Journal of Financial Services Research

, Volume 43, Issue 1, pp 99–126

Deconstructing Herding: Evidence from Pension Fund Investment Behavior

Authors

  • Claudio Raddatz
    • Banco Central de Chile
    • World Bank
Article

DOI: 10.1007/s10693-012-0155-x

Cite this article as:
Raddatz, C. & Schmukler, S.L. J Financ Serv Res (2013) 43: 99. doi:10.1007/s10693-012-0155-x
  • 466 Views

Abstract

In this paper, we examine herding across asset classes and industry levels. We also study what incentives managers at various layers of the financial industry face when investing. To do so, we use unique and detailed monthly portfolios between 1996 and 2005 from pension funds in Chile, a pioneer in pension-fund reform. The results show that pension funds herd more in assets that have more risk and for which pension funds have less market information. Furthermore, the results show that herding is more prevalent for funds that narrowly compete with each other, namely, when comparing funds of the same type across pension fund administrators (PFAs). There is much less herding across PFAs as a whole and in individual pension funds within PFAs. These herding patterns are consistent with incentives for managers to be close to industry benchmarks, and might be also driven by market forces and partly by regulation.

Keywords

Institutional investorsPortfolio allocationInvestment patternsCapital market development

JEL Classification

G11G12G23G28O16

1 Introduction

This paper uses a unique and rich micro data set of pension funds to shed new light on how incentives might affect institutional investors in their portfolio allocation decisions. In particular, we study two aspects discussed in the literature but still relatively unexplored: (i) how institutional investors trade in different types of assets, and (ii) what incentives managers at various levels of the financial industry face when implementing their investment strategies. We focus on herding statistics because they are widely used and are important measures to analyze. Among other things, herding can contribute to market volatility and might mean that managers are not generating independent assessments and not providing distinct services to the underlying investors. The analysis of herding also helps to more broadly understand the role of institutional investors in capital market activity, such as capital raising and trading. Furthermore, the paper offers new evidence on the importance of several factors linked to the behavior of institutional investors; such as information, liquidity, incentives related to organizational aspects of the financial industry, and the regulatory framework in which managers operate.

Institutional investors are interesting to analyze not only because they have become very large, but also because detailed asset-level portfolios over time (unavailable at the household or retail-investor level) are sometimes accessible. In particular, institutional investors are increasingly relevant for both asset management and the development of financial systems. In fact, they are likely to be among the most important conduits of private and public savings, intermediating funds and supplying capital for firms and countries to grow. As institutional investors became prevalent, the research on how they invest flourished.1

In this paper, we exploit new data and analyze the investment behavior of pension funds for which relatively little is known, although they have played a crucial role across countries.2 In particular, we use data from Chile that was the first country to embrace the new mandatory, privately managed, defined-contribution (DC) pension fund model by replacing the public, defined-benefit (DB) pension system in 1981. Many developed and developing countries have followed suit and reformed their pension regimes in a similar way (e.g., Argentina, Colombia, Hungary, Lithuania, Mexico, Peru, Slovakia, Sweden, Poland, and the UK).

The data we assemble contain the detailed portfolios of the universe of Chilean pension funds in all types of securities and asset classes at a monthly frequency for a 10-year period from 1996 to 2005. We also compile the monthly returns of each instrument included in these portfolios. The data set contains 3,869,290 observations with information on the holdings and returns of 24,322 different securities for up to 57 pension funds. We then compute different estimates of herding that have an association with funds buying or selling the same assets simultaneously.

This unique data set allows us to shed new light on a series of questions related to different aspects of pension fund investment strategies and their overall behavior. In particular, do pension funds herd by buying or selling the same assets simultaneously? Is their herding pattern different across asset classes with varying liquidity and information? Does their herding behavior vary at different levels of the pension fund organizational structure? What does their behavior tell about the incentives that managers face to compete with each other? Does herding differ on the buying and selling sides? Does it change with purchases in primary and secondary markets? Is trading activity associated with variations in returns?

By addressing these questions, we investigate at least three important aspects of herding discussed in the literature. First, traders might copy other traders in the process of extracting private information (Shiller and Pound 1989; Scharfstein and Stein 1990; Banerjee 1992; Bikhchandani et al. 1992). Because some assets are more opaque than others, the degree of herding should decrease with the transparency of the assets for a given level of risk. In other words, securities for which information is widely available and that entail less risk are less likely to induce herding patterns. Thus, we use our data set to analyze herding by institutional investors in the different asset types. In particular, we calculate herding in corporate bonds, financial-institution bonds, government bonds, mortgage bonds, and equity.

Second, herding might also be explained by managers following similar trading strategies like momentum (Froot et al. 1992; Gompers and Metrick 2001). Momentum trading, defined as the purchasing (selling) of assets whose returns are positive (negative), is a popular investment strategy. Its presence among US institutional investors (primarily mutual funds) has been widely documented in the literature (e.g., Grinblatt et al. 1995; Nofsinger and Sias 1999; Grinblatt and Keloharju 2000; Kaminsky et al. 2004; Sias 2004, 2007; Greenwood and Nagel 2009). Our data on returns allow us to test whether, to the extent that there is herding, it is driven by momentum strategies.

Third, managers might herd as a way to reduce risk. Although traditional theories of asset allocation focus on the problem of an isolated investor maximizing wealth or consumption, several papers study the incentive schemes that arise in the context of financial intermediation. In particular, the conflicts of interest between fund managers and the underlying investors can affect risk taking (Scharfstein and Stein 1990; Shleifer and Vishny 1990; Chevalier and Ellison 1999; Graham 1999; Stein 2003, 2005; Kapur and Timmermann 2005; Bolton et al. 2006). The underlying investors, the regulator, and the asset management companies monitor managers on a short-run basis to reduce principal-agent problems, generating incentives for managers to be averse to investments that (though potentially profitable) are different from those held by their competitors. Namely, deviating from the pack might entail costs because the principal cannot evaluate whether the agent deviated for good reasons. Herding behavior is thus a natural response by managers to avoid penalties. The data we assemble allow us to study herding at different levels of the pension fund industry. In particular, we study herding among pension fund administrators (PFAs), individual pension funds across the entire industry, individual pension funds within PFAs, and similar types of pension funds across PFAs. We expect herding to increase for fund managers of similar types of pension funds, as comparisons are easier to make and competition intensifies.

The main results from this paper can be summarized as follows. First, pension funds tend to herd; that is, they buy or sell the same assets at the same time.

Second, herding varies substantially across asset classes. In particular, herding is more pronounced in corporate bonds and financial-institution bonds (which are similar to corporate bonds), while there is less herding in equity and mortgage bonds. Relative to corporate bonds, investors know equity better (few, large companies dominate the trading activity). And equity tends to be traded frequently in secondary markets, sending continuous price signals. Instead, corporate bonds trade infrequently and part of their trading occurs over the counter (OTC).3 Mortgage bonds are safer than corporate bonds because they are backed by real estate. The asset class in which the least herding occurs is government bonds, which have the lowest risk. Across different levels of the industry, pension funds tend to herd more in the asset classes that are more opaque. This result is consistent with pension funds trying to copy each other in their portfolio decisions, especially for the assets for which there is less information from the markets.

Third, herding is the most intense when comparing funds of the same type across PFAs. That is, herding peaks as funds narrowly compete with each other across PFAs to retain pensioners and/or avoid market or regulatory punishment. PFAs as a whole also herd but less intensively, because administrators are not so narrowly compared with each other. The least intense herding occurs among funds within PFAs, where competition is less prevalent as the incentives are for PFAs to keep pensioners in any of the funds that the PFA has.

Fourth, we do not find evidence that momentum trading is the main cause of the herding observed in domestic assets.

Fifth, although the patterns in this paper might be influenced by the regulation that induces funds across PFAs to compare with each other, the investment decisions of fund managers cannot be neglected because there is no specific mandate for pension funds to trade in specific securities and herding does not decrease when regulations are relaxed. Moreover, the behavior does not seem to be explained by the lack of investable instruments because pension funds do not invest in all of the available and pre-approved assets.

The rest of the paper is structured as follows. Section 2 briefly summarizes the case of Chile and its pension fund system. Section 3 describes the data and some basic turnover statistics. Section 4 studies different turnover measures. Section 5 explores what other factors might be related to herding behavior. Section 6 concludes.

