice among different alternatives (e.g. on the choice among vehicles), this leads to positive WTP estimates. In contrast, for most of our financial variables (i.e. for the annual nominal interest rate in the case of the first SC experiment and for the net return in the last year and the average annual net return in the last five years in the case of the second SC experiment), we expect positive effects on the choice among the investment products (an exception is the consideration of the value of the subscription fee in the case of the second SC experiment). This leads to negative WTP estimates or alternatively to positive estimates of the willingness to sacrifice returns. However, in line with common language use, we do not interpret the WTP estimates in negative values, but in absolute terms, i.e. more negative WTP estimates are interpreted as higher estimated WTP.

values of the estimated means of these random parameters and the estimated fixed parameters of the financial variables. 3.1.2. Basic estimation results Tables 4 and 5 report the parameter estimates (including robust z-statistics) and the mean WTP estimates in mixed logit models for the choice among four three-year fixed-interest investment products and for the choice among four equity funds, respectively. While the SML estimation results in Table 4 are based on all N = 1001 respondents (and thus 1001 × 6 = 6006 observations) who were included in the first SC experiment, the SML estimation results in Table 5 are based on all N = 801 respondents (and thus 801 × 8 = 6408 observations) who were included in the second SC experiment. The left parts of both tables refer to the inclusion of the aggregated dummy variables for the consideration of sustainability criteria and a transparency logo, whereas the right parts refer to the inclusion of the corresponding disaggregated variables. The upper parts of the tables reveal that almost all standard deviations of the random parameters (with one exception for the consideration of sustainability criteria without certificate in the choice among the three-year fixed-interest investment products, see Table 4) are strongly significantly different from zero, which indicates high unobserved heterogeneity among the respondents.16 According to the upper part of Table 4, the annual nominal interest rate has an expected significantly positive effect on the choice among the fixed-interest investment products. Furthermore, according to the upper part of Table 5, the value of the subscription fee has a significantly negative effect, while the net return in the last year and the average annual net return in the last five years have expected significantly positive effects on the choice among the equity funds. In addition, considering the estimated means of the parameters, all aggregated and disaggregated variables of the consideration of sustainability criteria and a transparency logo have strong significantly positive effects. With respect to the provider in the choice among the fixed-interest investment products, the estimation results in Table 4 report a significantly higher stated preference for direct banks and especially for sustainability banks, municipal savings banks, and co-operative banks than for big banks, respectively. While the estimated means of the attribute parameters in mixed logit models can be interpreted in terms of their signs, the size of the estimated effect cannot be directly identified.17 Nevertheless, the estimated means of the different provider parameters surprisingly imply a rather weak stated preference for sustainability banks compared with municipal savings or co-operative banks. In order to analyze the dimension of the estimated effects, probabilities for the choice of an investment product have to be estimated. The corresponding results imply that the effects are not only statistically, but also economically significant. For example, based on the estimation results in Tables 4 and 5, the estimated average discrete probability effects of the consideration of sustainability criteria are about nine percentage points for the choice of fixed-interest investment products and more than eleven percentage points for the choice of equity funds. This means that the estimated average probability of the choice of an investment product with sustainability criteria is about nine or more than eleven 16 The estimated standard deviations for the random parameters should not be confused with the estimated standard deviations of the estimated parameters, which are the basis for the reported z-statistics, respectively. Furthermore, it should be noted that the sign of the estimated standard deviations for the random parameters is irrelevant, i.e. negative estimates are interpreted as being positive (e.g. Hole, 2007). 17 For example, a significantly positive parameter for the consideration of sustainability criteria only implies that a change from zero to one in the case of this dummy variable leads to an estimated average increase of the utility or preference for the corresponding investment product.

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

201

Table 4 SML estimation results in mixed logit models for the choice among four three-year fixed-interest investment products. Explanatory variables

Estimates (robust z-statistics) Mean of the parameter

Standard deviation of the parameter

Mean of the parameter

–

4.121∗ ∗ ∗ (25.85)

–

Consideration of sustainability criteria without certificate

3.973∗ ∗ ∗ (26.17) 0.833∗ ∗ ∗ (14.19) –

1.067 (14.46) –

Consideration of sustainability criteria with certificate

–

–

Transparency logo

1.052∗ ∗ ∗ (12.88) –

– −0.036 (−0.13) 1.238∗ ∗ ∗ (14.61) –

Transparency logo issued by an NGO

1.027∗ ∗ ∗ (16.54) –

– 0.460∗ ∗ ∗ (8.25) 1.025∗ ∗ ∗ (14.36) –

Transparency logo issued by the state

–

–

Municipal savings bank

0.487∗ ∗ ∗ (6.55) 0.502∗ ∗ ∗ (7.05) 0.113∗ (1.79) 0.215∗ ∗ ∗ (2.82)

1.288∗ ∗ ∗ (12.06) 1.070∗ ∗ ∗ (10.30) 0.509∗ ∗ ∗ (3.35) 1.141∗ ∗ ∗ (10.95)

0.768∗ ∗ ∗ (13.42) 0.976∗ ∗ ∗ (14.61) 0.460∗ ∗ ∗ (6.02) 0.503∗ ∗ ∗ (6.84) 0.112∗ (1.73) 0.204∗ ∗ (2.50)

−0.327∗ (−1.84) 0.842∗ ∗ ∗ (10.59) 1.353∗ ∗ ∗ (12.40) 1.125∗ ∗ ∗ (10.61) 0.461∗ ∗ (2.17) 1.327∗ ∗ ∗ (12.50)

Annual nominal interest rate Consideration of sustainability criteria

Co-operative bank Direct bank Sustainability bank

Standard deviation of the parameter

∗∗∗

Mean WTP estimates (based on the annual nominal interest rate) Consideration of sustainability criteria Consideration of sustainability criteria without certificate Consideration of sustainability criteria with certificate Transparency logo Transparency logo issued by an NGO Transparency logo issued by the state Municipal savings bank Co-operative bank Direct bank Sustainability bank N (number of observations)

−0.210 – – −0.259 – – −0.123 −0.126 −0.028 −0.054 10 01 (60 06)

– −0.112 −0.249 – −0.186 −0.237 −0.112 −0.122 −0.027 −0.049 10 01 (60 06)

Note: For the SML estimations 10 0 0 Halton draws were used. The basis of the estimation results in this table are data across M = 6 choice sets. While the left part of the table refers to the model specification that includes the two aggregated dummy variables for the consideration of sustainability criteria and a transparency logo, the right part refers to the model specification that includes the corresponding four disaggregated dummy variables besides the annual nominal interest rate and the four provider dummy variables, respectively. The upper part of the table reports the parameter estimates for each explanatory variable. For fixed parameters (i.e. for the annual nominal interest rate), only one value is estimated, whereas for random parameters (i.e. for the variables of the consideration of sustainability criteria, a transparency logo, and the providers), estimates for the mean and for the standard deviation are reported. The corresponding robust z-statistics on the basis of the estimated standard deviations of the estimated parameters are in parentheses. ∗ (∗ ∗ , ∗ ∗ ∗ ) means that the appropriate parameter is different from zero at the 10% (5%, 1%) significance level, respectively. The lower part of the table reports the mean WTP estimates for each non-financial attribute. They are calculated by dividing the negative values of the estimated means of the random parameters by the estimated parameter of the annual nominal interest rate.

percentage points (and accordingly almost 45% or even more than 61%) higher than the estimated average probability of the choice of an investment product without sustainability criteria, respectively. Furthermore, the estimated average discrete probability effects of the consideration of sustainability criteria with certificate is more than twice as high as the corresponding estimated effects of the consideration of sustainability criteria without certificate (the respective values are about five and more than 12 percentage points for the choice of the fixed-interest investment products and about 7.4 and 15.1 percentage points for the choice of the equity funds). However, we do not focus on estimated probability effects, but directly include the financial variables of the SC experiments in our consideration and thus especially examine the estimated WTP. The lower parts of Tables 4 and 5 report the corresponding results. In line with the previous discussion, Table 4 reveals a lower estimated mean willingness to sacrifice annual nominal interest rates for sustainability banks than for municipal savings or co-operative banks in the choice among the fixed-interest investment products. Furthermore, the estimated mean WTP for a transparency logo is considerable, whereby the corresponding value for a transparency logo issued by the state is slightly higher than for a transparency

logo issued by an NGO. These results are qualitatively strongly confirmed in the analysis of the estimated mean WTP with respect to all three financial attributes in the choice among the equity funds, i.e. the value of the subscription fee, the net return in the last year, and the average annual net return in the last five years (see Table 5). However, the main estimation results refer to the consideration of sustainability criteria. The estimated mean willingness to sacrifice annual nominal interest rates for the aggregated variable is more than 0.2 percentage points in the choice among the fixedinterest investment products. Furthermore, the mean WTP estimate of almost 0.25 percentage points for the consideration of sustainability criteria with certificate is much higher than the corresponding value of 0.112 percentage points for the consideration of sustainability criteria without certificate, which suggests an additional considerable WTP for the certification of sustainable investment products. Again, these results are qualitatively strongly confirmed in the choice among the equity funds. Table 5 reveals, for example, an estimated mean willingness to sacrifice average annual net returns in the last five years of 2.519 percentage points, an estimated mean WTP for the consideration of sustainability cri-

202

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

Table 5 SML estimation results in mixed logit models for the choice among four equity funds. Explanatory variables

Estimates (robust z-statistics) Mean of the parameter

Standard deviation of the parameter

Mean of the parameter

Standard deviation of the parameter

−0.199∗ ∗ ∗ (−8.25) 0.195∗ ∗ ∗ (14.40) 0.338∗ ∗ ∗ (23.35) 0.851∗ ∗ ∗ (16.38) –

–

–

–

−0.212∗ ∗ ∗ (−8.28) 0.207∗ ∗ ∗ (14.70) 0.358∗ ∗ ∗ (23.30) –

–

–

– 0.965∗ ∗ ∗ (15.57) –

−0.392∗ ∗ ∗ (−4.07) 1.090∗ ∗ ∗ (15.84) –

Transparency logo issued by an NGO

0.840∗ ∗ ∗ (15.01) –

0.519∗ ∗ ∗ (11.42) 0.987∗ ∗ ∗ (16.13) –

Transparency logo issued by the state

–

–

0.620∗ ∗ ∗ (11.77) 0.820∗ ∗ ∗ (13.94)

