DOCUMENTOS DE TRABAJO FCEA Departamento de Contabilidad

DOCUMENTOS DE TRABAJO FCEA
ISSN 1909-4469 / ISSNe 2422-4642
Año 2016
No.20
Departamento de Contabilidad y Finanzas
ARE PENSION FUNDS DETERMINANTS OF
FINANCIAL MARKETS STABILITY? A DYNAMIC
ANALYSIS OF OECD COUNTRIES
Jesús Ancizar Gómez Daza
Luis Ferruz Agudo
Facultad de Ciencias Económicas y Administrativas, FCEA
1
Año 2016
DOCUMENTOS DE TRABAJO FCEA
No.20
ISSN 1909-4469 / ISSNe 2422-4642
Documento de Trabajo FCEA
ISSN 1909-4469 / ISSNe 2422-4642
Año 2016 No. 20
Are pension funds determinants of financial markets stability? A dynamic analysis of OECD countries
Autores: Jesús Ancizar Gómez Daza. [jancizarg@javerianacali.edu.co]
Luis Ferruz Agudo. [lferruz@unizar.es]
Departamento de Contabilidad y Finanzas
WEBSITE: wp_fcea.javerianacali.edu.co
Comité editorial
Alina Gómez Mejía
Julián Piñeres
Luis Fernando Aguado
Pedro Pablo Sanabria Pulido
Correspondencia, suscripciones y solicitudes
Calle 18 No. 118-250 Vía Pance
Santiago de Cali, Valle del Cauca, Colombia
Pontificia Universidad Javeriana Cali
Facultad de Ciencias Económicas y Administrativas
Teléfonos: (57+2) 3218200 Ext.: 8694
Correo electrónico: dt.fcea@javerianacali.edu.co
Sello Editorial Javeriano - 2016
Coordinador: Iris Cabra
icabra@javerianacali.edu.co
Concepto Gráfico: William Fernando Yela Melo
Formato 28 x 21 cms.
©Derechos Reservados
©Sello Editorial Javeriano
Enero de 2016
La serie de Documentos de Trabajo FCEA pone a disposición para el análisis, discusión y retroalimentación de la comunidad académica los avances y
resultados preliminares del trabajo académico de los profesores de la Facultad de Ciencias Económicas y Administrativas. Estos documentos no han sido
sometidos a procesos de evaluación formal por pares internos ni externos a la Facultad. Se espera que muchos de estos documentos posteriormente sean
sometidos a evaluación en publicaciones especializadas.
Las opiniones expresadas en este documento son de exclusiva responsabilidad de los autores y no comprometen institucionalmente a la Facultad de
Ciencias Económicas y Administrativas, ni a la Pontificia Universidad Javeriana Cali.
2
Contenido
1. INTRODUCTION
2. PENSION FUNDS AND THE VOLATILLITY OF FINANCIAL
MARKETS
3. LITERATURE REVIEW
3.1 IN FAVOR OF THE EFFICIENT MARKET HYPOTHESIS (EMH)
3.2 AGAINST THE EFFICIENT MARKET HYPOTHESIS
4. MODEL, DATA, VARIABLES AND METHODOLOGY
5. CONCLUSIONS
6. BIBLIOGRAPHY
3
ARE PENSION FUNDS DETERMINANTS OF FINANCIAL
MARKETS STABILITY? A DYNAMIC ANALYSIS OF OECD
COUNTRIES
Jesús Ancizar Gómez Daza
jancizarg@javerianacali.edu.co
Departamento de Contabilidad y Finanzas
Pontificia Universidad Javeriana Cali
Luis Ferruz Agudo
lferruz@unizar.es
Departamento de Contabilidad y Finanzas
Universidad de Zaragoza
ABSTRACT
This paper studies the relationship between the volatility of financial markets and
positions in government bonds, corporate bonds, and stocks held by institutional
investment portfolios, particularly pension funds. Panel data corresponding to the 34
countries of the Organization for Economic Co-operation and Development (OECD) in
the period 2001-2012 are used. Other methodologies for estimating panel data, including
fixed effects, random effects, and the generalized method of moments (GMM), are
considered. First, fixed models are estimated, followed by dynamic models. In terms of
results, the study finds empirical evidence in favor of the arguments made by Thomas et
al. (2014) for the fixed case; however, regarding the dynamic case, the results show that
variations in the position of pension fund portfolios have a significant and direct impact
on the stability of financial markets.
Key words: Volatility, Pension Fund, Financial Markets, Panel Data, Random Effects.
JEL Classification: G10, G23, C18.
4
¿SON LOS FONDOS DE PENSIONES DETERMINATES DE LA
ESTABILIDAD DE LOS MERCADOS FINANCIEROS? UN
ANALISIS DINAMICO PARA LOS PAISES DE LA OCED
RESUMEN
Este trabajo investiga la relación entre volatilidad de los mercados financieros y las
posiciones en deuda pública, deuda privada y acciones en las carteras de inversión
institucionales, concretamente fondos de pensiones. Para tal propósito se utiliza un panel
de datos correspondiente a los 34 países de la OECD en un periodo comprendido entre el
año 2001 y el año 2012. Se utilizan metodologías de estimación de panel de datos que
incluyen modelos de efectos fijos, efectos aleatorios y estimación GMM. En una primera
etapa se estiman modelos de carácter estáticos y posteriormente modelos dinámicos. A
nivel de resultados de la investigación se encuentra evidencia empírica a favor de lo
planteado por Thomas, et al. (2014) para el caso estático, pero para el caso dinámico los
resultados muestran que las variaciones de las posiciones de las carteras de los fondos de
pensiones sí impactan de manera significativa y directa la estabilidad de los mercados
financieros.
Palabras clave: Volatilidad, Fondo de Pensiones, Mercados Financieros, Panel de Datos,
Efectos Aleatorios.
Clasificación JEL: G10, G23, C18.
5
1.
Introduction
Pension funds, investment funds, and insurance companies play an important role in the
development of capital markets. These companies have become the primary agents of
institutional investment. The evolution of the resources administered by these
institutional investors over the last 17 years is shown in Figure 1.
From a conceptual standpoint, whether public or private pension funds, compulsory or
voluntary, savings are the vehicles that allow those individual members of the fund to
obtain a pension at the end of their working lives to ensure them of a dignified old age.
However, beyond their social function, pension funds have become an important
investment and financing mechanism, which allows them to play a leading role in the
development of local capital markets.