2 Chile’s pension fund system

Chile is a good natural case study to analyze in depth the behavior of pension funds and, more broadly, institutional investors. In 1980, Chile decided to reform its pension fund system and replaced over time the pay-as-you-go scheme with a fully funded capitalization one based on individual accounts operated by private pension fund administrators (PFAs). Under the new system, pensioners choose their PFA and the specific funds in which they invest. Pensioners can switch their PFA at any point in time and reallocate their investments across funds. Since 2002, the multi-fund period begins, when each PFA offers five different funds (Funds A–E), with Fund A (Fund E) being the most (least) risky.

The mandate of each pension fund is to provide the highest possible returns to pensioners given the set of risk parameters and investment regulations. For example, there are no restrictions on the amount and type of trading activity. Pension fund managers do not have liabilities with the pensioners; they simply manage their assets. Pension funds are subject to a minimum return regulation that establishes that the PFAs are responsible for ensuring an average real rate of return over the preceding 36 months (12 months before October 1999).

The portfolio of each fund is managed by managers that compete on a frequent basis to obtain the best possible returns. PFAs hire the managers and might fire them if their performance lags behind. Within a PFA, pension fund portfolios are managed separately, but the PFA provides market analysis to all its funds.

Over time, pension fund administrators have grown substantially and have become the largest institutional investors in Chile. As a share of GDP, assets managed by pension funds increased by 1.85 times, from 38 % in 1996 to 71 % in 2005. Figure 1 shows the evolution of pension system holdings as a share of GDP. Appendix 1 provides more information about the regulations and evolution of the system.
https://static-content.springer.com/image/art%3A10.1007%2Fs10693-012-0155-x/MediaObjects/10693_2012_155_Fig1_HTML.gif
Fig. 1

Pension System Holdings. This figure shows the size of the total assets of pension funds across all PFAs relative to Chile’s GDP, by fund type, for the entire sample period (July 1996 to December 2005). The fund types reflect different risk profiles, from the riskiest fund (Fund A) to the most conservative fund (Fund E). The nominal values are computed for December of each year and scaled by GDP

3 Data and turnover statistics

The data set used in this paper comes from Chile’s Superintendency of Pensions (Superintendencia de Pensiones, SP). It consists of a monthly panel of all the portfolio investments of PFAs in operation and for each of their funds during the period of July 1996 to December 2005. However, because of the richness of the data, a large part of the analysis in this paper focuses on the multi-fund period (2002–2005). The data set has information on the price and quantity for every security held by a fund per unit of time. We define a fund as a pair PFA/fund type (e.g., Fund C of PFA Aporta configures a single fund). After cleaning the data set, we use 3,869,290 observations that represent all domestic fixed-income securities and domestic equity held during each month by at least one fund and that contain information on the holdings of 24,322 different securities for up to 57 funds. These securities are divided into 20 different instrument types. We group all the instrument types into five general asset classes: corporate bonds, financial-institution bonds, government bonds, mortgage bonds, and equity.4, 5 Table 1 displays the average portfolio holding in each class by type of fund.
Table 1

PFA Holdings by Asset Class and Fund Type. This table presents the average portfolio share of each asset class by fund type across PFAs. The sample periods are July 1996 to December 2005 for Fund C, May 2000 to December 2005 for Fund E, and September 2002 to December 2005 for Funds A, B, and D. Portfolio shares are calculated considering only the asset classes shown in the table. Derivatives, investment and mutual funds quotas, former pension system bonds, deposits, and foreign assets are excluded from the analysis. Portfolio weights of each asset class are calculated per PFA and fund type for each month, and averaged over time. For Fund E, no investments are allowed in equity

Average PFA Portfolio Share by Asset Class and Fund Type (1996–2005)

 

Fund Type

Fund A

Fund B

Fund C

Fund D

Fund E

Domestic Assets

     

  Corporate Bonds

4.6 %

9.5 %

9.1 %

13.2 %

12.3 %

  Financial-Institution Bonds

1.3 %

2.9 %

3.4 %

3.3 %

3.4 %

  Government Bonds

14.1 %

29.4 %

43.3 %

47.9 %

61.6 %

  Mortgage Bonds

5.2 %

13.6 %

19.8 %

19.0 %

24.0 %

  Equity

74.8 %

44.6 %

24.3 %

16.6 %

The securities analyzed in this paper vary across different dimensions associated with the availability of market information on issuing companies and the availability of quoted and realized market prices for institutional investors. Table 2 shows characteristics for the multi-fund period of issuances, trading activity, and size of issuers for some of the asset classes analyzed in this paper: corporate bonds (including those issued by financial institutions, which tend to be similar to corporate bonds), government bonds, and equity.6
Table 2

Evolution of Chilean Equity and Fixed-Income Markets. This table shows the evolution of the issuances and turnover of Chilean equity and fixed-income markets from 2002 to 2005. Panel A presents the total amount issued, the number of companies, and the median amount issued by company during each year for corporate bonds, government bonds, and equity. Panel B presents the turnover ratio for each asset class as the annual value traded divided by end of the year market value. The turnover ratio for government and corporate bonds is obtained from Eterovic et al. (2011). The turnover ratio for equity is obtained from the World Development Indicators and it is adjusted by free-float market capitalization using Dahlquist et al. (2003). The corporate bonds categories in Panels A and B comprises both financial-institution bonds and corporate bonds. Panel C presents the trading frequency for each asset class as the average percentage of trading days in which instruments are traded. Panel D presents the median amount of assets for the main companies, considering the 40 listed companies that compose the main Chilean stock market index (IPSA) for equity, and the biggest 40 companies that have outstanding corporate bonds

Panel A. Issuances

  

2002

2003

2004

2005

Corporate Bonds

     

  Total Amount Issued

(Millions of US Dollars)

6,064

7,693

11,206

11,560

  Number of Companies

 

36

40

54

53

  Median Amount per Company

(Millions of US Dollars)

95.0

114.1

150.0

175.3

Government Bonds

     

  Total Amount Issued

(Millions of US Dollars)

2,087

4,238

4,981

4,967

  Number of Issuances

 

80

232

263

294

  Median Amount per Issue

(Millions of US Dollars)

4.0

2.4

4.0

5.0

Equity

     

  Total Amount Issued

(Millions of US Dollars)

318

44

654

1,126

  Number of Companies

 

14

2

4

7

  Median Amount per Company

(Millions of US Dollars)

2.6

22.0

165.1

155.0

Panel B. Turnover Ratio

  

2002

2003

2004

2005

Corporate Bonds

 

32 %

34 %

38 %

40 %

Government Bonds

  

420 %

460 %

290 %

Equity

 

38 %

52 %

62 %

76 %

Panel C. Trading Frequency (Percentage of Trading Days)

  

2002

2003

2004

2005

Corporate Bonds

 

46 %

Government Bonds

 

100 %

Equity

 

92 %

90 %

92 %

95 %

Panel D. Size of Main Companies Issuing Equity and Corporate Bonds (Millions of US Dollars)

  

2002

2003

2004

2005

Corporate Bonds

 

1,008

1,118

1,110

1,696

Equity

 

979

998

1,367

1,888

During the sample period, issuance per year is highest for corporate bonds, followed by government bonds, and lastly equity (Panel A). This is expected because many companies issue bonds, and they have to continue issuing them over time as bonds mature and firms seek refinancing. But the amount issued in corporate bonds per company is much smaller than the total amount issued by the government, and similar to the equity issued per company.7 Panel B shows data on turnover ratios (annual value traded divided by end-of-the-year market capitalization) across these three asset classes. Clearly, government bonds are the asset class with the highest turnover, followed by equity, and corporate bonds. Government bonds and equities not only have higher turnover than corporate bonds, but also are more frequently traded in open exchanges (Panel C). For instance, equities from the 40 listed companies that comprise the main Chilean stock market index (IPSA) traded on 92 % of the trading days in 2004. Government bonds with maturities of between 8to 10 years also traded almost every day. In contrast, corporate bonds with the same maturities traded 46 % of the time during that year.8 Finally, companies listed in the stock exchange are typically larger than those that issue corporate bonds. Panel D compares the median size of the main listed companies and of those companies that have issued corporate bonds. It shows that, despite corporate bond issuers being relatively large in Chile, they are typically smaller than the main companies listed on the stock exchange (median assets of US$ 750 million versus US$ 1,900 million for listed firms in 2005). Even the 40 largest companies with corporate bonds outstanding during 2002 to 2005 were smaller than the 40 main listed companies included in the IPSA (median assets US$ 1,700 versus US$ 1,900 million in 2005).