0.620∗ ∗ ∗ (8.58) 0.851∗ ∗ ∗ (11.96)

Value of the subscription fee Net return in the last year Average annual net return in the last five years Consideration of sustainability criteria Consideration of sustainability criteria without certificate Consideration of sustainability criteria with certificate Transparency logo

–

0.995∗ ∗ ∗ (18.29) –

– –

Mean WTP estimates

Consideration of sustainability criteria Consideration of sustainability criteria without certificate Consideration of sustainability criteria with certificate Transparency logo Transparency logo issued by an NGO Transparency logo issued by the state N (number of observations)

Value of the subscription fee

Net return in the last year

Average annual net return in the last five years

Value of the subscription fee

Net return in the last year

4.273 –

−4.364 –

−2.519 –

– 2.455

– −2.513

– −1.451

–

–

–

4.668

−4.779

−2.758

−2.485 – –

– 2.929 3.878

– −2.999 −3.971 801 (6408)

– −1.731 −2.292

4.216 – –

−4.305 – – 801 (6408)

Average annual net return in the last five years

Note: For the SML estimations 10 0 0 Halton draws were used. The basis of the estimation results in this table are data across M = 8 choice sets. While the left part of the table refers to the model specification that includes the two aggregated dummy variables for the consideration of sustainability criteria and a transparency logo, the right part refers to the model specification that includes the corresponding four disaggregated dummy variables besides the value of the subscription fee, the net return in the last year, and the average annual net return in the last five years, respectively. The upper part of the table reports the parameter estimates for each explanatory variable. For fixed parameters (i.e. for the value of the subscription fee, the net return in the last year, and the average annual net return in the last five years), only one value is estimated, whereas for random parameters (i.e. for the variables of the consideration of sustainability criteria and a transparency logo), estimates for the mean and for the standard deviation are reported. The corresponding robust z-statistics on the basis of the estimated standard deviations of the estimated parameters are in parentheses. ∗ (∗ ∗ , ∗ ∗ ∗ ) means that the appropriate parameter is different from zero at the 10% (5%, 1%) significance level, respectively. The lower part of the table reports the mean WTP estimates for each non-financial attribute. They are calculated by dividing the negative values of the estimated means of the random parameters by the estimated parameters of the value of the subscription fee, the net return in the last year, and the average annual net return in the last five years.

teria with certificate of 2.758 percentage points, and an estimated mean WTP for the consideration of sustainability criteria without certificate of 1.451 percentage points. Overall, these estimation results suggest a high valuation of sustainable investments. As discussed above, however, it can naturally be argued that the stated preferences and WTP are overestimated due to the hypothetical character of the SC experiments without financial constraints. In general, hypothetical biases especially arise if participants respond strategically or give socially desirable answers. However, sustainable investments do not enjoy an exceptionally good reputation in the population, for example, due to announcements about violation of ethical standards or environmental pollution by firms that had good sustainability ratings and thus were components of sustainable equity funds or indexes like the Dow Jones Sustainability Indexes (e.g. BP with respect to the Deepwater Horizon oil spill or Volkswagen with respect to the emissions scandal). Therefore, we do not expect a strong hypothetical bias for our WTP estimates. Furthermore, the possible problem of hypothetical biases only refers to the absolute levels of the WTP estimates, whereas relative WTP estimates for different attributes or attribute levels, for example, the consideration of sustainability criteria with and without certificate in our case, are unaffected, as discussed above.

Since we cannot completely exclude possible hypothetical biases for our WTP estimates, we have conducted a series of robustness checks in order to examine the reliability of our estimation results, especially with respect to the choice among the fixed-interest investment products. Our first robustness check is based on an understanding question, i.e. the respondents were asked subsequent to the SC experiment whether they generally found the choice sets and the description of the choice situations comprehensible. A second even more relevant robustness check is based on another ex post technique to mitigate hypothetical biases, namely the use of certainty scales (e.g. Fifer et al., 2014), i.e. the respondents were asked subsequent to each choice set for the degree of certainty that they would purchase the chosen investment product in a real investment situation. Finally, the third robustness check combines these two approaches, i.e. the corresponding subsample only includes respondents who stated after the SC experiments that they understood the description of the choice situations and separately for each choice set that they were rather or very sure that they would purchase the chosen investment product in reality. The corresponding estimation results are qualitatively extremely similar to the estimation results in Table 4, i.e. the estimates for the mean willingness to sacrifice annual nominal interest

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

rates for the consideration of sustainability criteria and especially for the consideration of sustainability criteria with certificate are very high.18 These results suggest that hypothetical biases should not play an important role. Our result of a high WTP for sustainable investment products is further strengthened in additional robustness checks. We have, for example, analyzed several alternative numbers R of Halton draws for the SML estimation in the mixed logit models and the inclusion of alternative-specific constants. Furthermore, we have analyzed mixed logit model specifications that only include data of the first choice set for each respondent in order to exclude possible biased choices due to fatigue. However, the estimation results for the consideration of sustainability criteria on the basis of these additional robustness checks19 do not qualitatively differ from the previous estimation results and thus confirm the considerable WTP for sustainable investment products. In sum, our estimation results suggest that investors have on average strong preferences for sustainable investments, which is in line with previous studies on money flows of investments (e.g. Bollen, 2007; Benson and Humphrey, 2008; Renneboog et al., 2011; Hartzmark and Sussman, 2018) and with previous econometric studies on the basis of data at the individual level (e.g. Nilsson, 2008; Bauer and Smeets, 2015; Wins and Zwergel, 2016; Riedl and Smeets, 2017; Gutsche et al., 2019). Due to the methodological approach of our SC experiments, which directly considers returns (including interest rates) as attributes of investment products, our estimation result cannot be explained by the perception of sustainability criteria as signal for higher future returns as, for example, discussed in Hartzmark and Sussman (2018). Instead, our econometric analysis of sustainable investment products directly controls for returns, i.e. we identify estimated stated preferences for sustainability criteria in different investment products with identical returns. Interestingly, our estimated mean willingness to sacrifice annual nominal interest rates for sustainability criteria in the choice among the fixed-interest investment products is lower than the estimated WTP in Henke (2015) as discussed above, which is based on an analysis of bond fund flows and thus finally on revealed instead of stated preferences as in our case. This comparison strengthens our already discussed view that hypothetical biases should not play an important role. As mentioned in Hartzmark and Sussman (2018), one possible explanation for the positive valuation of sustainable investments refers to the salience theory according to Bordalo et al. (2012, 2013a, 2013b). The main idea behind this psychologically founded theory is that individual behavior is based on the attention to the most salient aspects in a real choice context, which leads to an overweighting of these attributes. In this view, the salience detection is an essential attentional mechanism that enables individuals to focus their cognitive limitations on some relevant choice attributes. In our case it is therefore possible that the attention of the respondents is drawn to sustainability criteria, which are more salient, for example, through the extensive public discussion of sustainability topics like climate change. Similarly or even more obviously, the attention of the respondents might be more strongly drawn to certified sustainable investment products since certificates are simply more salient. This would explain the higher mean WTP estimates for the consideration of sustainability criteria with certificate compared to the counterparts without certificate. This argumentation is supported by the high mean WTP estimates for transparency logos, which might also be salient attributes due to controversially discussed past events in the financial sector such as 18 These estimation results are not reported due to brevity, but are available upon request. 19 These estimation results are not reported due to brevity, either, but are also available upon request.

203

the Libor scandal in 2011 or some obscure practices in the context of the financial crisis of 2007 and 2008. As discussed in many previous studies (e.g. Nilsson, 2008; Bauer and Smeets, 2015; Wins and Zwergel, 2016; Riedl and Smeets, 2017; Hartzmark and Sussman, 2018; Gutsche et al., 2019), another popular explanation for the positive valuation of sustainable investments is that investors expect that these investments are less risky. In this case, our variable “high perceived risk” should have a negative effect on the stated preferences for sustainability criteria. In this respect, it can be speculated that especially certified sustainable investment products are expected to be less risky. We discuss the corresponding estimation results in the following subsection. However, the following subsection particularly focuses on a final explanation for the positive valuation of sustainable investments, namely the relevance of non-pecuniary motives, which are extensively analyzed in previous investment flow studies as well as econometric studies on the basis of individual level data as discussed above. In contrast to most of these studies, we consider different non-financial psychological motives, values, and norms, i.e. warm glow feelings, expectations of the social environment, environmental awareness, and political identification. 3.1.3. Relevance of risk perceptions and psychological motives, values, and norms Tables 6 and 7 report the parameter estimates (including robust z-statistics) and the mean WTP estimates in mixed logit models for the choice among the fixed-interest investment products and for the choice among the equity funds,20 respectively, that include interaction terms between the ten individual characteristics and the aggregated dummy variable for sustainability criteria besides the aggregated dummy variables for the consideration of sustainability criteria and a transparency logo. While the SML estimation results in Table 6 are based on N = 599 respondents (and thus 3594 observations), the SML estimation results in Table 7 are based on N = 516 respondents (and thus 4128 observations). This implies that overall 402 participants who were included in the first SC experiment and 285 participants who were included in the second SC experiment refused to answer at least one of the ten questions for the individual characteristics, especially with respect to the perceived risk of sustainable investments and political identification (see Table 3). While it might be argued that this can lead to selection problems, a comparison between the estimation results in Tables 4 and 5 on the basis of all N = 1001 or N = 801 respondents and the corresponding estimation results with the smaller numbers of N = 599 or N = 516 respondents reveals some quantitative, but almost no qualitative differences so that this restricted number of observations does obviously not lead to biased estimation results.21 Tables 6 and 7 reveal that most standard deviations of the random parameters are again significantly different from zero. In line with Tables 4 and 5, Tables 6 and 7 show a significantly positive effect of and similar mean WTP estimates for a transparency logo. In addition, Table 6 still reveals significantly higher stated preferences for municipal savings banks, co-operative banks, and sustainability banks than for big banks, whereas the mean of the parameter for direct banks becomes insignificant, which is possibly due to the lower number of observations. Furthermore, the means of the

20 For brevity, we now only consider the WTP on the basis of the average net return in the last five years in the choice among the equity funds since they seem to be most appropriate to reflect the belief in future returns of equity funds. 21 The estimation results are not reported due to brevity, but are available upon request. An additional comparison of the descriptive statistics in Table 3 and the descriptive statistics on the basis of the smaller number of N = 599 respondents for the choice among the fixed-interest investment products also shows only some slight differences for single individual characteristics, i.e. for “warm glow”, “female”, and “high education”.