The importance of these funds raises the question of whether these types of institutional
investors impact the stability of financial markets. This impact on stability, defined as the
volatility of markets, can be approached from two perspectives that are explained below.
 Positive impact: The performance of pension funds serves as a market stabilizer,
and therefore, they are vehicles that calm euphoria in times of volatility shocks
and help reassure other agents participating in these markets. From this
perspective, it is assumed that investment decisions or, rather, decisions that shape
pension fund portfolios follow a very long-term outlook whose main goal, beyond
seeking abnormal returns, is to achieve the satisfactory management of risk to
ensure that fund holders do not lose capital and thus earn an adequate return on
the savings they have made throughout their working lives. In this respect, Meng
and Pfau (2010) confirm that pension fund investments have a positive impact on
the stock market and the market for corporate bonds. However, they also stress
that this impact differs considerably depending on the level of a country’s
financial development.
 Negative impact: Decisions that shape pension fund portfolios serve as an
instrument to exacerbate the euphoria of financial markets and thus add volatility
shocks due to the impact of investment or divestment decisions. According to this
perspective, it is assumed that a significant proportion of pension fund portfolios
are used for short-term investments that seek high returns, liquid assets, and
greater desire for risk. In this regard, Hu (2006) shows that pension funds inject
volatility shocks that undermine financial markets.
Thus, taking into account the significant development of pension funds and their key
importance as the main agents of local capital markets, the main objective of this study is
to conduct fixed and dynamic analyses to determine whether such institutional investors
act as mechanisms that moderate or destabilize the financial markets of the countries that
belong to the Organization for Economic Co-operation and Development (OECD) in the
6
period 2001-2012. Thus, the significance of pension funds is based on two fundamental
factors.
The first factor relates to the great capacity of pension funds to intervene in financial
markets as a function of the resources they manage. In some countries such as Iceland,
the Netherlands, Switzerland, and the United Kingdom, the resources managed by these
funds are even higher than their GDP (see Figure 2).
Fig. 1. Total assets by type of institutional investors in the OECD, 1995-2012
In USD trillion
Source: Pension Markets in Focus 2013, the annual report of the OECD.
The high level of managed assets provides these types of institutional investors with
sizeable bargaining power, and consequently, their actions or interventions in financial
markets are particularly relevant.
This level of managed assets allows pension funds either to act as mediators of financial
market stability or to negatively impact market volatility and thus destabilize markets.
Thomas et al. (2014) corroborates the argument that, when acting as institutional
investors, pension funds may control the volatility of the stock market.
The second factor relates to the performance of these agents as a significant source of
financing for companies and governments. A large proportion of the resources managed
by pension funds are long-term resources that, in theory, allow them to intervene in
financial markets in a structural and not speculative manner.
Due to the high degree of liquidity these types of agents have, their investment decisions
are long term, which makes them an interesting financing mechanism for those private or
public companies that have decided to issue long-term debt or equity.
On the other hand, governments also benefit from the growth that such agents have been
experiencing since 1995 because pension funds are precisely one of the main buyers of
government debt.
7
Fig 2. Importance of pension funds relative to the size of the economy in selected OECD countries, 2012
As a percentage of GDP
Source: Pension Markets in Focus 2013, the annual report of OECD.
However, given the high amount of managed resources and their high degree of liquidity,
pension funds are beginning to show problems due to imbalances between assets and
liabilities.
These imbalances are a consequence of the flow of people receiving their pensions or
about to receive them being much greater than the flow of new contributors to the fund,
i.e., those entering the labor market and beginning to contribute to their pensions.
Considering the above and the research question presented, the hypothesis to be verified
through empirical work is the following:
Hypothesis 1: Do investment decisions that shape pension fund portfolios in countries
belonging to the OECD act as stabilizers or destabilizers of volatility in the markets to
which these funds belong?
The rest of the paper is structured as follows: Section 2 gives a description of the
influence of pension funds on markets volatility. Section 3 the literature review. The
section 4 includes the model, data and methodology and finally the conclusions in
Section 5 are made.
8
2.
Pension funds and the volatility of financial markets
In the scientific literature, there are two major schools of thought and types of empirical
work. On one hand are those who believe that pension funds, acting as rational agents,
serve as market volatility stabilizers. On the other hand, some works propose and
empirically demonstrate that institutional investors, such as pension funds, act as
financial market destabilizing agents.
Regarding the position that pension funds act as stabilizers, studies such as Thomas et al.
(2014) argue that the presence of pension funds in stock markets is beneficial for the
financial markets of countries that belong to the OECD because this type of institutional
investor contributes to reducing market volatility. Similarly, Walker and Lefort (2002)
analyze the impact of pension funds on emerging economies. They reach the conclusion
that there is a positive influence in areas such as regulatory framework, efficient
allocation processes, setting asset prices, liquidity, and market depth, among others.
According to these authors, these advances in financial markets have positive
implications for economic wellbeing. Beck et al. (2006) construct a theoretical model
based on Bachetta and Caminal (2000) that predicts the behavior of well-developed
financial intermediaries facing the effects of shocks in the real sector and monetary
shocks.
Furthermore, Bohl et al. (2009) find strong empirical evidence in favor of the hypothesis
that the increase in institutional ownership has changed the autocorrelation and structure
of volatility in aggregate stock returns. Their results support the hypothesis that such
investors act as stabilizers of stock index returns. They conclude that the herding effect
and positive feedback trading1 (PFT) implemented by institutional investors do not
necessarily imply that such agents destabilize stock prices. Bohl et al. (2009) use the
Markov-Switching-GARCH methodology as the basis of a hypothesis that states that
investments by pension funds in Poland reduce market volatility. That is, their empirical
evidence favors the idea that such agents act as vehicles of stabilization rather than as
destabilizing agents of financial market volatility. Similarly Choe et al. (1999) find no
evidence that the positions taken by foreign investors had a destabilizing effect on the
Korean stock market in the period and sample analyzed.
Information is a key component in the investment process, specifically when making
decisions on the allocation and structuring of investment portfolios. Dennis and Weston
(2001) find evidence that institutional investors are better informed than individual
investors, and therefore, information-based trading is positively and significantly related
to the number of institutional investors.
The volatility of financial markets has direct consequences on their liquidity levels.