In summary, the data show that government bonds are widely available, frequently traded, and have easily available price information. Equity markets are dominated by large corporations, whose stocks trade frequently in open exchanges. Corporations issuing corporate bonds are smaller than those that issue equity; some issuances are large, but they are infrequently traded, and a nontrivial part of this activity occurs over the counter. These differences suggest that corporate bonds are probably the most opaque of the Chilean asset classes, followed by equity, and finally government bonds.

To complement the analysis, we display here some basic measures of turnover or trading activity on different types of securities by pension funds. Turnover generally has a relation to market liquidity, which is vital for the emergence of new instruments, capital raising activity, and the functioning of secondary markets. More trading reduces the cost of immediate execution, which lowers bid-ask spreads and reduces the firm’s opportunity cost of capital.9

Panel A of Table 3 shows that pension funds tend to trade infrequently. In particular, Panel A shows what fraction of its assets a given PFA trades at any point in time. The table presents two simple statistics: the number of total assets traded by a PFA in a given period relative to the total number of holdings in the PFA’s overall portfolio (column 1) and the value of the aggregate portfolio that experiences some activity in a given month (column 2), both averaged over time.10 On average, a PFA trades only 15.6 % of its assets and the monthly changes in positions in those assets correspond to just 3.4 % of the initial total value of the PFA’s assets. This low number contrasts with the 88 % of the mean turnover ratio found in Kacperczyk et al. (2008) for a sample of 2,543 actively managed US equity mutual funds between 1984 and 2003, suggesting that Chilean PFAs are rather passive in their trading behavior. There is important variation across asset classes in the degree of PFA trading activity. The most traded assets are equities and mortgage bonds. On the other hand, there is a low degree of trading in corporate and financial-institution bonds.11
Table 3

Pension Fund Trading Activity Measures. Panel A presents trading statistics using data from the multi-fund period (2002 to 2005). Column (1) presents the average percentage of assets traded each month by PFAs as a share of the total number of assets held in their portfolios. Column (2) presents the average across PFAs of the difference in weights (contemporaneous weights minus lagged weights using lagged prices for both) for the traded portfolio, calculated at the PFA level. Columns (3) presents the average percentage of assets traded by funds as a share of the total number of assets held in their portfolios. Column (4) reports the average across funds of the difference in weights for the traded portfolio, calculated at the PFA-fund level. Panel B presents the average proportion of units of a given security that a PFA incorporates into its portfolio in its first purchase and the proportion of units of the security that a PFA liquidates at the security’s maturity date. Both measures are relative to the maximum number of units of that security that the PFA holds in its portfolio at any time; they are calculated at the PFA level and the PFAfund level across all instruments for each asset class and averaged across PFAs or PFA-funds. For both ratios, the average and standard deviations are presented for each asset class

Panel A. Monthly Average Percentage of Assets Traded by PFAs

 

Trading Statistics

PFA Level

PFA-Fund Level

Percentage of Assets Traded Relative to Assets Held

Share of Traded Portfolio

Percentage of Assets Traded Relative to Assets Held

Share of Traded Portfolio

(1)

 

(2)

 

(3)

 

(4)

 

All Asset Classes

15.6 %

 

3.4 %

 

17.4 %

 

3.7 %

 

Corporate Bonds

12.8 %

 

0.2 %

 

13.3 %

 

0.3 %

 

Financial-Institution Bonds

11.4 %

 

0.1 %

 

12.6 %

 

0.1 %

 

Government Bonds

12.4 %

 

1.7 %

 

13.6 %

 

1.7 %

 

Mortgage Bonds

16.2 %

 

0.3 %

 

18.0 %

 

0.4 %

 

Equity

41.9 %

 

1.1 %

 

35.8 %

 

1.3 %

 

Panel B. Proportion of Fixed-Income Instruments Bought and Held Until Expiration

 

PFA Level

PFA-Fund Level

Ratio of Units at First Purchase to Maximum Units in Portfolio

Ratio of Units at Expiration to Maximum Units in Portfolio

Ratio of Units at First Purchase to Maximum Units in Portfolio

Ratio of Units at Expiration to Maximum Units in Portfolio

Average

Standard Deviation

Average

Standard Deviation

Average

Standard Deviation

Average

Standard Deviation

PFA level

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Corporate Bonds

0.91

0.13

0.91

0.09

0.87

0.14

0.87

0.17

Financial-Institution Bonds

0.90

0.10

0.94

0.04

0.87

0.13

0.90

0.10

Government Bonds

0.66

0.23

0.85

0.10

0.61

0.21

0.89

0.07

Mortgage Bonds

0.86

0.03

0.68

0.12

0.84

0.10

0.71

0.13

An alternative way to gauge the extent to which managers are actively trading their portfolios is to focus on fixed-income instruments (which are also of fixed term). The useful feature of these assets is that they do not need to be traded to recover the initial investment, as managers can wait until maturity and collect coupons in the meantime. Panel B of Table 3 presents two statistics per asset class: (i) the average proportion of units of a given security that a PFA incorporates into its portfolio in its first purchase, and (ii) the proportion of units of that security that a PFA liquidates at the security’s maturity date. Both measures are relative to the maximum number of units of that security that the PFA holds in its portfolio at any time. They show that PFAs tend to purchase most of their fixed-income assets at once and liquidate most of them upon maturity, not before. That is, although pension funds might hold a large fraction of the outstanding securities, they trade a small fraction of them in secondary markets. This buy-and-hold behavior is common in this type of institutional investor, although it runs contrary to the idea that pension funds provide liquidity to secondary markets. Nonetheless, even in fixed-income assets, pension funds still trade between 5 and 10 % of their holdings over the lifetime of the asset.

4 Do pension funds herd?

To formally test for the presence of herding, we compute different estimates of herding. These measures focus on whether funds simultaneously buy or sell the same assets at the same time. We measure the degree of herding using the approach by Lakonishok et al. (1992) in which, when there is no herding, the probability of buying has to be equal among the assets being traded. Therefore, a measure of the difference between the probabilities of buying across assets provides a test of the hypothesis of no herding.

In particular, Lakonishok et al. (1992) define the herding statistic H(i,t) as:
$$ H\left( {i,t} \right)=\left| {\frac{{B\left( {i,t} \right)}}{{N\left( {i,t} \right)}}-p(t)} \right|-AF\left( {i,t} \right), $$
(1)
where p(t) is the probability of buying any asset at time t, B(i,t) is the number of funds that increase their holdings of asset i at time t (buyers), S(i,t) is the number of sellers of asset i at time t, and \( N\left( {i,t} \right)=B\left( {i,t} \right)+S\left( {i,t} \right) \) the number of funds active on asset i at time t (i.e., either buying or selling), and AF(i,t) is an adjustment factor. To calculate the herding statistic, we identify a purchase (sale) as an increase (decrease) in the number of units of a given asset held by a PFA.

Under the hypothesis that no herding occurs, the number of buyers B(i,t) follows a binomial distribution with parameters p(t) and N(i,t), and the adjustment factor AF(i,t) is the expected value of the first term on the right-hand side of Eq. (1) under this hypothesis, which is positive because of the use of the absolute value. Therefore, if no herding occurs we should be unable to reject the null hypothesis that the herding statistic has a zero mean.

The adjustment factor AF(i,t) is:
$$ AF\left( {i,t} \right)=E\left( {\left| {p\left( {i,t} \right)-E\left[ {p\left( {i,t} \right)} \right]} \right|} \right), $$
(2)
where p(i,t) is the probability of buying an asset i at time t. The proportion of all funds that buy during period t is used as a proxy for E[ p(i,t)]. And due to the assumption that the number of buyers in each period follows a binomial distribution, then AF(i,t) can be calculated as:
$$ AF\left( {i,t} \right)=\sum\nolimits_{j=0}^{{N\left( {i,t} \right)}} {\left\{ {\left( {_j^{{N\left( {i,t} \right)}}} \right){{{\left[ {p(t)} \right]}}^j}{{{\left[ {1-p(t)} \right]}}^{{N\left( {i,t} \right)-j}}}\left| {\frac{j}{{N\left( {i,t} \right)}}-p(t)} \right|} \right\}}, $$
(3)
which can be further simplified in order to carry out the calculations.