204

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

Table 6 SML estimation results in a mixed logit model for the choice among four three-year fixed-interest investment products with interaction terms of aggregated dummy variables. Explanatory variables

Annual nominal interest rate Consideration of sustainability criteria Consideration of sustainability criteria × High perceived risk Consideration of sustainability criteria × Warm glow Consideration of sustainability criteria × Expectation social environment Consideration of sustainability criteria × Membership environmental organization Consideration of sustainability criteria × Affinity left-wing parties Consideration of sustainability criteria × Female Consideration of sustainability criteria × Age Consideration of sustainability criteria × High education Consideration of sustainability criteria × Living together or married Consideration of sustainability criteria × Western Germany Transparency logo Municipal savings bank Co-operative bank Direct bank Sustainability bank N (number of observations)

Estimates (robust z-statistics)

Mean WTP estimates (based on the annual nominal interest rate)

Mean of the parameter

Standard deviation of the parameter

4.017∗ ∗ ∗ (20.47) 0.239 (0.74) −0.116 (−0.85) 0.735∗ ∗ ∗ (5.29) 0.029 (0.13) 0.289 (1.24) 0.541∗ ∗ ∗ (4.06) 0.123 (0.94) 0.004 (0.77) 0.072 (0.53) 0.164 (1.19) −0.285∗ (−1.68) 1.064∗ ∗ ∗ (13.33) 0.359∗ ∗ ∗ (3.95) 0.474∗ ∗ ∗ (5.14) 0.124 (1.51) 0.189∗ ∗ (2.00) 599 (3594)

–

–

0.863∗ ∗ ∗ (9.34) –

n.s.

–

−0.183

–

n.s.

–

n.s.

–

−0.135

–

n.s.

–

n.s

–

n.s.

–

n.s.

–

0.071

0.962∗ ∗ ∗ (10.33) 1.041∗ ∗ ∗ (8.38) 1.091∗ ∗ ∗ (8.00) 0.275 (1.01) 0.932∗ ∗ ∗ (7.09)

−0.265

n.s.

−0.089 −0.118 n.s. −0.047

Note: For the SML estimation 10 0 0 Halton draws were used. The basis of the estimation results in this table are data across M = 6 choice sets. The model specification includes the two aggregated dummy variables for the consideration of sustainability criteria and a transparency logo besides the annual nominal interest rate and the four provider dummy variables. In addition, the model includes interaction terms between the ten individual characteristics and the aggregated dummy variable for the consideration of sustainability criteria. The left part of the table reports the parameter estimates for each explanatory variable. For fixed parameters (i.e. for the annual nominal interest rate and the interaction terms) only one value is estimated, whereas for random parameters (i.e. for the variables of the consideration of sustainability criteria, a transparency logo, and the providers) estimates for the mean and for the standard deviation are reported. The corresponding robust z-statistics on the basis of the estimated standard deviations of the estimated parameters are in parentheses. ∗ (∗ ∗ , ∗ ∗ ∗ ) means that the appropriate parameter is different from zero at the 10% (5%, 1%) significance level, respectively. The right part of the table reports the mean WTP estimates for each non-financial attribute. They are calculated by dividing the negative values of the estimated fixed parameters (in the case of the interaction terms) or the estimated means of the random parameters by the estimated parameter of the annual nominal interest rate (n.s. means that the underlying mean of the random parameter is not significantly different from zero).

parameters for the consideration of sustainability criteria in both mixed logit models are not significantly different from zero according to Tables 6 and 7. This result suggests that some of the ten individual characteristics are able to explain the estimated strong preferences for sustainable investments since the interpretation for this variable refers to the case that all interaction terms are zero. The main estimation results refer to the estimated parameters of the interaction terms with the individual characteristics, which imply that respondents from Western Germany have a weakly significantly lower stated preference for the consideration of sustainability criteria in the choice among the fixed-interest investment products. In contrast, a high perceived risk of sustainable investments has no significant effect on the stated preference for sustainability criteria. Instead, Table 6 shows a strong relevance of two non-pecuniary motives since respondents with an affinity to leftwing parties and especially with high feelings of warm glow have significantly higher stated preferences for the consideration of sustainability criteria. The corresponding estimated mean WTP differences are considerable, i.e. an affinity to left-wing parties leads to an estimated mean WTP for the consideration of sustainability criteria that is 0.135 percentage points higher than for respondents

without affinity to left-wing parties. For the two groups of respondents with and without high feelings of warm glow, the estimated mean WTP difference is even 0.183 percentage points, which implies an overall strong effect of psychological motives, values, and norms on sustainable investments. The latter main results are confirmed in Table 7, i.e. respondents with both an affinity to left-wing parties and with high feelings of warm glow also have significantly higher stated preferences for the consideration of sustainability criteria in the choice among the equity funds. In addition, another non-pecuniary motive obviously plays an important role for the choice among the equity funds since respondents who are members of environmental organizations also have a significantly higher stated preference for the consideration of sustainability criteria. Besides significant effects of gender and age, another difference between the results in Tables 7 and 6 refers to the relevance of perceived risk of sustainable investments. The estimation results for this variable in Table 7 reveals that respondents who expect a similar or lower risk for sustainable investments have a significantly higher stated preference for the consideration of sustainability criteria. Therefore, besides some socio-demographic variables and especially several

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

205

Table 7 SML estimation results in a mixed logit model for the choice among four equity funds with interaction terms of aggregated dummy variables. Explanatory variables

Value of the subscription fee Net return in the last year Average annual net return in the last five years Consideration of sustainability criteria Consideration of sustainability criteria × High perceived risk Consideration of sustainability criteria × Warm glow Consideration of sustainability criteria × Expectation social environment Consideration of sustainability criteria × Membership environmental organization Consideration of sustainability criteria × Affinity left-wing parties Consideration of sustainability criteria × Female Consideration of sustainability criteria × Age Consideration of sustainability criteria × High education Consideration of sustainability criteria × Living together or married Consideration of sustainability criteria × Western Germany Transparency logo N (number of observations)

Estimates (robust z-statistics)

Mean WTP estimates (based on the average annual net return in the last five years)

Mean of the parameter

Standard deviation of the parameter

−0.198∗ ∗ ∗ (−6.59) 0.184∗ ∗ ∗ (10.99) 0.354∗ ∗ ∗ (19.44) −0.111 (−0.37) −0.303∗ ∗ (−2.50) 0.666∗ ∗ ∗ (5.73) −0.119 (−0.57) 0.433∗ ∗ (2.10) 0.613∗ ∗ ∗ (5.15) −0.215∗ (−1.85) 0.010∗ ∗ (2.23) −0.001 (−0.01) 0.184 (1.51) 0.004 (0.03) 0.856∗ ∗ ∗ (12.96) 516 (4128)

–

–

–

–

–

–

0.873∗ ∗ ∗ (12.47) –

n.s. 0.856

–

−1.881

–

n.s.

–

−1.224

–

−1.732

–

0.608

–

−0.028

–

n.s.

–

n.s.

–

n.s.

0.875∗ ∗ ∗ (12.39)

−2.418

Note: For the SML estimation 10 0 0 Halton draws were used. The basis of the estimation results in this table are data across M = 8 choice sets. The model specification includes the two aggregated dummy variables for the consideration of sustainability criteria and a transparency logo besides the value of the subscription fee, the net return in the last year, and the average annual net return in the last five years. In addition, the model includes interaction terms between the ten individual characteristics and the aggregated dummy variable for the consideration of sustainability criteria. The left part of the table reports the parameter estimates for each explanatory variable. For fixed parameters (i.e. for the value of the subscription fee, the net return in the last year, the average annual net return in the last five years, and the interaction terms) only one value is estimated, whereas for random parameters (i.e. for the variables of the consideration of sustainability criteria and a transparency logo) estimates for the mean and for the standard deviation are reported. The corresponding robust z-statistics on the basis of the estimated standard deviations of the estimated parameters are in parentheses. ∗ (∗ ∗ , ∗ ∗ ∗ ) means that the appropriate parameter is different from zero at the 10% (5%, 1%) significance level, respectively. The right part of the table reports the mean WTP estimates with respect to the average annual net return in the last five years for each non-financial attribute. They are calculated by dividing the negative values of the estimated fixed parameters (in the case of the interaction terms) or the estimated means of the random parameters by the estimated parameter of the average annual net return in the last five years (n.s. means that the underlying mean of the random parameter is not significantly different from zero).

psychological motives, values, and norms, also a financial motive obviously matters for the choice among the equity funds. This result suggests that risk motives are only relevant for sustainable investments when very risky investment products like equity funds are considered. Tables 8 and 9 report the parameter estimates (including robust z-statistics) and the mean WTP estimates in mixed logit models for the choice among the fixed-interest investment products and for the choice among the equity funds, respectively, that now include interaction terms between the ten individual characteristics and the disaggregated dummy variables for sustainability criteria besides the disaggregated dummy variables for the consideration of sustainability criteria and a transparency logo. The SML estimation results are again based on N = 599 and N = 516 respondents, respectively. The main result in Table 8 is that respondents with an affinity to left-wing parties and with high feelings of warm glow have significantly higher stated preferences for both uncertified and certified sustainable fixed-interest investment products. However, the mean WTP estimate of respondents with an affinity to left-wing parties is slightly and the corresponding WTP estimate of respondents with high feelings of warm glow is strongly higher for certified compared to uncertified sustainable fixed-interest investment products. Furthermore, respondents who are member of en-

vironmental organizations have a weakly significantly higher stated preference for certified sustainable fixed-interest investment products, whereas the effect of the corresponding variable on uncertified sustainable investment products remains insignificant. Similarly, Table 9 reveals that respondents who are member of environmental organizations only have a significantly higher stated preference for certified sustainable equity funds. Furthermore, respondents with an affinity to left-wing parties and with high feelings of warm glow also have significantly higher stated preferences for both uncertified and certified sustainable equity funds, whereby the corresponding mean WTP estimates are again higher for certified sustainable equity funds. Interestingly, respondents who expect a similar or lower risk for sustainable investments only have a significantly higher stated preference for certified sustainable equity funds, whereas the effect of the corresponding variable on uncertified sustainable equity funds is insignificant. This result suggests that risk motives are only relevant for risky and certified sustainable investments. The estimation results might be influenced by correlations between the ten individual characteristics and thus by multicollinearity problems, which may make the identification of significant effects difficult. However, in spite of such possible problems, our estimation results still reveal some very stable significant effects of