Blume and Keim (2012) show that the participation of institutional investors in stock
markets has increased and has played an important role in explaining the variability of
illiquidity in these markets. Gompers and Metrick (2001) show that the demand for
1
On one hand is what is known as herding, which, according to Lakonishok et al. (1992), refers to
purchases or sales executed simultaneously on the same stocks that other investors buy or sell. On the other
hand, positive feedback trading is defined by the same author as the long positions taken today on assets
that have generated profits in the past and/or the sale today of assets that have generated losses in the past .
9
shares by institutional investors is different from that of other investor types; in essence,
institutional investors show a preference for the stocks of large companies, more liquid
shares, and those with relatively low returns in previous periods. These demands certainly
have consequences on the prices of their preferred assets and on the prices of those assets
not included.
The relationship between firm-level volatility and institutional ownership has been
examined by authors such as Dennis and Strickland (2002), Sias (1996), and Xu and
Malkiel (2003). Despite statements by Rubin and Smith (2009), these authors argue that
the relationship between institutional ownership and firm-level volatility depends on the
interaction of three factors: institutional sophistication, institutional preference for low
volatility assets, and increased levels of institutional activity.
The other stream of literature includes studies, both theoretical and empirical, that
propose a direct relationship between the positions taken in the portfolios of institutional
investors and their effects on the volatility of financial markets, i.e., the approach under
which these agents destabilize the markets through volatility shocks caused by the
positions they assume in financial markets. Gabaix et al. (2006) proposes a theoretical
model in which volatility is caused by the trading of large institutional investors that also
cause spikes in trading volumes and in the returns on assets. Hu (2006) presents a
dynamic model to determine the impact of pension funds for countries that belong to the
OECD on market volatility. He concludes that these agents in both OECD countries and
emerging countries increase market volatility. This result is consistent with the findings
of Davis and Hu (2005).
Dennis and Strickland (2002) find evidence to support the hypothesis that institutional
investors have a stronger reaction than individual investors when the absolute value of
market returns is substantial on particular days. This evidence is consistent with the
hypothesis that states that managers of these types of institutions are evaluated in the
short term, which sometimes requires them to make short-term decisions instead of the
structured decisions they actually ought to make. In the case of banks, the relationship is
negative with abnormal returns on days with an upward or downward market trend.
Lipson and Puckett (2006) find a contrasting conclusion when studying the behavior of
investment (bank) and pension fund managers in days when the market experiences
upward or downward changes above 2%. In this regard, they find strong evidence that
both types of managers behave as net sellers when the market experiences an upward
trend and as net buyers when the market experiences a downward trend.
Thus, the relationship between the volatility of stock prices and the positions assumed by
institutional investors has been extensively studied by modeling the two behaviors
followed by these agents that are considered typical.
Lakonishok et al. (1992) are perhaps among the pioneers in empirically corroborating
herding behavior and PFT for institutional investors. They conclude that there is weak
evidence to support herding and some strong evidence of PFT in the case of shares in
small companies. However, they believe that their data give no solid evidence that
institutional investors destabilize individual stocks prices. Conversely, DeLong et al.
(1990) conclude that, in the presence of PFT, the positions taken by rational speculators
can lead to exuberance among portfolio managers and consequently move prices beyond
their core values. Wermers (1999) finds higher levels of herding in stocks of small
10
companies and a positive relationship between herding and PFT in the case of the growth
strategies of investment funds.
Cai and Zheng (2004) investigate the dynamic relationship between the aggregate trading
activity of institutional investors and stock prices. They find that institutional trading is
closely related to contemporary returns. The evidence shows that institutions are more
likely to follow PFT in long positions than in short positions, thus relieving concerns that
the institutions may exacerbate or deepen a fall in stock prices after a general market
decline.
3.
Literature review: Theoretical and Empirical Background
The literature and background review, both theoretical and empirical, is structured
according to the two trends discussed above.
3.1
In favor of the efficient market hypothesis (EMH)
Among the works in favor of the market efficiency hypothesis is Fama (1965), who
initially demonstrated two key aspects in the behavior of stock returns. The first aspect
refers to the independent behavior that is evident in the series of these returns, and the
second aspect refers to the probability distribution followed by this variable.
Consequently, these findings confirm the previous evidence of Bachelier (1914) and
Osborne (1959) regarding the random or stochastic behavior followed by the returns of
financial assets.
Subsequently, Roberts (1967) coined the term “efficient market hypothesis” and made the
distinction between weak and strong efficiency. A few years later, this hypothesis would
become the taxonomy of efficient markets posed by Fama (1970)2, who in a brilliant
theoretical and empirical study proposes the first formal definition of the efficient market
hypothesis. The efficient market hypotheses proposed by Fama (1970) take the form of
weak, semi-strong, and strong.
Jensen (1968) evaluated the performance of 115 mutual funds and found that, on average,
none of the assessed funds was able to predict future asset prices and consequently failed
to overcome the simple buy-and-hold strategy. Scholes (1972) proposed the price
pressure hypothesis, the substitution hypothesis, and the information hypothesis to
explain the reasonableness of changes in asset prices in a formal market such as a stock
exchange.
In 1973, Malkiel (1973) published his book “A Random Walk down Wall Street”, which
is already considered a classic in the financial and stock market literature. This book set
out to confirm the random behavior of prices and of the returns on financial assets in a
formal market such as a stock exchange.
Samuelson (1973a, 1973b) reexamined his own work in Samuelson (1965) and
established the theoretical basis claiming that price behavior follows a random walk or
martingale; and for that reason, it is very difficult for investors, even when they are
experienced, to obtain above-average yields in a sustained manner.
2
Quoted by Martin Sewell in “History of Efficient Market Hypothesis”. Research Note RN/ 11/04 January
20, 2011. UCL Department of Computer Science.
11
Grossman (1976) presented a theoretical model where there is more than one asset
manager and two types of assets, a risk-free asset and a risky asset. He found that current
prices reveal information to each manager that is of higher quality than his or her own
information. That is, systems with efficient informational prices add information in a
perfect way but also remove the incentive for those managers who are willing to pay for
information.
Le Roy and Porter (1981) designed a series of tests that aim to assess compliance or noncompliance with the efficiency hypothesis. They concluded that caution should be taken
when interpreting the evidence because this evidence might contain some type of bias.
Fama and French (1988) found patterns of negative autocorrelation in the time series of
returns. Autocorrelation begins to be negative for 2-year returns, reaches minimum values
for 3- to 5-year returns, and then tends toward zero for long-term returns.
Malkiel (1989) clearly presented the efficient market hypothesis and outlined its defined
taxonomy of weak, semi-strong, and strong forms. He highlighted empirical studies
validating each of the forms and the work rejecting the fulfillment of these efficiencies.