As explained earlier, our data have information on the detailed portfolios of all pension funds managed by all of the PFAs. Furthermore, we know which PFA manages each of the funds. We use this information to test for the presence of herding at four levels of aggregation. First, we test for herding at the PFA level (aggregating all funds managed by a PFA in a single portfolio). This neglects within-PFA herding and only considers herding among administrators. Second, we also test for herding at the PFA-fund level that considers herding both within each administrator and across all administrators. Two or more funds within a PFA or across PFAs buying the same asset equally contribute to this herding statistic. Third, we consider herding at the within-PFA level, which only looks at whether funds managed by the same PFA tend to buy or sell the same assets together. Finally, we test for the presence of herding across PFAs, but within a given fund type. Only funds of the same type (from A to E) trading the same assets count for the computation of this statistic. Testing for herding at these different levels of aggregation provides valuable information on the determinants of herding and the incentives that managers have to engage in this behavior.

Table 4 reports the herding results at the PFA level with each entry displaying the mean of the herding statistic for each asset class and its corresponding standard error by using an asset-class-specific probability of buying an asset. Column (1) presents the results obtained computing the statistic across all assets traded by more than one PFA. To show the robustness of the results to different estimates of herding, columns (2) and (3) report the herding statistics computed over those assets traded by more than two or three PFAs. Column (4) reports the average asset-specific probabilities of buying an asset for each asset class (p(t)). For example, the average probability of buying instruments from domestic financial institutions (conditional on trading them) is 51 % and the average probability of buying mortgage bonds is 13 %.
Table 4

Herding at the PFA Level. This table presents the average Lakonishok et al. (1992) herding statistic by asset class at the PFA level considering each PFA as an individual entity. The herding statistic is calculated using the asset-specific probability of buying an asset at any point in time. The herding statistic over all asset classes is calculated based on the average portfolio share of each asset class. Columns (1), (2), and (3) present the results considering assets traded by more than one, more than two, and more than three PFAs, respectively. Numbers represent percentages (results are multiplied by 100). Column (4) presents the average asset-specific probability of buying an asset, calculated over the assets traded by more than one PFA, by asset class. T-tests are one-tailed

 

Herding Statistic

 

Assets Traded by More than One PFA

Assets Traded by More than Two PFAs

Assets Traded by More than Three PFAs

Average Probability of Buying an Asset

(1)

(2)

(3)

(4)

All Asset Classes

0.90***

2.41***

3.84***

49.05 %

 

(0.29)

(0.41)

(0.47)

 

Domestic Assets

    

  Corporate Bonds

3.10***

10.24***

13.78***

51.61 %

 

(0.64)

(0.92)

(0.06)

 

  Financial-Institution Bonds

6.16***

10.31***

9.21***

51.27 %

 

(0.92)

(1.38)

(1.81)

 

  Government Bonds

−2.11

0.79***

3.82***

64.58 %

 

(0.16)

(0.25)

(0.46)

 

  Mortgage Bonds

4.58***

2.21***

1.20***

12.66 %

 

(0.07)

(0.06)

(0.06)

 

  Equity

1.46***

1.94***

2.44***

53.44 %

 

(0.24)

(0.27)

(0.32)

 

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors are presented in parentheses

The results in Table 4 show that there is robust evidence of herding, both overall and across asset classes. Except for government bonds traded by more than one PFA, there are positive and statistically significant coefficients regardless of the number of PFAs trading a given asset. The results also show significant differences in the coefficients of herding across asset classes within each column. Herding seems to be stronger for corporate bonds and financial-institution bonds. This ranking of herding across asset classes closely resembles the differences in market transparency of different asset classes documented in Section 3. As shown in Table 2, while Chilean corporate bonds are typically issued by relatively large companies, they are much less frequently traded than equities and government bonds, and part of these trades occur in more opaque, over-the-counter markets rather than in open exchanges.12

Except in the case of mortgage bonds, the different columns show that the prevalence of herding increases as the number of PFAs trading an asset increases (columns (1) to (3)). When focusing on column (3), on those assets traded by more than half of the active PFAs, we find significant evidence of herding for all asset classes. The economic magnitude of the herding statistic is close to the evidence reported for mutual funds in developed countries in the literature, but still significantly higher in some asset classes when considering instruments traded by most PFAs (column (3)). As an example, herding in corporate bonds is 14 % when considering assets traded by more than three PFAs, up from 10 % when considering assets traded by more than two PFAs, and up from 3 % when considering assets traded by more than one PFA. In the case of mortgage bonds, we find less herding for the measures that consider bonds traded by more PFAs. This result is expected because the number of specific mortgage bonds in the markets is very large but each bond is small. Therefore, the probability of a mortgage bond being traded by more than two PFAs is small.

Overall, the results indicate that the presence of herding among Chilean PFAs in many asset classes increases as more PFAs trade an asset. In other words, although PFAs hold few assets, when various PFAs are active they tend to be on the same side of the trade.

Table 5 reports similar herding estimates to those in Table 4 (i.e., at the PFA level) but constrains the sample to the multi-fund period, 2002–2005, in which more funds become available. The results show that herding is still prevalent among corporate bonds and financial-institution bonds but significantly less so in other asset classes, except for a couple of instances for mortgage bonds and government bonds. Again, as more PFAs trade assets the herding statistics increase. The differences in results between Tables 4 and 5 suggest that part of the herding might be driven by competition between pension funds, not PFAs, since herding is stronger when including the period for which only one or two funds per PFA are available (Table 4).13 As the number of funds within PFAs increases, the degree of herding across asset classes diminishes.
Table 5

Herding at the PFA Level-Multi-Fund Period. This table presents the average Lakonishok et al. (1992) herding statistic by asset class at the PFA level by using data only from a multi-fund period (2002 to 2005). Each PFA is considered like an individual entity. The herding statistic is calculated using the asset-specific probability of buying an asset at any point in time. The herding statistic over all asset classes is calculated based on the average portfolio share of each asset class. Columns (1), (2), and (3) present the results considering assets traded by more than one, more than two, and more than three PFAs, respectively. Numbers represent percentages (results are multiplied by 100). Column (4) presents the average asset-specific probability of buying an asset, calculated over the assets traded by more than one PFA, by asset class. T-tests are one-tailed

 

Herding Statistic

 

Assets Traded by More than One PFA

Assets Traded by More than Two PFAs

Assets Traded by More than Three PFAs

Average Probability of Buying an Asset

(1)

(2)

(3)

(4)

All Asset Classes

−1.01

2.00***

4.02***

45.65 %

 

(0.47)

(0.71)

(0.77)

 

Domestic Assets

    

  Corporate Bonds

1.65**

12.52***

20.55***

51.32 %

 

(0.79)

(1.33)

(0.06)

 

  Financial-Institution Bonds

7.49***

13.17***

11.46***

33.21 %

 

(1.18)

(1.77)

(2.48)

 

  Government Bonds

−5.06

−0.83

1.88**

55.44 %

 

(0.29)

(0.44)

(0.86)

 

  Mortgage Bonds

1.06***

−0.63

−0.81

3.94 %

 

(0.08)

(0.05)

(0.05)

 

  Equity

0.34

0.42

0.49

57.54 %

 

(0.41)

(0.43)

(0.50)

 

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors are presented in parentheses

Given that trading at the fund level seems to explain part of the herding, Table 6 shows herding statistics using all funds across PFAs, without distinguishing the PFA or type of each fund. That is, this herding measure is computed at the most disaggregate level by taking into account the within-PFA and across-PFA variation across any type of fund. The results in Table 6 show again that herding is more prevalent in corporate bonds and financial-institution bonds. However, the point estimates are noticeably smaller than those in Table 5. For example, in the case of corporate bonds traded by more than three PFAs, the herding statistic in Table 6 is 4.58, while in Table 5 it is 20.55. The only result that does not follow this pattern is the degree of herding in equities for which coefficients become statistically significant in Table 6. In other words, part of the herding in equities is explained by the fund-level behavior. Because different fund types face different regulatory limits on their portfolio allocations, it is not surprising that the herding statistic is lower when considering the trades conducted by different fund types (given that they invest in different asset classes).
Table 6