206

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

Table 8 SML estimation results in a mixed logit model for the choice among four three-year fixed-interest investment products with interaction terms of disaggregated dummy variables. Explanatory variables

Annual nominal interest rate Consideration of sustainability criteria without certificate Consideration of sustainability criteria without certificate × High perceived risk Consideration of sustainability criteria without certificate × Warm glow Consideration of sustainability criteria without certificate × Expectation social environment Consideration of sustainability criteria without certificate × Membership environmental organization Consideration of sustainability criteria without certificate × Affinity left-wing parties Consideration of sustainability criteria without certificate × Female Consideration of sustainability criteria without certificate × Age Consideration of sustainability criteria without certificate × High education Consideration of sustainability criteria without certificate × Living together or married Consideration of sustainability criteria without certificate × Western Germany Consideration of sustainability criteria with certificate Consideration of sustainability criteria with certificate × High perceived risk Consideration of sustainability criteria with certificate × Warm glow Consideration of sustainability criteria with certificate × Expectation social environment Consideration of sustainability criteria with certificate × Membership environmental organization Consideration of sustainability criteria with certificate × Affinity left-wing parties Consideration of sustainability criteria with certificate × Female Consideration of sustainability criteria with certificate × Age Consideration of sustainability criteria with certificate × High education Consideration of sustainability criteria with certificate × Living together or married Consideration of sustainability criteria with certificate × Western Germany Transparency logo issued by an NGO Transparency logo issued by the state Municipal savings bank Co-operative bank Direct bank Sustainability bank N (number of observations)

Estimates (robust z-statistics)

Mean WTP estimates (based on the annual nominal interest rate)

Mean of the parameter

Standard deviation of the parameter

4.358∗ ∗ ∗ (19.94) 0.355 (1.03) −0.137 (−0.95) 0.312∗ ∗ (2.06) 0.104 (0.41) 0.020 (0.08) 0.537∗ ∗ ∗ (3.78) 0.091 (0.66) 0.003 (0.48) 0.058 (0.40) 0.047 (0.31) −0.321∗ (−1.87) 0.106 (0.24) −0.183 (−1.02) 1.110∗ ∗ ∗ (6.20) −0.154 (−0.54) 0.518∗ (1.80) 0.618∗ ∗ ∗ (3.54) 0.154 (0.88) 0.006 (0.80) 0.058 (0.31) 0.246 (1.35) −0.225 (−0.99) 0.867∗ ∗ ∗ (11.91) 1.078∗ ∗ ∗ (11.98) 0.341∗ ∗ ∗ (3.59) 0.492∗ ∗ ∗ (5.02) 0.086 (1.01) 0.166 (1.64) 599 (3594)

–

–

0.147 (0.30) –

n.s.

–

−0.072

–

n.s.

–

n.s.

–

−0.123

–

n.s.

–

n.s

–

n.s.

–

n.s.

–

0.074

1.167∗ ∗ ∗ (11.15) –

n.s.

–

−0.255

–

n.s.

–

−0.119

–

−0.142

–

n.s.

–

n.s.

–

n.s.

–

n.s.

–

n.s.

−0.168 (−0.89) 0.915∗ ∗ ∗ (8.68) 1.121∗ ∗ ∗ (8.13) 1.218∗ ∗ ∗ (8.59) −0.382∗ (−1.87) 1.072∗ ∗ ∗ (8.00)

−0.199

n.s.

n.s.

−0.247 −0.078 −0.113 n.s. n.s.

Note: For the SML estimation 10 0 0 Halton draws were used. The basis of the estimation results in this table are data across M = 6 choice sets. The model specification includes the four disaggregated dummy variables for the consideration of sustainability criteria and a transparency logo besides the annual nominal interest rate and the four provider dummy variables. In addition, the model includes interaction terms between the ten individual characteristics and the two disaggregated dummy variables for the consideration of sustainability criteria. The left part of the table reports the parameter estimates for each explanatory variable. For fixed parameters (i.e. for the annual nominal interest rate and the interaction terms) only one value is estimated, whereas for random parameters (i.e. for the variables of the consideration of sustainability criteria, a transparency logo, and the providers) estimates for the mean and for the standard deviation are reported. The corresponding robust z-statistics on the basis of the estimated standard deviations of the estimated parameters are in parentheses. ∗ (∗ ∗ , ∗ ∗ ∗ ) means that the appropriate parameter is different from zero at the 10% (5%, 1%) significance level, respectively. The right part of the table reports the mean WTP estimates for each non-financial attribute. They are calculated by dividing the negative values of the estimated fixed parameters (in the case of the interaction terms) or the estimated means of the random parameters by the estimated parameter of the annual nominal interest rate (n.s. means that the underlying mean of the random parameter is not significantly different from zero).

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

207

Table 9 SML estimation results in a mixed logit model for the choice among four equity funds with interaction terms of disaggregated dummy variables. Explanatory variables

Value of the subscription fee Net return in the last year Average annual net return in the last five years Consideration of sustainability criteria without certificate Consideration of sustainability criteria without certificate × High perceived risk Consideration of sustainability criteria without certificate × Warm glow Consideration of sustainability criteria without certificate × Expectation social environment Consideration of sustainability criteria without certificate × Membership environmental organization Consideration of sustainability criteria without certificate × Affinity left-wing parties Consideration of sustainability criteria without certificate × Female Consideration of sustainability criteria without certificate × Age Consideration of sustainability criteria without certificate × High education Consideration of sustainability criteria without certificate × Living together or married Consideration of sustainability criteria without certificate × Western Germany Consideration of sustainability criteria with certificate Consideration of sustainability criteria with certificate × High perceived risk Consideration of sustainability criteria with certificate × Warm glow Consideration of sustainability criteria with certificate × Expectation social environment Consideration of sustainability criteria with certificate × Membership environmental organization Consideration of sustainability criteria with certificate × Affinity left-wing parties Consideration of sustainability criteria with certificate × Female Consideration of sustainability criteria with certificate × Age Consideration of sustainability criteria with certificate × High education Consideration of sustainability criteria with certificate × Living together or married Consideration of sustainability criteria with certificate × Western Germany Transparency logo issued by an NGO Transparency logo issued by the state N (number of observations)

Estimates (robust z-statistics)

Mean WTP estimates (based on the average annual net return in the last five years)

Mean of the parameter

Standard deviation of the parameter

−0.211∗ ∗ ∗ (−6.60) 0.200∗ ∗ ∗ (11.49) 0.380∗ ∗ ∗ (19.49) −0.046 (−0.16) −0.187 (−1.58) 0.546∗ ∗ ∗ (4.71) 0.110 (0.60) 0.259 (1.31) 0.298∗ ∗ ∗ (2.59) −0.283∗ ∗ (−2.46) 0.005 (1.14) 0.019 (0.15) 0.161 (1.37) 0.051 (0.35) −0.390 (−0.98) −0.359∗ ∗ (−2.33) 0.824∗ ∗ ∗ (5.58) −0.275 (−0.97) 0.565∗ ∗ (2.19) 0.865∗ ∗ ∗ (5.75) −0.158 (−1.07) 0.017∗ ∗ ∗ (2.97) 0.003 (0.02) 0.180 (1.19) −0.044 (−0.22) 0.649∗ ∗ ∗ (10.09) 0.884∗ ∗ ∗ (12.08) 516 (4128)

–

–

–

–

–

–

0.148 (0.56) –

n.s. n.s.

–

−1.437

–

n.s.

–

n.s.

–

−0.785

–

0.744

–

n.s.

–

n.s.

–

n.s.

–

n.s. ∗∗∗

n.s.

0.952 (11.84) –

0.946

–

−2.170

–

n.s.

–

−1.488

–

−2.278

–

n.s.

–

−0.045

–

n.s.

–

n.s.

–

n.s.

0.592∗ ∗ ∗ (6.29) 0.867∗ ∗ ∗ (10.20)

−1.708 −2.328

Note: For the SML estimation 10 0 0 Halton draws were used. The basis of the estimation results in this table are data across M = 8 choice sets. The model specification includes the four disaggregated dummy variables for the consideration of sustainability criteria and a transparency logo besides the value of the subscription fee, the net return in the last year, and the average annual net return in the last five years. In addition, the model includes interaction terms between the ten individual characteristics and the two disaggregated dummy variables for the consideration of sustainability criteria. The left part of the table reports the parameter estimates for each explanatory variable. For fixed parameters (i.e. for the value of the subscription fee, the net return in the last year, the average annual net return in the last five years, and the interaction terms) only one value is estimated, whereas for random parameters (i.e. for the variables of the consideration of sustainability criteria and a transparency logo) estimates for the mean and for the standard deviation are reported. The corresponding robust z-statistics on the basis of the estimated standard deviations of the estimated parameters are in parentheses. ∗ (∗ ∗ , ∗ ∗ ∗ ) means that the appropriate parameter is different from zero at the 10% (5%, 1%) significance level, respectively. The right part of the table reports the mean WTP estimates with respect to the average annual net return in the last five years for each non-financial attribute. They are calculated by dividing the negative values of the estimated fixed parameters (in the case of the interaction terms) or the estimated means of the random parameters by the estimated parameter of the average annual net return in the last five years (n.s. means that the underlying mean of the random parameter is not significantly different from zero).