Finally, he related the different anomalies detected in light of EMH through empirical
exercises. He concluded that, even though there are irregularities in the price formation
process that may even persist for periods of time, the markets can simultaneously be
influenced by trends, fashions, or fads; the market is responsible for correcting itself.
Laffont and Maskin (1990) indicated compliance with the efficiency hypothesis only for
cases of perfect competition where each individual agent is too small to affect market
prices. They reaffirmed that, in the process of price formation, efficiency is met only
under these conditions.
Malkiel (2003) recounted all the empirical work resulting from the random walk
approach and the theory of efficient markets. The paper concluded that, although there
have been periods of irrationality as shown by Shiller (2000) and acknowledging the
existence of market anomalies, the impossibility of investors’ making above-average
profits in the long term or for long periods of times without affecting their degree of risk
aversion is the main test of compliance with the efficiency hypothesis or the inability to
predicting the prices of financial assets.
Hu (2006) began from the fundamental principle that pension funds are more
sophisticated and better informed than individual investors, making this type of
institutional investor capable of restoring the balance in the long term.
Similarly, Lakonishok et al. (1992) specifically evaluated the potential effect of pension
fund trading on stock prices. Finally, they concluded that there is no solid evidence that
institutional investors destabilize the prices of individual stocks.
De Long et al. (1990) suggested that rational speculators must stabilize asset prices. They
argued that, in the presence of PFT strategies, rational speculators act as market
stabilizers, inviting other speculators to break rather than follow the trend. Figlewski
(1979) addressed the same problem involving subjective information or expert opinion as
a new feature. Given this new feature, he investigated the validity of market efficiency
when behaving as a competitive market in which the disregard of subjective information
in the formation of current asset prices is assumed. De Long et al. (1990) followed the
same line of thought, showing that the unpredictability of irrational behavior may itself
be a risk source that makes rational speculation less effective.
12
3.2. Against the efficient market hypothesis
Bearing in mind that the financial system and its various institutions do not fall under the
Walrasian context of competitive markets, much less under an Arrow-Debreu context, in
financial markets, there are transaction costs, markets are not complete, and information
is asymmetric. This last point is perhaps the most important feature for conducting this
study. Grossman and Stiglitz (1981) propose a model that considers the information costs
and two types of agents involved in the market, agents who are informed and agents who
are uninformed. They conclude that, although theorists of efficient markets, particularly
Fama (1970), consider that lower information costs constitute a sufficient condition for
prices to be fully reflected in the information available, they do not consider that
condition to be more than sufficient because it should be a necessary condition. So
Grossman and Stiglitz (1981) propose a model where they find that prices reflect the
information informed agents but only partially, because the costs of acquiring
information. In this regard, given that information is expensive, prices are not perfectly
reflected in the information available, in which case, the agents who spend money or
resources to acquire information would not have the incentive to seek it out because they
would not receive compensation for doing so.
Stiglitz (1981) highlights the role that information plays in the search for an efficient and
competitive market. He argues that, although efficient markets may exist, it is possible
that efficiency is not reached from the perspective of Pareto because information is not
considered symmetrical or complete, which creates incentives for agents involved in the
market to acquire and manage information.
De Bondt and Thaler (1985), supported by previous authors such as Beaver (1981),
Kahneman and Tversky (1982), Shiller (1979), and others, perhaps represent the best
scientific work covering the overreaction of agents to recent news and the excessive
volatility shown by asset prices. This study led to a new and crucial field of study in
modern finance known as behavioral finance. De Bondt and Thaler (1985) find evidence
for the overreaction of agents to new news and announcements, which constitutes an
important test for the rejection of the efficient market hypothesis.
Through an empirical study using data from the period 1962-1985 and several periods of
aggregate returns, Lo and Mackinlay (1988) reject the hypothesis of random walk
behavior followed by the series of assessed returns. However, they indicate that the
rejection of the random walk model does not imply inefficiencies in the formation of
stock prices. Poterba and Summers (1988) indicate that stock returns show positive
autocorrelation over short periods and negative autocorrelation over long
periods.Campbell and Kyle (1993) find that stock prices are heavily influenced by the
presence of some investors who do not maximize profits but negotiate exogenously
instead.
Shiller (1981), aware of the high variation in prices due to the random nature of how new
information that affects them is presented, proposes alternative models to measure
volatility that could allow for greater variation in prices. Shiller (1987) suggests that
changes in the prices of financial assets not only obey psychological factors. In a later
13
work, Shiller (1988) highlights the emergence of so-called “anomalies”, emphasizing the
anomaly related to how the prices of financial assets appear to be too volatile, with the
consequence that their behavior fails to be explained by the efficiency hypothesis. Shiller
(1990) examines the relationship between stock prices and dividends. In this way, he
questions the approach of the efficient market theory in which the current price of a stock
corresponds to the current value of estimated future dividends.
Campbell and Shiller (1988) use a forecasting model that employs the Value at Risk
(VAR) methodology to show that stock returns and dividend/price ratios are too volatile
to be contained in new news about future dividends. This excess in volatility is closely
related to the predictability of multi-period returns.Davis (2005) shows that the increase
in the assets managed by such institutional investors push share prices upward.
Because financial information is asymmetric, which means that there will always be
some agents who are better informed than others and, more importantly, that access to
and administration of information has a cost, one can infer that the acquisition and
management of such information would be less costly for institutional investors, in this
case pension funds, given their economies of scale, on one hand, and given the
administrative and operational structure that grants them greater access to information on
the other hand. Scholes (1972) argues that there are high costs in the search for and
management of valuable information. He argues that one hopes that, when the trading of
shares occurs in large quantities, it is because the agents buying or selling possess
valuable information.
4.
Model, Data, Variables, and Methodology
The empirical study was prepared with the aggregate information of each country’s
respective pension fund portfolios. Initially, the sample corresponds to the 34 countries of
the OECD. Special emphasis is placed on the participation of such portfolios in variable
income assets, debt, and other funds. The main objective of the study is to show whether
the interventions of these institutional investors in financial markets positively or
negatively impacted their stability. Market stability is measured as a function of
annualized volatility.
An unbalanced panel is formed with 34 individual (countries, i) and annual samples from
2001 to 2012 (t). The dependent variable of this panel is the annualized volatility that was
taken from the most representative stock index of each country. Daily observations of
index prices were taken, and the difference in logarithmic returns (compound returns)
was subsequently calculated. Finally, these returns were annualized using the root square
rule (
to match observations of the independent variables. Logarithmic returns were
used for two reasons: first, because these measurements show a probability function
closer to the normal distribution, and second, because their behavior is fixed and does not
have unit roots.
