* Professor of Finance, at INSEAD, Boulevard de Constance, Fontainebleau Cedex, France.

Transcription

1 "CUSTOMER LOYALTY, SCALE ECONOMIES AND ECONOMIES OF SCOPE IN FRENCH FUNDS: ADDITIONAL EVIDENCE" by Jean DERMINE* Lars Hendrik ROLLER** and Carole BONANNI*** 93/08/EPS/FIN * Professor of Finance, at INSEAD, Boulevard de Constance, Fontainebleau Cedex, France. ** Associate Professor of Economics, at INSEAD, Boulevard de Constance, Fontainebleau Cedex, France. *** Research Assistant, at INSEAD, Boulevard de Constance, Fontainebleau Cedex, France. Printed at INSEAD Fontainebleau, France

2 CUSTOMER LOYALTY, SCALE ECONOMIES, and ECONOMIES OF SCOPE IN FRENCH MUTUAL FUNDS : ADDITIONAL EVIDENCE by C. Bonanni, J. Dermine and L.H. Miler* INSEAD, Fontainebleau February 1993 * This project was funded by INSEAD International Financial Services Research Programme.

3 CUSTOMER LOYALTY, SCALE ECONOMIES, and ECONOMIES OF SCOPE IN FRENCH MUTUAL FUNDS : ADDMONAL EVIDENCE Abstract The paper evaluates the effects of rapid market growth and product-age on economies of scale and scope in the French mutual funds (SICAV) industry. This segment of the financial sector offers some unique characteristics. The first is that some firms specialize while others supply several products. The second is that the French mutual funds industry has grown by 35 % from 1987 to The third is that the age (time since creation) of each mutual fund is known, so that product-age effects can be identified. By using jointly the 1987 and 1989 cross-sectional data sets and the age of each mutual fund, we are able to test the stability of economies of scale and scope estimates reported in a previous study.

4 I. Introduction This paper contributes to the literature on economies of scale and scope in the financial services industry. Concerned mostly with depository institutions, previous US studies have found little evidence of scale economies, except for banks with assets under $ 100 million. The evidence also fails to support the conclusion of global economies of scope'. In a recent paper, Dermine and Roller (1992) have estimated the degree of economies of scale and scope in the mutual funds industry. Using a 1987 data set for France, they demonstrate evidence of economies of scale and scope for small institutions. For larger institutions, they are unable to find evidence of economies of scale or scope. Additional empirical evidence is provided in this paper. Firstly, we evaluate the effect of rapid market growth on the stability of these estimates. The French mutual funds (SICAV2) industry has grown by 35 % from 1987 to Secondly, having access to the time of creation of each mutual fund, we can attempt to differentiate empirically age effect from pure scale and scope economies. These issues are investigated with 1987 and 1989 crosssectional data for the French mutual funds. The characteristics of the data set are particularly well suited for this study. First of all, our data on French mutual funds allow to study the degree of economies of scope as some institutions specialize completely, while other offer a large range of services. Moreover, there are large and small institutions, and the industry has grown rapidly between 1987 and Finally, the time of creation of each product is well identified. The paper is organized as follows. Section 2 provides a description of the data. The model is developed in section 3, and the empirical results follow in Section 4. Section 5 concludes. 'Gilbert (1984), Kolari, and Zardkoohi (1987), and Clarke (1988) provide surveys. A recent study of the American securities industry (Goldberg et al, 1991) reports similar conclusions. 2A SICAV, Societe d'investissement a Capital Variable, is an open-end mutual fund. This paper is among the first to study the mutual funds industry. Ferris and Chance (1987) provide evidence on the American mutual funds industry, but do not test for economies of scope, nor do they use any of the flexible functional forms.