208

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

risk perceptions and especially psychological motives, values, and norms so that multicollinearity problems obviously play no important role. We have nevertheless examined the correlations between the individual characteristics in more detail. In fact, they are mostly rather moderate, especially in terms of the relationships between the socio-demographic variables on the one hand and the variables for risk perceptions and psychological motives, values, and norms on the other hand. For the latter variables the strongest relationships refer to “warm glow” and “membership environmental organization”, “warm glow” and “high perceived risk”, and especially “warm glow” and “expectation social environment” with (Pearson) correlations coefficients of 0.148, −0.184, and 0.287, respectively. We have therefore examined restricted mixed logit models that exclude the interaction terms between “warm glow” and the disaggregated dummy variables for sustainability criteria. The corresponding estimation results are qualitatively mostly very similar to the estimation results in Tables 8 and 9. 22 3.2. Latent class logit model analysis 3.2.1. Econometric approach The previous mixed logit model analysis allows the direct estimation of the mean WTP for different sustainable investment products. Methodologically, mixed logit models provide a flexible framework that allows for taste heterogeneity among the participants in our SC experiments. In fact, the previous analysis reveals a large extent of such unobserved heterogeneity. In mixed logit models taste heterogeneity is generally included by specific assumptions about the continuous distribution of the parameters. However, the underlying heterogeneity can sometimes be better reflected as discrete rather than continuous (e.g. Adamowicz et al., 2011) such as possibly in our case if a specific investor group has a general higher WTP for sustainable investment products. Latent class logit models do not require specific assumptions about the distributions of the parameters, but generally assume discrete mixing distributions (e.g. Greene and Hensher, 20 03; Train, 20 09), whereby the heterogeneity in the parameters can be explained by individual characteristics. While latent class logit models cannot directly provide the estimation of the mean WTP for sustainable investment products and also not directly analyze the average relevance of individual characteristics such as non-pecuniary motives in our case, an important advantage is that a potential problem of our mixed logit models analysis, namely that the parameters of the financial variables and the interaction terms are assumed to be fixed, is directly avoided. Therefore, we complement our mixed logit model analysis by a corresponding latent class logit model analysis, whereby we focus on our data on fixed-interest investment products (examining the aggregated dummy variables for the consideration of sustainability criteria and a transparency logo), which generally allow a

22 Only the positively estimated parameters of the interaction terms with “expectation social environment” and “membership environmental organization” are slightly more often significantly different from zero, which supports our model specifications with all ten individual characteristics (i.e. especially with “warm glow”) since otherwise some parameter estimates and thus estimated effects might be slightly distorted due to omitted variable biases. These additional estimation results are not reported due to brevity, but are available upon request. In order to test the robustness of our estimation results, we have also examined restricted mixed logit models that only include those interaction terms between the individual characteristics and the disaggregated dummy variables, for which the corresponding parameters are significantly different from zero in the two models as discussed before. The corresponding estimation results for these interaction terms are qualitatively mostly very similar to the respective estimation results in Tables 8 and 9, especially for the variables for risk perceptions and psychological motives, values, and norms. The estimation results are not reported due to brevity, either, but are available upon request.

cleaner setting for the estimation of WTP compared to the consideration of past return profiles of equity funds as discussed above.23 Specifically, latent class logit models assume that respondents are implicitly sorted into a set of Q classes and that the error terms εijm are still independently and standard (type 1) extreme value distributed. In our case, the probability that respondent i chooses investment product j in choice set m if i belongs to investor class q (q = 1,…, Q), is: 

Pi jmq (βq ) =

eβq xi jm 4   e βq x i j  m

j =1

Now β q = (β q1 ,…, β q7 )’ is a class-specific vector of parameters for the seven attributes for investor class q. However, the class membership is unknown. The respondents are probabilistically assigned to the Q different investor classes by a class membership model. It is assumed that the membership to an investor class q depends on the vector zi = (zi1 ,…, zi,11 )’ which generally includes individual characteristics, i.e. in our case the perceived risk of sustainable investments, four variables for psychological motives, values, and norms, and the five sociodemographic variables as discussed above besides a constant. The corresponding unknown parameter vector is θ q = (θ q1 ,…, θ q,11 )’. By additionally assuming that the error terms in the class membership model are independently and standard (type 1) extreme value distributed, the probability that respondent i belongs to investor class q is: 

Hiq =

e θq z i Q  θ z e q i

q =1

The maximization of the corresponding loglikelihood function thus refers to the Q structural parameter vectors β 1 ,…, β Q and Q1 latent class parameter vectors θ 1 ,…, θ Q-1 (i.e. θ Q is normalized to the zero vector to ensure the formal identification of the other latent class parameter vectors). However, this optimization problem is numerically not trivial compared to other common ML estimations (e.g. Greene and Hensher, 2003). In line with Train (2009), we used the Expectation-Maximization (EM) algorithm to guarantee numerical stability and convergence of the loglikelihood function to a maximum even for a higher number of classes. The corresponding algorithm for the EM ML estimation has recently been included in a Stata module written by Pacifico and Yoo (2013). In our econometric analysis, we compare the estimation results in latent class logit models with Q = 2 and Q = 3 investor classes.24 On the basis of the estimation results, we finally estimated the mean WTP by dividing for each class the negative values of the estimated parameters of the six non-financial attributes (especially including the variable for the consideration of sustainability criteria) by the estimated parameter of the annual nominal interest rate. 3.2.2. Estimation results Table 10 reports the estimation results in the latent class logit model with Q = 2 investor classes, whereby the lower part of the table shows that more than 60% of the respondents are members

23 While we have also conducted a latent class logit model analysis for our data on equity funds, we do not report the estimation results due to brevity. In fact, these estimation results, which are available upon request, provide very similar conclusions as the corresponding results for fixed-interest investment products. 24 This choice is based on a content-related motivation of the class memberships as discussed below and thus not solely on statistical information criteria such as the Bayesian information criterion (BIC) or the consistent Akaike information criterion (CAIC).

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

209

Table 10 EM ML estimation results in a latent class logit model with Q = 2 investor classes for the choice among four three-year fixed-interest investment products. Variables

Annual nominal interest rate Consideration of sustainability criteria Transparency logo Municipal savings bank Co-operative bank Direct bank Sustainability bank

Class 1

Class 2

Parameter estimates (robust z-statistics)

Mean WTP estimates (based on the annual nominal interest rate)

Parameter estimates (robust z-statistics)

1.741∗ ∗ ∗ (13.12) 1.006∗ ∗ ∗ (16.00) 0.881∗ ∗ ∗ (13.84) 0.663∗ ∗ ∗ (7.34) 0.648∗ ∗ ∗ (7.08) 0.281∗ ∗ ∗ (3.04) 0.487∗ ∗ ∗ (5.33)

–

9.950∗ ∗ ∗ (14.67) 0.697∗ ∗ ∗ (4.83) 0.944∗ ∗ ∗ (6.92) 0.113 (0.62) 0.382∗ ∗ (2.13) −0.196 (−1.08) −0.297 (−1.44)

−0.578 −0.506 −0.381 −0.372 −0.161 −0.280

Mean WTP estimates (based on the annual nominal interest rate) – −0.070 −0.095 n.s. −0.038 n.s. n.s.

Parameter estimates (robust z-statistics) High perceived risk Warm glow Expectation social environment Membership environmental organization Affinity left-wing parties Female Age High education Living together or married Western Germany Constant Class share N (number of observations)

0.109 (0.48) 1.082∗ ∗ ∗ (4.51) 0.853∗ (1.89) 1.300∗ ∗ ∗ (2.68) 0.662∗ ∗ ∗ (2.98) 0.325 (1.42) 0.014 (1.63) −0.500∗ ∗ (−2.10) 0.027 (0.11) −0.568∗ ∗ (−1.98) −0.520 (−0.93) 0.615

– – – – – – – – – – – 0.385 599 (3594)

Note: The basis of the estimation results in this table are data across M = 6 choice sets. The upper part of the table reports for both investor classes and each explanatory variable the (fixed) parameter estimates and the corresponding robust z-statistics that are based on the estimated standard deviations of the estimated parameters in parentheses. It additionally reports the mean WTP estimates for each non-financial attribute. They are calculated by dividing the negative values of the estimated parameters of the non-financial attributes by the estimated parameter of the annual nominal interest rate (n.s. means that the underlying parameter is not significantly different from zero). The lower part of the table reports the parameter estimates and the corresponding robust z-statistics in parentheses for the individual characteristics that explain the investor class membership. The parameters for the second investor class are normalized to zero so that only parameters for the first investor class are estimated. ∗ (∗ ∗ , ∗ ∗ ∗ ) means that the appropriate parameter is different from zero at the 10% (5%, 1%) significance level, respectively.

of the first class. The upper part of the table reveals strong differences in the estimated stated preferences between the two investor classes. The main difference refers to the estimated parameters for the annual nominal interest rate, which is almost six times higher in the second investor class. In combination with the estimated parameters for the other attributes, this implies a much higher sensitivity to this return variable in the second investor class. On the basis of the higher parameter estimate for the consideration of sustainability criteria, this leads to a mean WTP estimate of 0.578 percentage points in the first investor class that is more than eight times higher than in the second class. Similarly, the estimated mean WTP for a transparency logo is also strongly higher in the first investor class. In addition, the estimation results for the second investor class reveal a significantly higher stated preference for co-operative banks, but insignificant differences in the stated preferences for municipal savings banks, direct banks and sustainability banks compared to big banks, respectively, which also points to a strong heterogeneity between the two investor classes.