There are three main independent variables in the set: the first is the participation of
variable income assets (vi) in pension fund portfolios; the second is the participation of
government and corporate bonds in pension fund portfolios (bpp); and the third is the
participation of other funds in pension fund portfolios (fund).
Market capitalization to GDP (mcgdp) is a variable that shows in relative terms the size
of the local market in proportion to its respective gross domestic product. Government
bonds to GDP (gbgdp) is a variable that indicates the degree of indebtedness of local
14
governments in proportion to their gross domestic product. This is a key variable
whenever pension funds are a significant vehicle for financing local governments, which
means that, to the extent to which countries face higher debt levels, their pension funds
are also sure to maintain high government bond holdings in their investment portfolios.
The variable (infla) is a macroeconomic control variable. It is important to include it
because the level of the inflation rate is an important determinant of the level of interest
rates. These in turn are the main variables to consider when making an investment
decision, either in variable income or fixed income assets. The variable (gdp), which
actually shows the variation of GDP from year to year, is another macroeconomic control
variable that shows the dynamics of growth or decline in the economy between two
periods, which in this case are annual. The variable (intreal) shows the behavior of
interest rates once the effect of inflation is discounted. As with inflation, this is an
important variable when making investment decisions because it allows one to evaluate
real returns on investment decisions made, in this case, by pension funds.The variable
(gei) represents the Global Equity Index Fund, a benchmark index that functions as a
basis of more than 650 funds throughout the world. In this case, it is used as a proxy
variable for assessing the behavior of the positions held by pension funds in other funds,
such as participation in short-term liquidity. And finally, the variable (asgdp) is a variable
that shows the size of administered assets depending on the respective country's GDP.
This variable is an aggregate of funds in a given country.
Methodology
According to the structure of the data, there are panel data initially comprised of
information corresponding to the aggregate investments of pension fund portfolios of the
34 countries of the OECD. This information was obtained directly from the annual
reports published by the OECD. Other macroeconomic information was obtained from
the World Bank.
The study has two stages. The first stage proposes fixed models that provide conclusive
results regarding whether the hypothesis can be demonstrated. Initially, panel data models
of fixed and random effects are presented, and the best model is then evaluated according
to the results of Hausman test application.
The first general model is specified as follows:
where:

represents the logarithm of market volatility. Market volatility is calculated
from the price performance of stock indexes in each of the countries i in our sample.

represents the proportion of the pension fund portfolio invested in debt
securities. Government bonds and corporate bonds are included.

represents the proportion of the pension fund portfolio invested in variable
income assets.

represents the proportion of the pension fund portfolio invested in other
funds, such as highly liquid investments.
15

represents the market capitalization of country i divided by GDP.

represents the amount of government bonds issued as a proportion of GDP.





represents the variation in the consumer price index for country i.
represents the annual variation in GDP of country i.
represents the magnitude of the of the real interest rate in country i.
represents the Global Equity Index in country i.
represents the total administered assets as a proportion of the GDP of the
respective country.
Estimates and Results
OLS estimates: As shown in the following three tables, the estimate under Ordinary
Least Squares (OLS) does not verify the central hypothesis of the study.
The specification of the models to be estimated by OLS is as follows:
Model 1: lineal model
Model 2: logarithmic model
Model 3: logarithmic difference model
16
Table 1
Estimation by Ordinary least squares (OLS)
Independ. Var.
Mod. Lineal (Panel A)
Mod. Log. (Panel B)
Mod. Dif_Log. (Panel C)
coef
se
coef
se
coef
se
0.001
0.003
0.170*
0.101
-0.039
0.125
vi
-0.005
0.004
-0.091
0.075
-0.004
0.090
fund
-0.005*
0.003
-0.040
0.064
0.008
0.070
mcgdp
0.000
0.001
0.094
0.065
-0.013
0.046
gbgdp
-0.000
0.001
-0.012
0.061
-0.043
0.054
infla
0.028
0.034
-0.028
0.108
-0.069
0.092
gdp
-0.062***
0.016
-0.041
0.086
-0.080
0.066
gei
-0.006***
0.001
-0.004
0.050
-0.016
0.055
intreal
0.005
0.020
0.027
0.112
-0.091
0.117
asgdp
-0.000
0.002
-0.014
0.054
0.176
0.234
_cons
3.146***
0.234
2.168***
0.659
2.529***
bpp
R2
0.336
0.240
0.068
0.190
Notes: Estimates are based on equations of the models 1, 2 and 3. The dependent variable is lvol and independent
variables in panel A are taken linear, logarithmics in panel B and logarithmics difference in panel C. The column shows
the standard errors SE.
*** Denote statistical significance at 1% level
** Denote statistical significance at 5% level
* Denote statistical significance at 10% level
As noted above, OLS estimates do not provide conclusive results regarding the proposed
modeling. Taking into account the panel structure of the data, these estimates also show
biased and inefficient estimators.
Estimates through the Fixed Effects Model
The fixed effects model is proposed to investigate whether there are observable effects in
the panel data set. As shown in the following table, variables that do not contribute to the
model and therefore those that have a significance level of at least 90% were excluded.
It is emphasized that, of the three fundamental variables of the model, the stakes of
government bonds in pension fund portfolios (bpp), in assets (vi), and in other investment
funds (fund), only the statistical variable (vi) is significant. This behavior is more than
logical because the dependent variable proxies the behavior of the main index of country
i; thus, variable equity has a direct relationship because stock indexes are comprised of
the behavior of the main market shares.
The specification of the fixed effects model under estimates WITHIN is as follows:
Model 4: Fixed Effects Lineal Model
Model 5: Fixed Effects logarithmic Model
Model 6: Fixed Effects Logarithmic Difference Model
17
Table 2
Estimation by Effects Fixed Model
Independ. Var.