Herding at the PFA-Fund Level. This table presents the average Lakonishok et al. (1992) herding statistic by asset class at the PFA-fund level by using data from a multi-fund period (2002 to 2005). Each fund in each PFA is considered like an individual entity. The herding statistic is calculated using the asset-specific probability of buying an asset at any point in time. The herding statistic over all asset classes is calculated based on the average portfolio share of each asset class. Columns (1), (2), and (3) present the results considering assets traded by more than one, more than two, and more than three funds, respectively. Numbers represent percentages (results are multiplied by 100). Column (4) presents the average asset-specific probability of buying an asset, calculated over the assets traded by more than one fund, by asset class. T-tests are one-tailed

 

Herding Statistic

 

Assets Traded by More than One Fund

Assets Traded by More than Two Funds

Assets Traded by More than Three Funds

Average Probability of Buying an Asset

(1)

(2)

(3)

(4)

All Asset Classes

−1.46

0.63*

1.48***

52.83 %

 

(0.31)

(0.37)

(0.36)

 

Domestic Assets

    

  Corporate Bonds

−0.96

2.46***

4.58***

54.95 %

 

(0.47)

(0.58)

(0.07)

 

  Financial-Institution Bonds

1.42**

6.09***

8.37***

43.97 %

 

(0.76)

(1.03)

(1.23)

 

  Government Bonds

−4.56

−0.97

0.22

57.65 %

 

(0.19)

(0.25)

(0.32)

 

  Mortgage Bonds

0.18**

−0.17

−0.11

9.29 %

 

(0.08)

(0.07)

(0.07)

 

  Equity

0.50**

1.15***

1.33***

54.44 %

 

(0.29)

(0.28)

(0.29)

 

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors are presented in parentheses

These differences in regulatory constraints cannot fully account for the observed decline in herding because they only restrict the composition of a fund’s portfolio across asset classes. Whereas these constraints could reduce the degree of overall herding computed by pooling all asset classes, Table 6 shows that the decline in herding occurs within each asset class. In sum, while the results in Table 6 suggest that fund-level herding is important, they leave unanswered the question of how funds specifically interact with each other in their trading and herding activity.

Table 7 reports the results from herding among funds within PFAs. As above, the comparison within asset classes eases concerns about the different compositions of the portfolios of different types of funds. While there is still significantly more herding for corporate bonds and financial-institution bonds, the herding statistics are also significant for government bonds and, in one instance, for mortgage bonds. These results hold when more than two and more than three funds trade assets. The results then suggest that part of the herding in government bonds is driven by PFAs purchasing those securities for several of their funds. In fact, PFAs participate actively in government bond auctions, demanding a significant proportion of the securities that come to markets (Opazo et al. 2009).
Table 7

Herding within PFAs across Funds. This table presents the average Lakonishok et al. (1992) herding statistic by asset class by using data from a multi-fund period (2002 to 2005). The herding statistic is calculated within PFAs and across funds and then averaged across PFAs, using the asset-specific probability of buying an asset at any point in time. The herding statistic over all asset classes is calculated based on the average portfolio share of each asset class. Columns (1), (2), and (3) present the results considering assets traded by more than one, more than two, and more than three funds, respectively. Numbers represent percentages (results are multiplied by 100). Column (4) presents the average asset-specific probability of buying an asset, calculated over the assets traded by more than one fund, by asset class. T-tests are one-tailed

 

Herding Statistic

 

Assets Traded by More than One Fund

Assets Traded by More than Two Funds

Assets Traded by More than Three Funds

Average Probability of Buying an Asset

(1)

(2)

(3)

(4)

All Asset Classes

−2.15

2.49***

5.36***

48.77 %

 

(0.47)

(0.69)

(0.84)

 

Domestic Assets

    

  Corporate Bonds

−0.62

5.84***

11.85***

58.15 %

 

(0.71)

(1.01)

(0.24)

 

  Financial-Institution Bonds

0.27

8.63***

12.38***

44.77 %

 

(0.97)

(1.38)

(1.85)

 

  Government Bonds

−3.26

4.87***

9.28***

56.32 %

 

(0.38)

(0.68)

(1.03)

 

  Mortgage Bonds

−2.93

−0.83

1.22***

10.35 %

 

(0.10)

(0.12)

(0.25)

 

  Equity

−1.39

−1.03

−1.25

58.16 %

 

(0.45)

(0.54)

(0.76)

 

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors are presented in parentheses

Table 8 shows the results from comparing funds within fund types across PFAs. Interestingly, the herding statistics increase noticeably across the board in this case, both in terms of the point estimates and the statistical significance of the coefficients. For example, relative to the estimates at the PFA level, the average herding across asset classes in Table 8 is 5.22 for assets traded by more than two funds versus 2.00 in Table 5, and 3.71 for assets traded by more than one fund versus -1.01 in Table 5. The asset classes that experience more herding are corporate bonds and financial-institution bonds. The ones that experience less herding are government bonds and mortgage bonds. Equity is in the middle. What is also clear from this table is that the herding in equity is driven almost exclusively by herding within fund types across PFAs.
Table 8

Herding within Fund Types across PFAs, with Buy and Sell Decomposition. This table presents the total average Lakonishok et al. (1992) herding statistic and the statistics for the buy and sell subgroups, following the Grinblatt et al. (1995) methodology and using only data from the multi-fund period (2002 to 2005). The herding statistic is calculated within fund type and across PFAs, and then averaged across funds by using the asset-specific probability of buying an asset at any point in time. The herding statistic over all asset classes is calculated based on the average portfolio share of each asset class. Columns (1), (2), and (3) present the results considering assets traded by more than one PFA. Columns (4), (5), and (6) present the results for assets traded by more than two PFAs. Columns (7), (8), and (9) show the results for more than three PFAs. Numbers represent percentages (results are multiplied by 100). Column (10) presents the average asset-specific probability of buying an asset, calculated over the assets traded by more than one fund, by asset class. T-tests are one-tailed

 

Herding Statistic

 

Assets Traded by More than One Fund

Assets Traded by More than Two Funds

Assets Traded by More than Three Funds

Average Probability of Buying an Asset

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

Total

Buy

Sell

Total

Buy

Sell

Total

Buy

Sell

All Asset Classes

3.71***

5.88***

2.01***

5.22***

8.85***

5.60***

5.80***

9.97***

4.74***

48.44 %

 

(0.29)

(0.50)

(0.33)

(0.41)

(0.76)

(0.50)

(0.57)

(1.15)

(0.77)

 

Domestic Assets

          

  Corporate Bonds

12.33***

11.65***

15.01***

19.57***

16.14***

22.54***

24.03***

20.30***

26.80***

52.73 %

 

(0.68)

(0.76)

(0.04)

(0.85)

(1.36)

(0.05)

(1.02)

(1.64)

(0.08)

 

  Financial-Institution Bonds

12.51***

14.34***

12.18***

15.49***

17.57***

18.63***

14.47***

19.81***

18.38***

37.53 %

 

(1.01)

(1.29)

(1.99)

(1.51)

(2.51)

(2.15)

(2.62)

(5.28)

(3.70)

 

  Government Bonds

1.20***

1.19***

1.22**

3.43***

0.28

6.24***

3.10***

3.26***

2.96*

57.94 %

 

(0.35)

(0.41)

(0.70)

(0.67)

(0.88)

(0.98)

(1.18)

(1.36)

(1.93)

 

  Mortgage Bonds

1.93***

10.73***

−0.81

0.21***

24.91***

−0.94

−0.10

19.25***

−1.07

4.66 %

 

(0.08)

(0.31)

(0.03)

(0.07)

(0.82)

(0.04)

(0.08)

(0.83)

(0.05)

 

  Equity

5.20***

6.69***

0.47

5.88***

7.19***

2.42***

7.54***

8.92***

3.64***

54.79 %

 

(0.32)

(0.33)

(0.77)

(0.35)

(0.39)

(0.75)

(0.43)

(0.47)

(0.91)

 

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors are presented in parentheses

To further understand where the trading behavior is coming from, Table 8 also presents a decomposition of herding into herding in buying and herding in selling, following Grinblatt et al. (1995).14 In general, the results suggest that herding occurs at both sides of the transactions. That is, pension funds herd both when they purchase securities and when they sell them in secondary markets. The only exception is the case of mortgage bonds that show herding just on the buying side. This might be due to pre-payment and restructuring of those bonds, which might lead pension funds to remove them from their portfolios at different points in time.