The lower part of Table 10 especially reveals the composition of the two investor classes. The estimation results for the sociodemographic variables imply that lower education groups and participants from Eastern Germany tend to be members of the first investor class with a higher WTP for sustainable investment products. However, the main result refers to the other individual characteristics. Respondents with high feelings of warm glow from sustainable investments, an affinity to left-wing parties, and a strong environmental awareness have a significantly higher probability to be members of the first investor class. In addition, “expectation social environment” has a weakly significantly positive effect on the membership in the first investor class, whereas “high perceived risk” has no significant effect. In sum, these estimation results strongly confirm the main results from the mixed logit model analysis with fixed-interest investment products that respondents with an affinity to left-wing parties and with high feelings of warm glow have significantly higher stated preferences for the consideration of sustainability criteria, whereas the perceived risk of

210

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

Table 11 EM ML estimation results in a latent class logit model with Q = 3 investor classes for the choice among four three-year fixed-interest investment products. Variables

Annual nominal interest rate Consideration of sustainability criteria Transparency logo Municipal savings bank Co-operative bank Direct bank Sustainability bank

Class 1 (“sustainability investor group”)

Class 2 (“transparency investor group”)

Parameter estimates (robust z-statistics)

Mean WTP estimates (based on the annual nominal interest rate)

Parameter estimates (robust z-statistics)

Mean WTP estimates (based on the annual nominal interest rate)

Parameter estimates (robust z-statistics)

2.812∗ ∗ ∗ (12.75) 1.672∗ ∗ ∗ (12.72) 1.368∗ ∗ ∗ (12.27) 0.576∗ ∗ ∗ (3.86) 0.673∗ ∗ ∗ (4.47) 0.545∗ ∗ ∗ (3.69) 0.925∗ ∗ ∗ (6.26)

–

0.789∗ ∗ ∗ (3.44) 0.283∗ ∗ (2.46) 0.385∗ ∗ ∗ (2.95) 0.754∗ ∗ ∗ (4.77) 0.691∗ ∗ ∗ (4.29) −0.038 (−0.23) −0.398∗ (−1.86)

–

10.292∗ ∗ ∗ (13.89) 0.663∗ ∗ ∗ (4.35) 0.998∗ ∗ ∗ (6.61) 0.124 (0.65) 0.314∗ (1.67) −0.253 (−1.34) −0.335 (−1.55)

−0.595 −0.487 −0.205 −0.239 −0.194 −0.329

−0.358 −0.488 −0.955 −0.875 n.s. 0.505

Class 3 (“financial performance investor group”) Mean WTP estimates (based on the annual nominal interest rate) – −0.064 −0.097 n.s. −0.031 n.s. n.s.

Parameter estimates (robust z-statistics) High perceived risk Warm glow Expectation social environment Membership environmental organization Affinity left-wing parties Female Age High education Living together or married Western Germany Constant Class share N (number of observations)

−0.099 (−0.35) 1.758∗ ∗ ∗ (6.05) 0.480 (0.93) 1.377∗ ∗ (2.45) 1.023∗ ∗ ∗ (3.81) 0.369 (1.33) 0.023∗ ∗ (2.09) −0.401 (−1.37) 0.200 (0.68) −0.585∗ (−1.74) −1.947∗ ∗ ∗ (−2.66) 0.419

0.467 (1.64) 0.171 (0.53) 1.457∗ ∗ ∗ (2.82) 1.509∗ ∗ ∗ (2.67) 0.236 (0.82) 0.240 (0.84) 0.005 (0.44) −0.660∗ ∗ (−2.25) −0.145 (−0.49) −0.391 (−1.09) −0.576 (−0.84) 0.213 599 (3594)

– – – – – – – – – – – 0.368

Note: The basis of the estimation results in this table are data across M = 6 choice sets. The upper part of the table reports for all three investor classes and each explanatory variable the (fixed) parameter estimates and the corresponding robust z-statistics that are based on the estimated standard deviations of the estimated parameters in parentheses. It additionally reports the mean WTP estimates for each non-financial attribute. They are calculated by dividing the negative values of the estimated parameters of the non-financial attributes by the estimated parameter of the annual nominal interest rate (n.s. means that the underlying parameter is not significantly different from zero). The lower part of the table reports the parameter estimates and the corresponding robust z-statistics in parentheses for the individual characteristics that explain the investor class membership. The parameters for the third investor class are normalized to zero so that only parameters for the first and second investor classes are estimated. ∗ ∗∗ ∗∗∗ ( , ) means that the appropriate parameter is different from zero at the 10% (5%, 1%) significance level, respectively.

sustainable investments obviously only plays a negligible role. Furthermore, the result for the membership in environmental organizations in Table 10 suggests a stronger evidence for this nonpecuniary motive than in the mixed logit model analysis, where only a significantly higher stated preference for certified sustainable fixed-interest investment products is identified. The main results from the mixed logit and the first latent class logit model analysis with fixed-interest investment products are also strongly confirmed in the latent class logit model analysis with three investor classes. Table 11 reports the corresponding estimation results, whereby the lower part of the table shows that almost 42% of the respondents are members of the first investor class, more than 21% are members of the second class, and almost 37% are members of the third investor class. The estimation results reveal for the third investor class strong similarities with the estimation results for the second class in the latent class logit model with two investor classes (see Table 10). Both investor classes are

characterized by a large estimated sensitivity to the interest rate, low mean WTP estimates for the consideration of sustainability criteria and a transparency logo, and low differences in the estimated stated preferences for several providers. This investor class can thus be termed as “financial performance investor group”. As a consequence, the previous first investor class in the latent class logit model with two investor classes is more or less divided in the first two classes in the model with three investor classes. The first investor class in this latent class logit model is largely in line with the first class in the previous latent class logit model with high mean WTP estimates for a transparency logo and several providers and especially with very high mean WTP estimates for the consideration of sustainability criteria and also (in contrast to the first class in the previous latent class logit model) for sustainability banks. This investor class can thus be termed as “sustainability investor group”. In contrast, the second investor class can be termed as “transparency investor group” due to the high

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

mean WTP estimates for a transparency logo and also for municipal savings and co-operative banks, which are considered as much more transparent and reliable in Germany due to the public and co-operative ownership. While the mean WTP estimate for sustainable fixed-interest investment products in this investor class is clearly lower than in the first class, it is still much higher than in the third investor class, even when the members have a weakly significantly lower stated preference for sustainability banks. According to the estimation results in the lower part of Table 11, respondents from Eastern Germany, older respondents, and especially participants with high feelings of warm glow from sustainable investments, a strong environmental awareness, and an affinity to left-wing parties tend to be members of the “sustainability investor group” with a high estimated mean willingness to sacrifice interest rates for sustainable fixed-interest investment products compared to the membership in the base investor class, i.e. the “financial performance investor group”. While “expectation social environment” has no significant effect on the membership in the “sustainability investor group”, it has a significantly positive effect on the membership in the “transparency investor group” with strong estimated stated preferences for transparency and reliability criteria. The estimation results also imply that respondents with a strong environmental awareness and lower education groups tend to be members of the “transparency investor group” compared to the membership in the “financial performance investor group”. 4. Conclusion This paper empirically examines the willingness of private financial decision makers in Germany to pay for sustainable investments. Traditional finance theory suggests a WTP of zero since sustainable investment products would only be considered if they are at least as attractive as other investments in terms of risk and returns. However, several empirical studies reveal the relevance of non-pecuniary motives for sustainable investments and thus indirectly a WTP for sustainable investment products. Our study contributes to this literature by directly analyzing the WTP for sustainable investment products. Methodologically, our econometric analysis with data from two SC experiments for fixed-interest investment products and equity funds is based on flexible discrete choice models, namely mixed and latent class logit models. This enables us to examine the relevance of unobserved heterogeneity among different respondents and investor groups, which would not be possible if restricted discrete choice models like the multinomial logit model were used. Both our mixed and latent class logit model analyses imply a high extent of unobserved heterogeneity among financial decision makers and especially strong stated preferences and a considerable WTP for sustainable fixed-interest investment products and sustainable equity funds. Furthermore, our mixed logit model analysis reveals that the estimated mean WTP for certified sustainable investment products is strongly higher than the estimated mean WTP for the uncertified counterparts. In addition, our estimation results on the basis of both mixed and latent class logit models imply a strongly higher mean willingness to sacrifice returns for sustainable investment products among respondents with specific psychological motives, values, and norms, i.e. for investors with high feelings of warm glow from sustainable investments, an affinity to left-wing parties, and a strong environmental awareness, compared with their unsustainable counterparts. While risk perceptions seem to be additionally relevant for certified sustainable equity funds, they obviously play a negligible role for less risky sustainable fixed-interest investment products. This relevance of psychological motives, values, and norms is in line with previous investments studies (e.g. Bauer and Smeets, 2015; Riedl and Smeets, 2017), but also with other empirical

211

environmental economics studies that generally consider proenvironmental behavior including climate protection activities (e.g. Kotchen and Moore, 2008; Schwirplies and Ziegler, 2016). Indeed, a positive effect of an affinity to left-wing parties seems to contradict the estimation results in Gutsche et al. (2019), who report rather negative effects of a left-wing orientation on sustainable investments. However, they specifically examine the share of sustainable investments among all investments, which are currently mainly characterized by very risky investments. The results of Gutsche et al. (2019) can therefore be explained by the general aversion of left-wing parties and their supporters to risky investments and the participation in stock markets (e.g. Kaustia and Torstila, 2011), i.e. left-wing oriented financial decision makers are rather more skeptical toward general equity investments than toward sustainable investments. Thus, the difference is that their analysis does not disentangle the sustainability and risk dimensions of current sustainable investment products. In contrast, our empirical analysis on the basis of the SC experiments is able to identify the relevance of the consideration of sustainability criteria in the specific choice among different equity funds or especially among different fixed-interest investment products. This sophisticated picture of left-wing investors and our general identification of investor groups with a higher WTP for sustainable investment products have important implications for the development of specific sustainable investments products and marketing activities by banks or other providers of financial investments in order to attract new sustainable investments customers. In this respect, it would certainly be interesting to analyze uncertified or certified sustainable fixed-interest investment products, which are currently only an extremely small segment in the investments universe, in more detail in future empirical studies. In order to test the generalizability of our specific estimation results for Germany, analyses for other capital markets, with other investment products, and/or alternative return structures are also an interesting direction for further research. However, it might be argued that the reliability and external validity of our estimation results is uncertain due to the hypothetical character of our SC experiments. One concern refers to the general hypothetical biases of stated preferences studies. However, it should be noted that our WTP estimates are very stable across several robustness checks with mixed logit models including techniques to mitigate possible hypothetical biases. While overestimations of the WTP can nevertheless not be completely excluded, this possible problem only refers to the absolute levels. In contrast, it is argued in the experimental literature (e.g. Goett et al., 20 0 0; Cohn et al., 2015) that relative WTP estimates for different attributes or attribute levels and especially comparative static effects of individual characteristics are unaffected by possible hypothetical biases. In this respect, we assume a high reliability of the estimated WTP for sustainable investment products without and with certificate as well as for of the estimated effects of non-pecuniary psychological motives, values, and norms. Another concern might refer to the specific setting of our SC experiments. In line with the salience theory according to Bordalo et al. (2012, 2013a, 2013b) suggesting that individual behavior is based on the attention to the most salient aspects in a choice context, it can be argued that the attention of the respondents in the experiments is more strongly directed to sustainability criteria than to the relatively few other attributes including returns. Due to the even higher salience, the attention of the respondents can be especially strong for (sustainability) certificates and (transparency) logos. Therefore, it might be argued that the WTP estimates for these attributes are overvalued and thus overestimated. However, in line with Hartzmark and Sussman (2018), it can be replied that the overweighting of some salient attributes is not necessarily a