Mod. Lineal (Panel A)
Mod. Log. (Panel B)
Mod. Dif_Log. (Panel C)
coef
-0.002
se
0.002
coef
0.012
se
0.056
coef
0.031
se
0.028
-0.007**
0.003
-0.082**
0.041
-0.044*
0.027
0.002
0.001
-0.006
0.059
-0.020
0.026
gei
-0.006***
0.001
0.022
0.042
0.001
0.020
gdp
-0.051***
0.010
-0.014
0.062
0.038
0.035
asgdp
-0.017***
0.004
-0.019
0.068
-0.065
0.136
_cons
3.743***
0.178
2.902***
0.365
2.620***
bpp
vi
gbgdp
Number of observations
R2
0.030
240
119
70
0.353
0.046
0.117
Notes: Estimates are based on equations of the models 4, 5 and 6. The dependent variable is lvol and independent
variables in panel A are taken linear, logarithmics in panel B and logarithmics difference in panel C. The column shows the
standard errors SE.
*** Denote statistical significance at 1% level
** Denote statistical significance at 5% level
* Denote statistical significance at 10% level
According to Table 2, it is evident that the variable vi has statistical significance at 95%
and also that the negative sign of the estimated coefficient indicates an inverse
relationship between market volatility and the participation of pension fund portfolios in
variable income assets. This result is consistent with the results obtained by Thomas et al.
(2014) because the participation of the portfolios of these institutional investors in
variable income assets acts as a market stabilizer. Additionally, the linear model (lineal) is
showing greater statistical significance in different variables.
Specification Analysis
For the fixed effects model, the following specification tests exist:
Table 3
Temporal Intragroup Autocorrelation Contrast
xtserial lvol bpp vi gbgdp gdp gei asgdp
Wooldridge test for autocorrelation in panel data
Ho: no first-orden autocorrelation
F (1, 21) = 1.053
Prob > F = 0.3164
According to the Wooldridge test, which has the null hypothesis that there is no firstorder serial autocorrelation, it is not possible to reject Ho in the case of the proposed
model; consequently, the model does not present serial intragroup autocorrelation
problems.
18
 Heteroscedasticity contrast between groups:
Table 4
Modified Wald test for groupwise heterocedasticity
in fixed effect regression model
xttest3
Ho: sigma(i)^2 = sigma^2 for all i
chi2 (23) = 1.5e+33
Prob > chi2 = 0.000
The results of the Wald test for measuring heteroscedasticity among groups have a null
hypothesis of equal variances (homoscedasticity). According to the Stata output, one can
see that it is necessary to reject Ho; thus, the proposed fixed effects model presents
problems of heteroscedasticity.
Estimation using the Random Effects Model
Given that the fixed effects model presents problems of heteroscedasticity, the same
model is proposed again under the assumption that there are non-observable random
effects.
The specification of the random effects model under Generalized Least Squares (GLS)
estimation is as follows:
Model 7: Random Effects Lineal Model
Model 8: Random Effects Logarithmic Model
Model 9: Random Effects Logarithmic Difference Model
19
Table 5
Estimation by Random Effects Model
Mod. Lineal (Panel A)
Mod. Log. (Panel B)
Mod. Dif_Log. (Panel C)
bpp
coef
0.002*
se
0.001
coef
0.061
se
0.044
coef
0.024
se
0.027
vi
-0.003*
0.002
-0.084***
0.031
-0.036
0.025
0.000
0.001
-0.004
0.044
-0.011
0.026
gei
-0.006***
0.001
0.015
0.037
0.008
0.020
gdp
-0.050***
0.010
-0.013
0.055
0.019
0.033
_cons
3.060***
0.093
2.707***
0.244
2.609***
Independ. Var.
gbgdp
Number of observations
240
119
0.039
70
Notes: Estimates are based on equations of the models 7, 8 and 9. The dependent variable is lvol and independent
variables in panel A are taken linear, logarithmics in panel B and logarithmics difference in panel C. The column shows the
standard errors SE.
*** Denote statistical significance at 1% level
** Denote statistical significance at 5% level
* Denote statistical significance at 10% level
The random effects model shows several interesting results. First, the bpp and vi variables
reach a statistical significance of 90%. The second interesting result is that the variable vi
retains its negative sign as in the estimation under fixed effects. Finally, the bpp variable
shows that the participation of pension fund portfolios in government and corporate
bonds has a positive sign. This result is very important because it means that investments
in this type of asset have a direct and positive impact on market volatility. That is, to the
extent that pension fund portfolios increase or decrease holdings in government and/or
corporate bond assets, market volatility behaves in the same manner. Therefore, it can be
concluded that these investments have a destabilizing effect on market volatility. These
results agree with those obtained by Hu (2006).
Specification Analysis
The following specification tests exist for the random effects model:
 Contrast of error components:
Table 6
Breusch and Pagan Lagrangian multiplier for
random effects
xttest0
lvol(id, t) = Xb + u(id) + e(id, t)
Estimated Results:
Var
sd = sqrt(var)
lvol
0.2631785 0.5130093
e
0.1751124 0.4184644
u
0.0037066 0.0608817
Test: Var(u) = 0
chibar2(01) = 3.46
Prob > chibar2 = 0.0314
20
The Breusch and Pagan (1979) test proposes, as a null hypothesis, that the error is simple
and, as an alternate hypothesis, that the error has two components: an idiosyncratic
component and another component that is associated with individuals. According to this
test, Ho is not rejected at the 99% confidence level (CL) but is rejected at the 95% and
lower CLs.
 Hausman Test:
The Hausman test propose by Hausman (1978) helps to identify which model is the most
appropriate, the fixed effects model or the random effects model.
Table 7.
Hausman Fixed Effects vs. Random Effects
Fixed Effects (b)Random Effects (B)Difference (b - B)
SE
bpp
-0.0014186
0.0024057
-0.0038243
0.001157
vi
-0.0039649
-0.0033874
-0.0005775
0.0023434
gbgdp
0.0015843
0.0004851
0.0010992
0.00075
gdp
-0.0510018
-0.0497182
-0.0012836
0
gei
-0.0063008
-0.0062627
-3.81E-05
0.0021583
b = consistent under Ho and Ha; obtained from xtreg
B = inconsistent under Ha, efficient under Ho; obtained from xtreg
Test: Ho: difference in coefficient not systematic
chi2(5) = (b - B)' [(V_b - V_B)^(-1)](b - B)
chi2(5) = 17.96
Prob>chi2 = 0.0030
(V_b - V_B is not positive definite)
Table 8.