Table 9 decomposes herding by type of fund and shows that the result of herding within fund types is not due to herding in only one type of fund. Instead, herding within fund types across PFAs occurs across all types of funds.
Table 9

Herding within Fund Types Across PFAs, by Fund Type. This table presents the average Lakonishok et al. (1992) herding statistic by asset class by using data from the multi-fund period (2002 to 2005). The herding statistic is calculated within fund types and across funds (one per PFA), by fund type, considering assets traded by more than one fund for each fund type. The herding statistic over all asset classes is calculated based on the average portfolio share of each asset class. Numbers represent percentages (results are multiplied by 100). For Fund E, no investments are allowed in equity. T-tests are one-tailed

 

Herding Statistic - Assets Traded by More than One Fund

Fund A

Fund B

Fund C

Fund D

Fund E

(1)

(2)

(3)

(4)

(5)

All Asset Classes

5.87***

3.54***

7.99***

5.65***

4.67***

 

(0.92)

(0.65)

(0.49)

(0.66)

(0.84)

Domestic Assets

     

  Corporate Bonds

13.61***

11.47***

20.80***

10.51***

13.02***

 

(1.93)

(0.85)

(0.08)

(0.88)

(1.06)

  Financial-Institution Bonds

6.63***

10.78***

15.33***

9.49***

13.56***

 

(2.61)

(1.29)

(1.21)

(1.25)

(1.70)

  Government Bonds

1.21

4.91***

2.96***

4.94***

2.08***

 

(1.72)

(0.84)

(0.44)

(0.67)

(0.80)

  Mortgage Bonds

5.02***

2.89***

1.24***

2.52***

3.26***

 

(0.85)

(0.17)

(0.08)

(0.14)

(0.32)

  Equity

6.32***

0.69*

10.43***

6.68***

 

(0.43)

(0.45)

(0.60)

(0.64)

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors are presented in parentheses

As mentioned in Section 3, pension funds tend to purchase fixed-income securities at issuance and hold them until maturity. The herding statistics reported here do not include the dates when instruments are removed from the markets. So they are not affected by the maturing fixed-income instruments. However, they do include initial purchases at issuance. While this does not pose a bias to the estimates, it raises the question of whether herding is driven mainly by these initial acquisitions. To answer this question, we re-compute the herding statistics excluding purchases at issuance for fixed-income assets. We do so for the estimates at the PFA level and within fund types across PFAs. Relative to Table 4, the estimates at the PFA level show that herding is prevalent even after the purchases at issuance, when securities are bought in secondary markets.15 However, as expected, the herding estimates for fixed-income securities are lower when initial acquisitions are removed from the sample. Relative to the previous tables, the ranking of herding across asset classes holds. Compared to Table 8, the estimates within fund types across PFAs yield similar conclusions. The results are reported in the working paper version of this paper (Raddatz and Schmukler 2011).

5 What other factors might be related to herding?

Aside from the herding studied so far, which refers to contemporaneous herding, there can also be dynamic herding if funds follow the herd with a lag. Therefore, assets that are more heavily traded in a given period are also more likely to be traded in subsequent periods. Sias (2004) studies this dimension of herding and tests the hypothesis that the intensity of trading is serially correlated. We do so by estimating the parameter βt in the following equation for each month t:
$$ {\varDelta_{i,t }}={\beta_t}{\varDelta_{i,t-1 }}+{\varepsilon_{i,t }}, $$
(4)
where \( {\varDelta_{i,t }}=\frac{{Ra{w_{i,t }}-\overline{{Ra{w_t}}}}}{{\sigma {{{\left( {Raw} \right)}}_t}}} \), Rawi,t is the fraction of PFAs buying asset i at time t among those active (\( B\left( {i,t} \right)/N\left( {i,t} \right) \) in the previous notation), and \( \overline{{Ra{w_t}}} \) and σ(Raw)t are the average and standard deviation of Rawi,t among all assets i, respectively. The parameter βt corresponds, therefore, to the serial correlation of the standardized fractions of PFAs that are buying an asset, which is permitted to vary with time.16
Table 10 reports the results on dynamic herding. Each entry in the table reports the average βt across months for various asset classes, its standard error, and the fraction of periods in which the coefficient is significantly different from zero at the 10 % level. When considering all the active assets across classes (first row in column 1), we find evidence of a significantly negative serial correlation in trades. Assets that are more intensively bought in a given month are significantly less likely to be bought during the next month. Moreover, this significantly negative coefficient occurs in all of the 1-month regressions. The rest of the results reported in column (1) indicate that the negative serial correlation is present in almost all asset classes with domestic equities being the only asset class in which there is significant evidence of positive dynamic herding. One possible explanation for this finding is that pension funds cannot quickly adjust their positions in domestic equity markets.
Table 10

Dynamic Herding by Fund Type. This table presents a measure of dynamic herding across time for all assets and by asset class for each fund type. For each moment in time, we run an ordinary least squares regression of the probability of buying an instrument at that time on the lagged probability of buying an instrument. The average coefficient of this exercise is shown in the table. Numbers represent percentages (results are multiplied by 100). T-tests are two-tailed

 

Herding Regressions-All Assets

Fund A

Fund B

Fund C

Fund D

Fund E

(1)

(2)

(3)

(4)

(5)

All Asset Classes

Average Coefficient

−15.49***

−32.75***

−28.12***

−29.59***

−41.08***

 

Standard Error

(2.06)

(2.82)

(2.70)

(2.69)

(3.21)

 

% of Positive Coefficients

0.00 %

0.00 %

10.53 %

0.00 %

0.00 %

 

% of Negative Coefficients

59.46 %

91.89 %

86.84 %

91.89 %

92.11 %

Domestic Assets

      

  Corporate Bonds

Average Coefficient

−15.33

−21.38**

−20.90***

−3.23

−10.82

 

Standard Error

(10.7)

(7.75)

(3.93)

(5.69)

(6.65)

 

% of Positive Coefficients

0.00 %

0.00 %

0.00 %

0.00 %

3.03 %

 

% of Negative Coefficients

12.50 %

12.50 %

20.00 %

85.71 %

12.12 %

  Financial-Institution Bonds

Average Coefficient

−29.47

−19.95

−9.58

−31.11*

−22.74

 

Standard Error

(40.9)

(25.9)

(12.8)

(16.1)

(20.5)

 

% of Positive Coefficients

0.00 %

0.00 %

12.50 %

0.00 %

0.00 %

 

% of Negative Coefficients

0.00 %

0.00 %

12.50 %

0.00 %

0.00 %

  Government Bonds

Average Coefficient

−33.95***

−36.56***

−28.33***

−31.89***

−38.50***

 

Standard Error

(5.43)

(3.30)

(3.22)

(2.97)

(4.64)

 

% of Positive Coefficients

0.00 %

0.00 %

2.63 %

0.00 %

2.63 %

 

% of Negative Coefficients

34.29 %

70.27 %

73.68 %

72.97 %

73.68 %

  Mortgage Bonds

Average Coefficient

−63.53***

−55.57***

−40.83***

−45.14***

−43.67***

 

Standard Error

(6.45)

(4.44)

(3.74)

(4.14)

(4.04)

 

% of Positive Coefficients

0.00 %

0.00 %

7.89 %

0.00 %

0.00 %

 

% of Negative Coefficients

89.29 %

91.89 %

86.84 %

89.19 %

86.49 %

  Equity

Average Coefficient

17.79***

21.73***

22.42***

0.65

 

Standard Error

(2.67)

(4.00)

(3.34)

(4.79)

 

% of Positive Coefficients

56.76 %

45.95 %

57.89 %

21.21 %

 

% of Negative Coefficients

2.70 %

0.00 %

2.63 %

6.06 %

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors of the average coefficient are presented in parentheses. In addition, this table presents the percentage of positive and negative coefficients that are statistically significant at a 10 % level. The dashes in column (5) indicate that equity is not traded by Fund E

As mentioned in the introduction, aside from herding, there can also be momentum trading. A fund is typically called a momentum trader if, on average, it sells assets with low past performance and purchases securities with high past returns. On the other hand, a fund that sells past winners and buys past losers is called a contrarian trader, and a fund that follows none of these strategies is a non-momentum trader. The working paper version of this study shows evidence of momentum trading.