212

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

specific characteristic of hypothetical SC experiments, but can also be found in real investment decisions and thus real market behavior, at least if these investment products are offered. Therefore, we expect that the overestimations due to this salience argument are rather weak, especially since our hypothetical SC experiments are formulated as realistic and practically relevant as possible with respect to (future) real investment choice situations. Nevertheless, we can naturally not completely exclude any biases with respect to real investment decisions in our SC experiments and thus restrictions in the external validity of our estimation results. More reliable analyses with an even higher external validity require revealed preferences and thus real market data. However, for our empirical analysis of new uncertified or certified sustainable investment products with transparency logos from different providers as well as of sustainable fixed-interest investment products, such data are not available so far. In general, we share the possible problem of low external validity of our empirical results with other previous stated preferences studies in this field, which, for example, directly asked for the WTP for sustainable investment products (e.g. Dorfleitner and Utz, 2014; Borgers and Pownal, 2014) or which are based on rather simple experimental settings (e.g. Webley et al., 2001; Pasewark and Riley, 2010; Berry and Yeung, 2013; Hartzmark and Sussman, 2018). In this respect, the main novelty and thus a main contribution of our study is the use of data from SC experiments as it is common for the estimation of WTP in other economic sub-disciplines like transportation economics. In contrast to these previous studies (including the empirical analyses of Nilsson, 2008, Wins and Zwergel, 2016, and Riedl and Smeets, 2017), our empirical analysis is additionally especially based on data from a representative sample, which should further increase the external validity of our estimation results, at least for German private financial decision makers. We additionally expect that the external validity of our sophisticated SC experiments is less restricted than the external validity of previous relatively simple incentivized laboratory experiments, especially if they are conducted with students. For example, in the experiment of Barreda-Tarrazona et al. (2011) the participants had to invest real money in real existing mutual funds. While their experimental approach is therefore certainly very attractive and an interesting starting point for future studies, it should be noted that their used amount of individual investments is rather small (i.e. 16 Euro). Further potential shortcomings of their approach are the artificial and thus also hypothetical scenarios for risk and returns of the funds. In contrast, field experiments for the analysis of WTP for uncertified or certified sustainable fixed-interest investment products, but also sustainable equity funds are an interesting alternative approach. A first (natural) field experiment can be found in Døskeland and Pedersen (2016). As they only consider the effect of wealth or morality framings on sustainable investments, field experimental analyses of the WTP for sustainable investment products and the relevance of psychological motives, values, and norms are left for future studies. While such field experiments are certainly an attractive direction of further research, it can be argued that even they are also restricted if, for example, only sustainable investors from a sustainable financial provider are considered, which are not representative for the whole capital market, as discussed in Hartzmark and Sussman (2018). Appendix: Questionnaire Appendix A. Description of the first SC experiment on three-year fixed-interest investment products (translated into English) On each of the following six pages you will be shown four different three-year fixed-interest investment products, which differ in terms of certain characteristics. Please indicate for each of the

six decision situations which of the four investment products is as attractive to you that you would acquire it most likely. It is possible that some of the investment products with their listed properties are currently not provided by banks. This should not bother you. Just imagine that these investment products are available in reality. Please additionally assume for your choice that the investment products are identical with respect to all characteristics that are not shown (e.g. type of investment, full deposit guarantee, etc.). Some of the properties of the three-year fixed-interest investment products are explained in the following: Annual nominal interest rate: The individual investment products differ in terms of the fixed annual nominal interest rate for the time of maturity of three years. Sustainability criteria: The individual investment products differ in terms of whether ecological, social, and/or ethical criteria are considered besides financial criteria in their construction process. Such sustainable investment products additionally differ in terms of whether they receive a sustainability certificate. In this case, the consideration of sustainability criteria is audited and confirmed by an independent organization. Transparency logo: The individual investment products differ in terms of whether they receive a transparency logo. Such investment products are characterized by the fact that extensive information on each investment strategy (also including the sustainability criteria if appropriate) is published. Investment products with a transparency logo further differ in terms of whether the logo is issued by a state agency or by a non-governmental organization (NGO). Provider: The individual investment products differ in terms of whether they are provided by a big bank, a municipal savings bank, a cooperative bank, a direct bank, or (in the case of the consideration of sustainability criteria) a sustainability bank. Direct banks are banks providing banking services without having an own branch-office network. Appendix B. Description of the second SC experiment on equity funds (translated into English) After you have chosen among different three-year fixed-interest investment products, on each of the following eight pages you will be shown four different equity funds, which differ in terms of certain characteristics. Please indicate for each of the eight decision situations, which of the four equity funds is as attractive to you that you would acquire it most likely. It is possible that some of the equity funds with their listed properties are currently not provided on the capital market. This should again not bother you. Just imagine that these equity funds are available in reality. Please additionally assume for your choice that the equity funds are identical with respect to all characteristics that are not shown (e.g. fund size, fund age, etc.). Some of the properties of the equity funds are explained in the following: Net return in the last year: The individual equity funds differ in terms of the net return (return minus all costs and fees except subscription fees) of the last year. Average annual net return in the last five years: The individual equity funds differ in terms of the net return (return minus all cost and fees except subscription fees) in the last five years. Value of subscription fees: The individual equity funds differ in terms of the level of the one-time subscription fee, measured in % of the amount of money invested.

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

213

Sustainability criteria: The individual equity funds differ in terms of whether ecological, social, and/or ethical criteria are considered besides financial criteria in their construction process. Such sustainable equity funds additionally differ in terms of whether they receive a sustainability certificate. In this case, the consideration of sustainability criteria is audited and confirmed by an independent organization. Transparency logo: The individual equity funds differ in terms of whether they receive a transparency logo. Such equity funds are characterized by the fact that extensive information on each investment strategy (also including the sustainability criteria if appropriate) is published. Equity funds with a transparency logo further differ in terms of whether the logo is issued by a state agency or by a nongovernmental organization (NGO).

Variable: “Affinity left-wing parties” Please indicate with which political party you are most likely affiliated (even when you vote for another party occasionally).

Appendix C. Survey questions for variables in the econometric analysis (translated into English)

 Male  Female

        

Variable: “Female” Are you … ?

Variable: “Age” Please indicate your age in years. ______ years

Variable: “High perceived risk” Please indicate your assessment of the average risk level of sustainable investments compared to conventional investments.  The average risk  The average risk  The average risk vestments.  The average risk  The average risk  No statement

is much lower for sustainable investments. is rather lower for sustainable investments. is neither higher nor lower for sustainable inis rather higher for sustainable investments. is much higher for sustainable investments.

Variable: “Warm glow” Please indicate how strongly you agree to the following statements regarding sustainable investments. “It makes me feel good to make sustainable investments.” I totally disagree

I rather weakly agree

I neither strongly nor weakly agree

I rather strongly agree

I totally agree

No statement

❍

❍

❍

❍

❍

❍

“I feel responsible for a sustainable development and want to contribute by making sustainable investments.” I totally disagree

I rather weakly agree

I neither strongly nor weakly agree

I rather strongly agree

I totally agree

No statement

❍

❍

❍

❍

❍

❍

Variable: “Expectation social environment” Please indicate how strongly you agree to the following statements regarding sustainable investments. “My social environment (e.g. family, friends, colleagues) expects me to make sustainable investments” I totally disagree

I rather weakly agree

I neither strongly nor weakly agree

I rather strongly agree

I totally agree

No statement

❍

❍

❍

❍

❍

❍

Variable: “Membership environmental organization” Are you member of a group or organization engaging in the conservation and protection of the environment and nature?  Yes  No  No statement

Christian Democrats Social Democrats Liberals Green Party Left Party AfD Pirate Party Another party No statement

Variable: “High education” Please indicate your highest educational level.         

No educational level German “Hauptschulabschluss” Secondary school certificate (German “Mittlere Reife”) Advanced technical college certificate, high school graduation (German “Abitur”) Bachelor degree / degree from a University of Applied Sciences Master degree / diploma from a University Doctorate / habilitation Other educational level (enter): ______ No statement Variable: “Living together or married” Please indicate your marital status.

     

Single Living together Married Divorced or living separated Widowed No statement

Variable: “Western Germany” Please indicate the federal state in which you have your primary residence.                