Hausman Fixed Effects vs. Random Effects
Fixed Effects (b) Random Effects (B) Difference (b - B)
SE
bpp
-0.0018178
0.0015275
-0.0033453
0.0009686
vi
-0.0070968
-0.0023384
-0.0047584
0.002196
gbgdp
0.0018214
0.0005606
0.0012608
0.0006616
gei
-0.0058729
-0.0062635
0.0003906
0
gdp
-0.0507633
-0.05113
0.0003667
0
asgdp
-0.0167267
-0.0015196
-0.0152071
0.0040451
b = consistent under Ho and Ha; obtained from xtreg
B = inconsistent under Ha, efficient under Ho; obtained from xtreg
Test: Ho: difference in coefficient not systematic
chi2(6) = (b - B)' [(V_b - V_B)^(-1)](b - B)
chi2(5) = 29.78
Prob>chi2 = 0.0000
(V_b - V_B is not positive definite)
The Hausman test is applied to two types of specifications. The first specification does
not contain the asgdp variable in the fixed effects model or in the random effects model.
The second specification contains this variable in both models. According to the results of
the Hausman test, the null hypothesis that states that there is no systematic difference
21
between the estimated coefficients under fixed effects or random effects is rejected. This
result indicates that, of the two models, the fixed effects model is the most appropriate.
However, taking into account the problems of heteroscedasticity presented by the fixed
effects model, in this particular case, it is recommended to take into account the results of
the random effects model.
The first stage of the empirical study ends with the conclusion that, of the three most
important variables in the proposed model, only two reached statistical significance; thus,
the bpp and vi variables are kept in the models and the fund variable is rejected.
Regarding the signs of the variables that remain in the model, the results are mixed; the
variable bpp has a positive sign, which means that there is a positive and direct
relationship between the investments of pension fund portfolios in government and
corporate bonds and market volatility. In the case of the vi variable, which has a negative
sign, the variable shows a negative relationship between the proportions of investments in
variable income assets that pension fund portfolios maintain and market volatility.
Estimates Using Dynamic Models
As noted above, dynamic models involve the time variable, i.e., they provide insight into
the behavior of both dependent and independent variables over time.
Table 9.
Estimation of the Dynamic Model, one step, two lags, orthogonal
regressors, predetermined and contemporaneously endogenous.
Mod_din
Independ. Var.
coef
se
L.lvol
0.748***
0.053
L2.lvol
-0.525***
0.069
diflgbgdp
0.050***
0.007
diflinfla
-0.197***
0.035
diflgdp
0.030***
0.006
diflbpp
0.065***
0.015
L.diflbpp
0.025**
0.012
diflvi
0.020**
0.008
L.diflvi
0.013*
0.007
diflegi
-0.008
0.008
diflintreal
-0.021
0.016
diflfund
-0.036***
0.012
diflmcgdp
0.052***
0.012
diflinasgdp
-0.785***
0.145
_cons
2.198***
0.180
The letter L denotes the contemporary variable. The letters L1 and L2 denotes the
lagged variable periods 1 and 2 respectively
*** Denote statistical significance at 1% level
** Denote statistical significance at 5% level
* Denote statistical significance at 10% level
22
The specification of the dynamic model to be estimated by GMM is as follows:
The estimation results show several interesting features. First, the three most important
variables of the model reach statistical significance, both in their early lags and in the
value of the contemporary coefficient. Second, when volatility, which in principle is the
dependent variable, is dynamically modeled, i.e., the same variable is lagged for two
periods, it becomes an independent or explanatory variable and also reaches statistical
significance.
The signs of the lvol variable coefficients are mixed; however, it is highlighted that their
values are of significant magnitudes in both the first lag (0.748) and the second lag (0.525). This result has a high correspondence level with the theory of portfolio selection,
in this specific case, of portfolios belonging to pension funds. Regarding this theory, it is
important to highlight the function of structuring and managing an investment portfolio.
This is a dynamic process that changes as economic environment variables require.
However, taking into account the long-term nature of this type of portfolio, it is assumed
that the variation should not be so great. This assumption is the key reason for justifying
the annual observations used in this study.
However, the statistical significance of the lagged variable lvol confirms that there is
enough statistical evidence that the estimated coefficients for these variables are different
from zero at the 99% CL.
The coefficients of the diflbpp variable reach statistical significance at the 99% CL in its
contemporary coefficient and the 95% CL in its first lag. Most importantly, the variable
maintains a positive sign in both cases, which shows a direct relationship with market
volatility.
The coefficients of the diflvi variable reach statistical significance at the 95% CL in its
contemporary variable and the 90% CL in its first lag; however, one sees that the signs in
both the contemporary variable and its first lag are positive. This result is different from
that found in the fixed estimates; however, given the greater robustness of the dynamic
estimation, this result is preferable to the fixed estimate result.
The only variables that do not reach statistical significance and that therefore do not
contribute to the model are diflgei and diflintreal. All other variables are significant.
Specification Analysis
The specification analysis of the dynamic panel data models consist of applying two tests.
The first test is autocorrelation, and the second test is orthogonality, as shown below.
Table 10.
Arellano-Bond test for zero autocorrelation in
first difference errors
Order
Z
Prob > Z
1
1.379
0.1679
2
-0.34029
0.7336
Ho: No autocorrelation
23
The autocorrelation tests developed by Arellano and Bond (1991) are based on sample
correlation coefficients to examine the hypothesis of non-correlation in the GMM
residues of a dynamic model. The null hypothesis (Ho) of this test suggests that there is
no j-level correlation in GMM residues. The alternative hypothesis (Ha) suggests that
there is a j-level correlation in GMM residues.
The Stata output and statistical values do not reject the null hypothesis, which furnishes
evidence that there is no j-level correlation in either first-order or second-order GMM
residues.
The Sargan (1958) test reveals whether the model is correctly identified, i.e., whether
orthogonality assumptions are met to estimate the k parameters. The null hypothesis (Ho)
argues that orthogonality restrictions are appropriate, and the alternative hypothesis (Ha)
argues that orthogonality restrictions are not appropriate.
Table 11.
Sargan test of overidentification restrictions
Ho: overidentification restrictions are valid
chi2 (4) = 5.824415
Prob > chi2 = 0.2127
According to the Stata output, the value of the test statistic indicates that one cannot
reject Ho; therefore, the model is correctly specified.
5.
Conclusions
The growth and development that pension funds have experienced over the last 12 years
are undeniable. One of the most significant factors in this development is represented by
the volume of managed resources, which in some countries reaches almost 200% of their
GDP. This significant amount of managed resources has allowed this type of institutional
investor to increasingly strengthen its presence and importance in financial markets for
stocks and bonds.
The main purpose of this study centered on confirming the hypothesis that pension funds
act as important players in financial markets and have an impact on their stability.