Momentum trading might explain the previous results on herding. If funds chase returns, they tend to buy assets when their returns are positive and look like they are following each other; when instead, they are following returns. However, the results from Table 11 suggest that momentum trading does not explain herding. In particular, the herding statistics are unrelated to the lagged returns of the assets included in each class. In unreported results, the same conclusions are reached if one analyzes dynamic herding instead. These conclusions suggest that managers’ common preferences over asset characteristics, such as stocks with high past returns, do not seem to drive herding behavior.
Table 11

Does Momentum Explain Herding? This table presents the results of ordinary least squares regressions of the herding statistic that uses the asset-specific probability of buying an asset, on a constant and the lagged rate of return. The regressions are computed over all asset classes and for each asset class separately, considering assets traded by more than two PFAs or funds. Numbers represent percentages (results are multiplied by 100). T-tests are two-tailed

 

Herding Statistic on Lagged Return

PFA Level

PFA Level (Multi-Fund Period)

PFA-Fund Level

Within Fund Types across PFAs

Constant

Lagged Return

Constant

Lagged Return

Constant

Lagged Return

Constant

Lagged Return

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

All Asset Classes

0.85***

−5.79*

−0.69

0.94

−0.48

−0.98

1.04***

10.64

 

(0.02)

(3.72)

(0.02)

(3.29)

(0.02)

(4.02)

(0.06)

(7.11)

Domestic Assets

        

  Corporate Bonds

8.54***

47.53

8.69***

87.42

−0.11

−8.09

19.89***

15.54

 

(0.20)

(88.22)

(0.07)

(97.59)

(0.00)

(13.95)

(0.03)

(21.45)

  Financial-Institution Bonds

8.12***

−82.76**

8.69***

−84.08**

2.87***

−35.90

12.49***

−29.25**

 

(0.15)

(47.14)

(0.05)

(45.72)

(0.08)

(33.74)

(0.05)

(12.56)

  Government Bonds

−1.67

12.38

−2.26

8.66

−2.52

15.7***

5.38***

8.67

 

(0.08)

(18.74)

(0.02)

(15.11)

(0.02)

(4.69)

(0.00)

(11.69)

  Mortgage Bonds

1.21***

−44.8***

−0.64*

−17.33*

−0.53

−12.60

0.17***

−9.02

 

(0.13)

(16.18)

(0.08)

(11.58)

(0.06)

(12.34)

(0.05)

(7.55)

  Equity

1.68***

−3.45

0.06

6.61

0.87***

2.70

6.48***

0.58

 

(0.02)

(4.03)

(0.10)

(5.26)

(0.06)

(2.71)

(0.10)

(4.88)

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors are presented in parentheses

Regulation might also play a role in the findings on herding. Chilean PFAs are subject to a minimum return requirement relative to the average return that might induce fund managers to mechanically herd around the average portfolio to avoid penalties. The time variation of the herding measures can help to determine the impact of changes in regulation. In October 1999, the average real rate of return to calculate the minimum return changed from 12 months to 36 months. This greater flexibility was expected to reduce the degree of herding, because the reform gave managers more time to converge to the average return. However, the data show no evidence of a decline in herding around the date of the reform. Table 12 compares the herding statistics computed in windows of 18 months before and after October 1999 for each asset class. For most asset classes, instead of a decline, we observe an increase in the herding statistic after the reform. Only among mortgage bonds, is there evidence of a small decline in herding. Thus, these findings do not support the claim that herding is mainly due to the tightness of the regulatory band.
Table 12

Evolution of Herding Statistic. This table presents the average Lakonishok et al. (1992) herding statistic by asset class considering 18 months before and after the regulatory reform in October 1999. Panel A shows the herding statistic for Fund C by using the asset-specific probability of buying an asset at any point in time. Numbers represent percentages (results are multiplied by 100). Panel B shows p-values for the one-sided t-test of equality of the herding statistic between the observations previous to the regulatory reform (April 1998 to October 1999) and the observations after the reform (October 1999 to April 2001). T-tests are one-tailed

 

Panel A. Herding Statistic

Assets Traded by More than One PFA

Assets Traded by More than Two PFAs

Assets Traded by More than Three PFAs

Before Regulatory Reform

After Regulatory Reform

Before Regulatory Reform

After Regulatory Reform

Before Regulatory Reform

After Regulatory Reform

Corporate Bonds

4.15**

7.07***

2.19**

8.85***

1.85**

8.29***

 

(1.81)

(1.94)

(0.98)

(2.38)

(0.90)

(2.85)

Financial-Institution Bonds

−0.57

7.01**

−0.43

8.03**

7.61

6.47**

 

(2.22)

(3.13)

(2.96)

(4.16)

(4.03)

(2.32)

Government Bonds

−0.44

−0.00

1.10

0.79**

3.40**

2.30***

 

(0.44)

(0.27)

(0.87)

(0.46)

(1.44)

(0.87)

Mortgage Bonds

6.56***

6.02***

3.46***

2.65***

1.70***

1.10***

 

(0.21)

(0.20)

(0.19)

(0.18)

(0.17)

(0.16)

Equity

0.81

2.64***

1.16**

3.15***

1.60**

4.14***

 

(0.61)

(0.68)

(0.70)

(0.79)

(0.83)

(0.96)

 

Panel B. P-Value for Hypothesis Testing: Herding Before the Reform > Herding After the Reform

Assets Traded by More than One PFA

Assets Traded by More than Two PFAs

Assets Traded by More than Three PFAs

Corporate Bonds

0.93

1.00

0.99

Financial-Institution Bonds

0.98

0.95

0.40

Government Bonds

0.79

0.40

0.28

Mortgage Bonds

0.01

0.00

0.00

Equity

0.98

0.98

0.99

*, **, and *** indicate statistical significance at the 10 %, 5 %, and 1 % levels, respectively. Standard errors are presented in parentheses

In addition, we analyze the possibility that herding behavior might be driven by the regulatory minimum return band by comparing the degree of herding observed across funds that face different regulatory bands according to their risk profiles. Although the band is typically larger for riskier funds, groups of funds with different risk profiles (i.e., funds investing different shares of their portfolios in riskier assets) face the same regulatory band. For instance, Funds C, D, and E face a band of two percentage points around the average return despite their different risk profiles; for Funds A and B, the band is four percentage points.

Because a given size of a regulatory band should be more binding for riskier funds due to their higher absolute return volatility, we expect to observe more herding in Fund C than in D and E, and in A relative to B.17 The results in Table 9 tend to support this prediction. When considering herding over all asset classes the table shows that herding in Fund A is higher than that in Fund B. Moreover, herding in Fund C is higher than that in Fund D, which in turn is higher than that in Fund E. Notice that the pattern in Table 9 cannot simply result from a relation between the riskiness of the assets and herding. If that were the case, there should be a decreasing degree of herding as the riskiness of the portfolio declines from Fund A to Fund E. Instead, the decreasing relation occurs only across funds that face similar regulatory bands. This decrease suggests that regulations that lead funds to follow industry benchmarks, such as a minimum return band, might impact the way that funds behave and induce herding. In sum, the evidence provides only mixed support for the idea that some aspects of the regulation might contribute to herding among pension funds.

Also, the lack of available instruments does not drive the herding results. This conclusion can be reached by comparing the number of instruments approved by the Risk-Rating Commission (Comisión Clasificadora de Riesgo, CCR) in various asset classes for the period of 2002 to 2005 and the fraction of approved instruments in which PFAs invest. On average, PFAs invest only in a subset of the available assets, 47 % in the case of corporate bonds. This investment pattern suggests that herding is not driven by the fact that all PFAs purchase the same assets when they become available because they have already exhausted the supply of investable assets. On the contrary, they select the same assets at the same time from a wide range of alternatives. Similar conclusions are reached if one looks at the auctions of government paper and the biddings by PFAs in those public offerings (Opazo et al. 2009).

6 Conclusions

Using unique pension fund data from Chile, this paper advances the understanding of herding behavior and, more generally, the investment practices of institutional investors. In particular, the paper exploits a rich data set to analyze herding behavior among pension funds across different asset classes and levels of the industry. In doing so, it sheds light on the underpinnings of herding and the implications for capital market development.

The paper shows that pension funds herd significantly in their investment decisions. In particular, herding is more pronounced in instruments that are more opaque, which suggests that pension funds copy each other in their investment decisions as a way to overcome informational problems. Herding is more prevalent in corporate and financial-institution bonds, followed by equity, mortgage bonds, and government bonds. These findings are consistent with the view that asset characteristics matter for herding and highlight an important shortcoming of the literature that typically focuses on herding in a particular asset class (equity). The large prevalence of bonds in the portfolios of many institutional investors and the differences we document suggest that the evidence that uses just equity investments might lead to incorrect conclusions about the magnitude and potential consequences of herding.