5 3 2. The French SICAVs Industry The multiproduct firm is an institution that offers several types of SICAVs. Our data set includes two years, 1987 and The 1989 sample is described here 3. The sample includes 162 institutions. As Table 1 indicates, institutions range in size from FF 100 million to FF 220 billion. As the last quintile shows, there are a number of very large institutions. The five largest control 57 % of the market. The assets managed by the mutual fund industry has grown 35 % from 1987 to <Insert Table 1> Twelve categories of SICAV are recognized in France. To limit the number of parameter estimates, we have grouped them into five categories : Money Market Funds (MMF), French Bonds (FRB), French Equities (FRE), Real Estate (REE), and International (INT)4. Data on the number of SICAVs, their size, age, and fee structure' are given in Table 2. <Insert Table 2> The money market funds dominate the industry, ranging in size from FF 100 million to FF billion. The age (time since creation) of the SICAVs is dispersed, from 33 days to 23 years. Many institutions specialize, offering a limited variety of SICAVs. More than a third of the sample, sixty institutions, offer only one category of SICAV. Fifteen institutions offer all categories. The large number of 'zero' points is particularly welcome to study the degree of scope economies. This contrasts sharply with most banking studies which attempt to evaluate economies of scope with a sample of firms that offer the entire product range. 3The 1987 data set is described in Dennine and Roller (1992). 4A discussion of the grouping is available on request. In contrast with the US mutual fund industry, one cannot distinguish special purpose equity funds (such as incomeorientated, growth, or maximum capital gain). 'The fee covers the cost of promotion, auditing and legal fees, and the management fee. Separate data on the cost structure are not available.

6 4 3. The model The sample includes a cross section of 162 institutions operating in 1987 and We have excluded from the sample the institutions that have either been closed (ten firms), or have entered the industry (thirty five firms)6 after These institutions are all located in Paris and assumed to access the same technology, and face similar factor prices. Each institution is assumed to be a multiproduct firm that can sell up to five kinds of SICAVs. Output is defined as the total assets of each type. Since direct information on costs is not disclosed by these institutions, we use instead the total fee charged by each institution as our dependent variable. A potential draw back is that the total fee variable may not distinguish between variation in total cost across firms and higher fees due to different profit margins. In this paper, we attempt to distinguish these two effects by assuming that they are functionally separable, i.e. the fee is assumed to be the product of a cost function and a mark-up factor'. Furthermore, we allow the mark-up factor to vary across institutions by relating the profit margin to the age of each product. The rationale is the following. In the French mutual funds industry, consumers incur costs in switching from one bank offering its SICAVs to another bank. Switching costs could create a 'loyalty' that could be exploited overtime by an institution (Sharpe, 1992). The impact of age on product pricing is an assumption that will be tested empirically. The total fee charged by each institution is assumed to be the product of an age effect and a quadratic cost function, specified as follows, 6The institutions that have withdrawn were all small institutions belonging to the first or second quartile of firms. The new institutions essentially focused on the money market funds or bonds. 'Goldberg et al (1991) face a similar problem. Taking revenues as a measure of output in the securities industry, they assume that there is no systematic relationship between prices and the size or product mix of firms.

7 5 5 F(q,t) = [El C(q) i= = [ri Cal [00+E 13 1 qi EEPy qi4)] (1). i=1 i=1 1=1 i=1 where q=(mmf, FRB, FRE, REE, INT) is the output vector of the five mutual funds, F(q,t) is total fee, and t, is the age of a particular SICAV. This specification has the merit of simplicity. In particular, the age elasticities are constant, i.e. independent of quantities or time of creation of the other products. Moreover, the economies of scale or scope are independent of the profit margin factor. Although many banking studies have used a translog function specification, we use a quadratic cost function, because of the several limitations reported in the literature for the translog. Berger et al (1987) emphasized that the translog is not defined at zero output points, so the conclusions on economies of scope depend on the proximity of the zero output approximation. Roller (1990) argues that the translog is not a robust functional form because of its inherent flip-flop property. As one or more outputs approach zero, the estimated cost surface will either be zero or infinite, depending on the sign of the second-order parameter'. 4. Empirical Results Equation (1) is estimated as a sytem of two equations corresponding to the 1987 and 1989 data sets. Table 3 reports the constrained seemingly unrelated regression estimates where the quadratic cost component have been constrained to be equal in 1987 and 1989 g. The profit margin age-factor is left unconstrained because the competitive structure of the industry and the market power could have changed over time. The use of a SUR estimation 8Dermine and Willer (1992) demonstrate the superiority of the quadratic cost function to evaluate the economies of scope in the French mutual funds industry. 91n alternative estimations of equation (1), we have left unconstrained all the coefficients, and tested that the 1987 vector of costs coefficients is not statistically different from the 1989 vector. The restrictions are accepted at the 1 % level.