Baden-Wurttemberg Bavaria Berlin Brandenburg Bremen Hamburg Hesse Mecklenburg-Western Pomerania Lower Saxony North Rhine-Westfalia Rhineland-Palatinate Saarland Saxony Saxony-Anhalt Schleswig-Holstein Thuringia

214

G. Gutsche and A. Ziegler / Journal of Banking and Finance 102 (2019) 193–214

References Achtnicht, M., Bühler, G., Hermeling, C., 2012. The impact of fuel availability on demand for alternative-fuel vehicles. Transp. Res. Part D 17 (3), 262–269. Adamowicz, W., Dupont, D., Krupnick, A., Zhang, J., 2011. Valuation of cancer and microbial disease risk reductions in municipal drinking water: an analysis of risk context using multiple valuation methods. J. Environ. Econ. Manag. 61 (2), 213–226. Akerlof, G.A., Kranton, R.E., 20 0 0. Economics and identity. Q. J. Econ. 115 (3), 715–753. Andreoni, J., 1990. Impure altruism and donations to public goods: a theory of warm-glow giving. Econ. J. 100 (401), 464–477. Barreda-Tarrazona, I., Matallín-Sáez, J.C., Balaguer-Franch, M.R., 2011. Measuring investors’ socially responsible preferences in mutual funds. J. Bus. Ethics 103 (2), 305–330. Bauer, R., Smeets, P., 2015. Social identification and investment decisions. J. Econ. Behav. Organ. 117, 121–134. Belghitar, Y., Clark, E., Deshmukh, N., 2014. Does it pay to be ethical? Evidence from the FTSE4Good. J. Bank. Financ. 47, 54–62. Benson, K.L., Humphrey, J.E., 2008. Socially responsible investment funds: investor reaction to current and past returns. J. Bank. Financ. 32 (9), 1850–1859. Berry, R.H., Yeung, F., 2013. Are investors willing to sacrifice cash for morality? J. Bus. Ethics 117 (3), 477–492. Bollen, N.P.B., 2007. Mutual fund attributes and investor behavior. J. Financ. Quant. Anal. 42 (3), 683–708. Bordalo, P., Gennaioli, N., Shleifer, A., 2012. Salience theory of choice under risk. Q. J. Econ. 127 (3), 1243–1285. Bordalo, P., Gennaioli, N., Shleifer, A., 2013a. Salience and asset prices. In: Proceedings of the American Economic Review: Papers and Proceedings, 103, pp. 623–628. Bordalo, P., Gennaioli, N., Shleifer, A., 2013b. Salience and consumer choice. J. Polit. Econ. 121 (5), 803–843. Borgers, A., Pownall, R.A., 2014. Attitudes towards socially and environmentally responsible investments. J. Behav. Exp. Financ. 1, 27–44. Borgers, A., Derwall, J., Koedijk, K., ter Host, J., 2015. Do social factors influence investment behaviour and performance? Evidence from mutual fund holdings. J. Bank. Financ. 60, 112–126. Chrzan, K. and B. Orme (20 0 0), An overview and comparison of design strategies for choice-based conjoint analysis, Sawtooth Software research paper series. http: //www.sawtoothsoftware.com/download/techpap/desgncbc.pdf, Assessed March 17, 2019 Cohn, A., Engelmann, J., Fehr, E., Maréchal, M.A., 2015. Evidence for countercyclical risk aversion: an experiment with financial professionals. Am. Econ. Rev. 105 (2), 860–885. Derwall, J., Koedijk, K., ter Horst, J., 2011. A tale of values-driven and profit-seeking social investors. J. Bank. Financ. 35 (8), 2137–2147. Derwall, J., Guenster, N., Bauer, R., Koedijk, K., 2005. The eco-efficiency premium puzzle. Financ. Anal. J. 61 (2), 51–63. Dorfleitner, G., Utz, S., 2014. Profiling German-speaking socially responsible investors. Qual. Res. Financ. Mark. 6 (2), 118–156. Døskeland, T., Pedersen, L.J.T., 2016. Investing with brain or heart? A field experiment on responsible investment. Manag. Sci. 62 (6), 1632–1644. Eccles, R.G., Ioannou, I., Serafeim, G., 2014. The impact of corporate sustainability on organizational processes and performance. Manag. Sci. 60 (11), 2835–2857. Edmans, A., 2011. Does the stock market fully value intangibles? Employee satisfaction and equity prices. J. Financ. Econ. 101 (3), 621–640. Eurosif (2018), European SRI study 2018. Fabozzi, F.J., Ma, K.C., Oliphant, B.J., 2008. Sin stock returns. J. Portf. Manag. 35 (1), 82–94. Fifer, S., Rose, J., Greaves, S., 2014. Hypothetical bias in stated choice experiments: is it a problem? And if so, how do we deal with it? Transp. Res. Part A 61, 164–177. Goett, A.A., Hudson, K., Train, K.E., 20 0 0. Customers’ choice among retail energy suppliers: the willingness-to-pay for service attributes. Energy J. 21 (4), 1–28. Graham, J.R., Harvey, C.R., 2001. The theory and practice of corporate finance: evidence from the field. J. Financ. Econ. 60, 187–243. Greene, W.H., 2012. Econometric Analysis„ 7th ed. Pearson Education, Edinburgh Gate, Harlow. Greene, W.H., Hensher, D.A., 2003. A latent class model for discrete choice analysis: contrasts with mixed logit. Transp. Res. Part B 37 (8), 681–698. Guiso, L., Sapienza, P., Zingales, L., 2013. The determinants of attitudes toward strategic default of mortgages. J. Financ. 68 (4), 1473–1515. Gutsche, G., A. Köbrich León, and A. Ziegler (2019), On the relevance of psychological motives, values, and norms for socially responsible investments: an econometric analysis, Oxford Economic Papers, DOI: https://doi.org/10.1093/oep/ gpy051.

Hartzmark, S.M. and A.B. Sussman (2018), Do investors value sustainability? A natural experiment examining ranking and fund flows, Finance Working Paper No. 565/2018, ECGI Working Paper Series in Finance. Heinkel, R., Kraus, A., Zechner, J., 2001. The effect of green investment on corporate behavior. J. Financ. Quant. Anal 36 (4), 431–449. Henke, H.-M. The influence of social preferences on bond investor flows. Hensher, D.A., Greene, W.H., 2003. The mixed logit model: the state of praxis. Transportation 30 (2), 133–176. Hensher, D.A., Rose, J.M., Greene, W.H., 2010. Applied Choice Analysis: A Primer. Cambridge University Press, Cambridge. Hensher, D.A., Shore, N., Train, K., 2005. Households’ willingness to pay for water service attributes. Environ. Resour. Econ. 32 (4), 509–531. Hole, A.R., 2007. Fitting mixed logit models using maximum simulated likelihood. Stata J. 7 (3), 388–401. Hong, H., Kacperczyk, M., 2009. The price of sin: the effects of social norms on markets. J. Financ. Econ. 93 (1), 15–36. Hong, H., Kostovetsky, L., 2012. Red and blue investing: values and finance. J. Financ. Econ. 103 (1), 1–19. Hood, M., Nofsinger, J.R., Varma, A., 2014. Conservation, discrimination, and salvation: investors’ social concerns in the stock market. J. Financ. Serv. Res. 45 (1), 5–37. Hoyos, D., 2010. The state of the art of environmental valuation with discrete choice experiments. Ecol. Econ. 69 (8), 1595–1603. Kaustia, M., Torstila, S., 2011. Stock market aversion? Political preferences and stock market participation. J. Financ. Econ. 100 (1), 98–112. Kempf, A., Osthoff, P., 2007. The effect of socially responsible investing on portfolio performance. Eur. Financ. Manag. 13 (5), 908–922. Kotchen, M.J., Moore, M.R., 2008. Conservation: from voluntary restraint to a voluntary price premium. Environ. Resour Econ. 40 (2), 195–215. Louviere, J.J., Flynn, T.N., Carson, R.T., 2010. Discrete choice experiments are not conjoint analysis. J. Choice Model. 3 (3), 57–72. McFadden, D., 1973. Conditional logit analysis of qualitative choice models. In: Zarembka, P. (Ed.), Frontiers in Econometrics. Academic Press, New York, pp. 105–142. McFadden, D., Train, K., 20 0 0. Mixed MNL models for discrete response. J. Appl. Econom. 15 (5), 447–470. Murakami, K., Ida, T., Tanaka, M., Freedman, L., 2015. Consumers’ willingness to pay for renewable and nuclear energy: a comparative analysis between the US and Japan. Energy Econ. 50, 178–189. Nakai, M., Honda, T., Nishino, N., Takeuchi, K., 2018. Psychological characteristics of potential SRI investors and its motivation in Japan: an experimental approach. J. Sustain. Financ. Invest. 8 (4), 349–367. Nilsson, J., 2008. Investment with a conscience: examining the impact of pro-social attitudes and perceived financial performance on socially responsible investment behavior. J. Bus. Ethics 83 (2), 307–325. Nyborg, K., Rege, M., 2003. Does public policy crowd out private contributions to public goods. Public Choice 115 (3–4), 397–418. Oberndorfer, U., Schmidt, P., Wagner, M., Ziegler, A., 2013. Does the stock market value the inclusion in a sustainability stock index? An event study analysis for German firms. J. Environ. Econ. Manag 66 (3), 497–509. Pacifico, D., Yoo, H., 2013. lclogit: a stata command for fitting latent-class conditional logit models via the expectation-maximization algorithm. Stata J. 13 (3), 625–639. Pasewark, W.R., Riley, M.E., 2010. It’s a matter of principle: the role of personal values in investment decisions. J. Bus. Ethics 93 (2), 237–253. Renneboog, L., Horst, J.T., Zhang, C., 2011. Is ethical money financially smart? Nonfinancial attributes and money flows of socially responsible investment funds. J. Financ. Intermed. 20 (4), 562–588. Revelt, D., Train, K., 1998. Mixed logit with repeated choices: households’ choices of appliance efficiency level. Rev. Econ. Stat. 80 (4), 647–657. Riedl, A., Smeets, P., 2017. Why do investors hold socially responsible mutual funds? J. Financ. 72 (6), 2505–2550. Schwirplies, C., Ziegler, A., 2016. Offset carbon emissions or pay a price premium for avoiding them? A cross-country analysis of motives for climate protection. Appl. Econ. 48 (9), 746–758. Train, K.E., 2009. Discrete Choice Methods With Simulation, 2nd ed. Cambridge University Press, Cambridge, New York. US SIF (2018), Report on US sustainable, responsible and impact investing trends 2018, https://www.ussif.org/files/Trends/Trends%202018%20executive% 20summary%20FINAL.pdf, Assessed on March 17, 2019. Webley, P., Lewis, A., Mackenzie, C., 2001. Commitment among ethical investors: an experimental approach. J. Econ. Psychol. 22 (1), 27–42. Wins, A., Zwergel, B., 2016. Comparing those who do, might and will not invest in sustainable funds: a survey among German retail find investors. Bus. Res. 9, 51–99.