Our results also shed light on the relation between competition and herding. We find that herding is more intense when comparing similar types of funds across PFAs than when comparing aggregate PFA portfolios. Narrowly defined fund types are easily compared by the public, the regulator, the overall managers of PFAs, and peers, and thus compete directly with each other across PFAs. We still find evidence of herding at other levels (across PFAs, all PFA-funds together, and various fund types within the same PFA), although less strong. These results are consistent with herding being driven by incentives for managers to be with the pack of their direct competitors and not deviate from industry standards.

Overall, the incentives to herd seem to come from the opaqueness and riskiness of the assets in which funds invest, the endogenous and regulatory driven use of industry benchmarks (not exclusive of Chile), and the existence of a clearly defined group of competitors. Investors and regulators face a tradeoff between the need to monitor asset managers on a regular basis, on the one hand, and the need to give incentives and space to managers to engage in long-term arbitrage and asset discovery, on the other. These incentives on asset allocation by pension funds and other institutional investors might be important and have been typically overlooked by the literature.

Finally, the findings in this paper have implications for the general debate on capital market development, where pension funds are expected to play a key role.18 On the bright side, pension funds seem to absorb a large amount of the bonds in primary markets. However, the high degree of herding is consistent with pension funds following each other in their investment strategies.19 Moreover, because pension funds invest only in a fraction of the available assets, herding behavior does not seem to be explained by the lack of investable instruments. In addition, pension funds tend to display relatively little turnover, which does not seem to square well with the idea that they contribute to the liquidity of secondary markets. In sum, the evidence suggests that at least the initial ideas that motivated the introduction of pension funds as a driving force in secondary capital market development need to be revisited.

Footnotes
1

Many papers in the literature focus on equity mutual funds, whose portfolio data are mostly available publicly, and study their investment patterns (e.g., Grinblatt et al. 1995; Wermers 1999; Kacperczyk et al. 2005). Others analyze general data on institutional investors (e.g., Sias and Starks 1997; Nofsinger and Sias 1999; Grinblatt and Keloharju 2000; Sias 2004; Choi and Sias 2009).

 
2

The few studies available on pension funds typically use quarterly data for a subsample of the funds and focus almost exclusively on equity holdings. Lakonishok et al. (1992), Badrinath and Wahal (2002), and Ferson and Khang (2002) analyze US data. Blake et al. (2002) and Voronkova and Bohl (2005) use data from the UK and Poland, respectively. Langford et al. (2006) use return data to study Australia’s superannuation funds.

 
3

For the relation between size, trading activity, and opacity, see Livingston et al. (2007) and Bessembinder and Maxwell (2008).

 
4

The original data also contain information on the holdings of derivatives, investment and mutual fund quotas, former pension system bonds, deposits, and foreign assets, but we exclude them from the analysis for various reasons. For instance, former pension system bonds are securities that were issued to the workers that moved from the old pay-as-you-go system to the new pension system when the reform was implemented in 1981. Thus, they are highly idiosyncratic and are increasingly disappearing as the system matures. Also, while quotas of investment and mutual funds are variable income instruments, the underlying assets are in many cases fixed income (bond funds) or a combination of bonds and equity, so they cannot be easily mapped into the standard categories. Foreign assets are excluded because, for regulatory reasons, most foreign investment carried out by Chilean pension funds occurs through the purchase of quotas of foreign investment funds.

 
5

While mortgage bonds (letras hipotecarias) represent 73.3 % of the observations, they only stand for 19.6 % of the investment when considering the entire period 1996–2005.

 
6

In this table, corporate bonds and financial institutions bonds are grouped together for data availability reasons.

 
7

The median amounts per issue are smaller for corporate bonds than for equity because companies typically issue several series of a corporate bond. Median amounts per issue for government bonds are small because the government tries to issue regularly to provide liquidity to the market and establish the benchmark yield curve. However, this should not make each issuance less transparent because the underlying debtor is the same. Summary statistics on amounts per issue are not reported but are available upon request.

 
8

Data on the trading frequency of corporate and government bonds come from Lazen (2005). Trading frequency for equities come from the Santiago Stock Exchange.

 
9

See, for example, Amihud and Mendelson (1986), Chordia et al. (2001), and Bekaert et al. (2007).

 
10

Infrequent trading does not necessarily mean that PFAs do not actively change the relative composition of their portfolios because, even if most assets are not traded, their relative importance depends on the changes experienced by those that are active.

 
11

The turnover measures described above are useful to determine the extent to which PFAs rebalance their portfolios, but they do not appropriately capture the extent to which that rebalancing is passive or active. In other words, part of the turnover might just be the consequence of passive trading due to: (i) the constant net inflows PFAs receive from current contributors that have not yet retired, or (ii) outflows due to pensioners retiring and leaving the system. Passive trading might also occur because some assets mature and, in order to reinvest them, PFAs need to purchase new instruments. Therefore, the amount of active turnover and the number of managers willing to change positions over time to maximize returns is lower than the turnover measures reported above.

 
12

Some papers propose that financial-institution bonds are more opaque than standard corporate bonds (Morgan 2002). However, others have shown that large banks are not more opaque than comparable corporations (Flannery et al. 2004). In Chile, only large banks issue corporate bonds, so in principle there is not a large difference in opacity between these two asset classes.

 
13

During 1996–1999, pension funds administrators offered a single fund (corresponding to Fund C in the current classification), and during 2000–2002 they offered two funds (corresponding to Funds C and D in the current classification).

 
14

The working paper version of this paper, Raddatz and Schmukler (2011), shows the buy and sell decomposition at the PFA level as well.

 
15

These results might slightly overestimate the role of purchases at issuance. The reason is that we do not have the issuance date for the fixed-income assets, so we assume that it corresponds to the date when the assets first appear in our data set. This is a good approximation because PFAs tend to quickly absorb fixed-income assets in their portfolio. According to market participants, PFAs actively demand assets during the underwriting process. However, it is possible that, in a few cases, we might exclude the first purchase of an asset in secondary markets.

 
16

The reason Sias (2004) standardizes the statistics is that it conducts inference on βt based on the time-variation of the parameters only (à la Fama and MacBeth 1973), and the standardization of the variables controls for changes in their mean and variance over time. Sias’ approach is reasonable but cannot be directly applied to the Chilean data because Chilean PFAs trade infrequently and a large fraction of the assets that are active in a month are not traded in the following one. This means that the sample over which the regressions in equation (2) can be estimated (i.e. the sample of assets traded in two consecutive periods) is different from the sample of traded assets in each period. Moreover, the mean and variance of the standardized statistics are different from zero and one, respectively, in the regression sample. Since the regression sample changes over time, the correct standardization in our case is time varying. We achieve this time-varying standardization by estimating the regressions of the raw fractions (Rawi,t) including a constant (to remove the mean of the dependent and independent variable) and then correcting the estimated coefficients, multiplying them by the ratio of the standard deviation of the dependent to the independent variable in each regression sample.

 
17

The idea is that funds investing in riskier assets will have a higher degree of idiosyncratic volatility if they do not follow the herd, making them more likely to hit the regulatory band. On the other hand, if they always follow the crowd, all risk in their portfolios would be aggregate risk. So, even if their absolute returns are volatile, their relative returns are not.

 
18

For a discussion, see Bodie (1990), Piñera (1991), Davis (1995), Vittas (1999), Arrau and Chumacero (1998), Zurita (1999), Catalán et al. (2000), Impavido and Musalem (2000), Lefort and Walker (2002a, b), Blommestein (2001), Davis and Steil (2001), Corbo and Schmidt-Hebbel (2003), Impavido et al. (2003), Catalán (2004), Olivares (2005), Yermo (2005), Berstein and Chumacero (2006), de la Torre and Schmukler (2006), Andrade et al. (2007), The Economist (2008), and Raddatz and Schmukler (2011).

 
19

Moreover, pension funds invest heavily short term. Opazo et al. (2009) show that Chilean pension fund portfolios are short term relative to those of insurance companies and US mutual funds.

 

Copyright information

© Springer Science+Business Media New York 2012