8 6 procedure is motivated by the correlation of 0.73 between the residuals of the two equations. The estimates of the coefficients of equation (1) are reported in Table 3. <Insert Table 3> Ray Scale Economies (RSCE) measure the behavior of cost when all outputs are expanded by the same proportion ; that is scale economies exist if and only if the sum of cost output elasticities is smaller than one. The RSCE are given by RSCE = v a In co. t a Inoo Table 4 reports the RSCE estimates evaluated at eight different points. Economies of scale can be identified for small institutions, but we are unable to identify scale economies for larger firms. This result is consistent with those obtained with the 1987 data set (Dermine and Roller, 1992), and robust as they hold whether an age variable is included or not. <Insert Table 4> Overall Economies of Scope (OEOS) exist if single output production leads to higher fee than multioutput production. The degree of OEOS is given by 5 E C(q,,O) - C(q) 1=1 C(q) The degree of OEOS is the percentage increase in total costs from specialized production. The results are reported in Table 4. We observe significant economies of scope at the eight points, indicating that joint production of SICAVs leads to lower costs. This is in contrast with our previous study which did not find significant economies of scope for large institutions. Thus, it seems that a more flexible specification like equation (1), allowing for dynamic time effect in a reduced form model, spreads some doubt on the result of no or low economies of scope in mutual funds.

9 7 The product-age elasticities, which measure the increase in total fees for an increase in the age of a SICAV, are reported in Table 5. We observe that the age elasticity of the money market funds is positive and statistically significant in 1987 and This result is consistent with the 'switching cost and customer loyalty' rationale. Institutions managing money market funds would charge a low fee to attract a customer base, raising it afterwards. However, a negative elasticity is observed for the age of SICAVs specialized in French Bonds. Hence, we can not find any empirical evidence consistent with the customer loyalty hypothesis for these funds. Similar inconclusive evidence is displayed by the remaining three funds, displaying either insignificant age elasticities (French Equities), or change in sign (Real Estate and International). Alternative hypotheses which would explain a negative age elasticity include the existence of significant learning effects or a skimming pricing strategy. The appropriate specification and testing of these alternative explanations are subject to further investigation. In conclusion, we find evidence consistent with the customer loyalty hypothesis for the dominant segment of the French Sicav industry, namely the money market funds. For other funds, however, we fmd no systematic such evidence. <Insert Table 5> 5. Conclusions This paper has provided additional evidence on economies of scale and scope in the French mutual funds industry. The very large growth observed from 1987 and 1989 has not altered our main conclusion that economies of scale exist for relatively small institutions, but are not significant for very large ones. Economies of scope appear to be significant for small and large firms. These results have important policy implications. The absence of economies of scale indicates that large firms are unlikely to drive out smaller firms, and that financial regulation which separate equity management, from bond or real estate management (functional regulation) cannot dismiss the possible existence of economies of scope in financial services. Furthermore, the age of the SICAVs is a statistically significant variable which affects substantially the estimates of scale and scope economies.

10 8 REFERENCES Berger, A. N., Hanweck, G.A. and Humphrey D.B. (1987) : "Competitive Viability in Banking", Journal of Monetary Economics, 20, Clarke J.A. (1988) : "Economies of Scale and Scope at Depository Financial Institutions : A Review of the Literature", Federal Reserve bank of Kansas City, Dermine J. and L.H. Roller (1992) : "Economies of Scale and Scope in French Mutual Funds, Journal of Financial Intermediation, 2, Ferris S.P. and Chance D.M. : "The Effects of Plan 12b-1 on Mutual Funds Expense Ratio : a Note", Journal of Finance 4, Gilbert R.A. : "Bank Market Structure and Competition, a Survey", Journal of Money, Credit and Banking, 16, Goldberg L.G., Hanweck G.A., Keenan M., and Young A. (1991) :"Economies of Scale and Scope in the Securities Industry", Journal of Banking and Finance, 15, Kolari and Zardkoohi (1987) : "Bank Costs, Structure and Performance". Lexington Books, Lexington. Roller L.H. (1990) : "Modelling Cost Structure : The Bell System Revisited". Applied Economics, 22, Sharpe S.A. (1992) : "Consumer Switching Costs, Market Structure and Prices : The Theory and its Application in the Bank Deposit Market", Finance and Economic Discussion Series, Board of Governors, Washington D.C.

11 TABLE 1 Size and Growth of the 162 Institutions Offering SICAV in 1989 (FF Millions) Quintiles Mean Minimum Maximum Growth from 1987 to 1989 (%) , Note : The five largest institutions control 57 % of the market : Credit Agricole (18.2 %), Caisse des Depots et Consignations (11.1 %), Banque Nationale de Paris (11.1 %), Credit Lyonnais (9.5 %), and Societe Generale (6.5 %). Source : TGF, Paris, February 1989.

12 TABLE 2 The 1989 SICAV Market In France Size of SICAVs (FF million) Age (days since creation) Fee per SICAV (percentage of assets) Market Number of Market size Mean Min Max Mean Min. Max. Mean Min. Max. SICAVs (FF billion) Money market funds French bonds French equities , , , , , , Real Estate , , Internation al , , Note. Source : TGF, Paris, February 1989.

13 TABLE 3 CONSTRAINED SUR PARAMETER ESTIMATES OF EQUATION I Explanatory variable' Parameter estimate t stat. Constant MMF FRB FRE REE INT MMF X MW 2.2E MMF X FRB 1.76E MMF X FRE -2.84E MW X REE 6.71E MMF X INT 1.01E FRB X FRB -1.73E FRB X FRE -I.06E FRB X REE FRB X INT -3.32E FRE X FRE 2.13E FRE X REE FRE X INT 3.63E REE X REE REE X INT -9.82E INT X INT 4.69E AGEMMF AGEFRB AGEFRE AGEREE AGEINT AGEMMF AGEFRB AGEFRE AGEREE AGEINT R Money Market Funds (MW), French bonds (FRB), French equities (FRE), real estate (REE), and international (INT).

14 TABLE 4 Ray Scale (RSCE), and Overall Scope (OEOS) Economies Institution Size Ray Scale Economies Overall Economies (FF million) of Scope 100 < assets < (25.73)b (3.74') 300 < assets < (7.08) (3.74) 700 < assets < (0.77) (3.7) 1300 < assets < (0.09) (3.2) 2100 < assets < (0.08) (3.20) 3600 < assets < (0.54) (2.99) 6700 < assets < (2.23) (2.75) Five largest (0.8) (1.83) RSCE and OEOS are evaluated at the mean of each of the five groups. b t statistics in parentheses refer to significance test around 1. t statistics in parentheses refer to significance test around 0.

15 TABLE 5 Product-Age Elasticities Sicav Age Elasticity Age Elasticity (1987) (1989) Money Market Funds (2.97)a (2.86) French Bonds (-2.91) (-1.93) French Equity (-0.48) (0.46) Real Estate (0.46) (-2.27) International (2.8) (0.31) at statistics in parenthesis.