Further, rather than acting as stabilizers, their actions of buying and selling financial
assets allow them to inject volatility shocks that could destabilize the markets in which
they participated. For this purpose, it was primarily investigated whether variations in the
participation of pension fund portfolios positively or negatively impacted market
volatility.
Through the empirical exercise, it was possible to verify that such variations in pension
fund portfolios in countries belonging to the OECD have a direct and positive
relationship with market stability, as measured by market volatility.
Consequently, the main conclusion is that these institutions do in fact impact the stability
of the market, making them more or less volatile, according to the positions that these
institutional investors hold in their portfolios. The results are presented at the aggregate
level for the sample countries and for the period analyzed.
24
6.
Bibliography
Arellano, M., Bond, S. (1991). Some Tests of Specification for Panel Data: Monte Carlo
Evidence and an Application to Employment Equations. Review of Economic
Studies. 58 (2), 277-297.
Bachelier, L. (1914). Le Jeu, la Chance et le Hasard (The Game, the Chance and the
Hazard). Bibliotheque de Philosophie Scientifique, Ernest Flammarion, Paris.
Reprinted by Editions Jacques Gabay, Paris, 1993.
Bacchetta, P., Caminal, R., (2000). Do capital market imperfections exacerbate output
fluctuations? European Economic Review, 44 (3), 449-468.
Beaver, W. H. (1981). Market Efficiency. The Accounting Review, 56 (1), 23-37.
Beck, T., Lundberg, M., Majnoni, G., (2006). Financial intermediary development and
growth volatility: do intermediaries dampen or magnify shocks? Journal of
International Money and Finance, 25 (7), 1146-1167.
Blume, M., Keim, D., (2012). Institutional investors and stock market liquidity: Trends
and relationships. Working Paper, Available at SSRN 2147757.
Bohl, M. T., Brzeszczynski, J., Wilfling, B., (2009). Institutional investors and stock
returns volatility: empirical evidence from a natural experiment. Journal of Financial
Stability, 5 (2), 170–182.
Breusch, T. S., Pagan, A. R. (1979). A simple test for heteroscedasticity and random
coefficient variation. Econometrica, 47 (5), 1287-1294.
Cai, Fang, Zhen, Lu. (2004). Institutional trading and stock returns. Finance Research
Letters, 1, 178-189.
Campbell, J. Y., Shiller, J. R. (1988). Stock prices, earnings and expected dividends. The
Journal of Finance, 43 (3), 661-676.
Campbell, J. Y., Kyle, A. S., (1993). Smart money, noise trading and stock Price
behaviour. The Review of Economic Studies, 60 (1), 1-34.
Choe, H., Kho, B. C., Stulz, R. M., (1999). Do foreign investors destabilize stock
markets? The Korean experience in 1997. Journal of Financial Economics, 54 (2),
227-264.
Davis, E. Philip. (2005). The role of pension funds as institutional investors in emerging
markets. Paper provided by Economics and Finance Section, School of Social
Sciences, Brunel University in its series Economics and Finance Discussion Papers.
# 05-18.
Davis, E., Hu, Yu. (2005). Is there a link between pension fund assets and economic
growth? A cross country study. Brunel University and NIESR, mimeo.
De Bondt, W. F. M., Thaler, R. (1985). Does the stock market overreact? The Journal of
Finance, 40 (3), 793-805.
De Long, J. B., Shleifer, A., Summers, L. H., Waldmann, R. J. (1990). Noise trader risk in
financial markets. The Journal of Political Economy, 98 (4), 703-738.
Dennis, P. J., Strickland, D. (2002). Who blinks in volatile markets, individuals or
institutions? The Journal of Finance, 57 (5), 1923-1949.
Dennis, Patrick, Weston, James. (2001). Who’s informed? An analysis of stock ownership
and informed trading. Working paper, University of Virginia and Rice University.
Fama, E. F. (1965). The behavior of stock market prices. The journal of business, 38 (1),
34-105.
25
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. The
Journal of Finance, 25 (2), 383-417.
Fama, E. F., French, K. R. (1988). Permanent and temporary components of stock prices.
Journal of Political Economy, 96 (2), 246-273.
Figlewski, S. (1979). Subjective information and market efficiency in a betting market.
Journal of Political Economy, 87 (1), 75-88.
Gabaix, X., Gopikrishnan, P., Plerou, V., Stanley, H. E. (2006). Institutional investors and
stock market. The Quarterly Journal of Economics, 461-504.
Gompers, P.A., Metrick, A. (2001). Institutional investors and equity prices. The
Quarterly Journal of Economics, 116 (1), 229-259.
Grossman, S. (1976). On the efficiency of competitive stock markets where traders have
diverse information. The Journal of Finance, 31 (2), 573-585.
Grossman, S. J., Stiglitz, J. E. (1980). On the impossibility of informationally efficient
markets. The American Economic Review, 70 (3), 393-408.
Hausman, J. A. (1978). Specification test in econometrics. Econometrica, 46, 1251 - 1271
Hu, Y. (2006). The impact of pension funds on financial markets. Financial Market
Trends, 2, 145-162.
Jensen, M. C. (1968). The performance of mutual funds in the period 1945-1964. The
Journal of Finance, 23 (2), 389-416.
Kahneman, D., Tversky, A. (1982). Intuitive Prediction: Biases and Corrective
Procedures. In D. Kahneman, P. Slovic, and A. Tversky, (eds.), Judgment Under
Uncertainty: Heuristics and Biases. London: Cambridge University Press.
Lakonishok, J., Shleifer, A., Vishny, R. W. (1992). The impact of institutional trading on
stock prices. Journal of Financial Economics, 32 (1), 23-43.
Laffont, J. J., Maskin, E. S. (1990). The efficient market hypothesis and insider trading on
stock market. Journal of Political Economy, 98 (1), 70-93.
LeRoy, S. F., Porter, R. D. (1981). The present-value relation: Tests based on implied
variance bounds. Econometrica, 49 (3), 555-574.
Lo, A. W., MacKinlay, A. C. (1988). Stock market prices do not follow random walks:
Evidence from a simple specification test. The Review of Financial Studies, 1 (1),
41-66.
Lipson, M., Puckett, A. (2006). Volatile Markets and Institutional Trading. Unpublished
Working Paper. University of Missouri, Darden Graduate School of Business.
Malkiel, B. G. (1973). A Random Walk Down Wall Street. Norton, New York.
Malkiel, B. G. (1989). Is the stock market efficient? Science New Series, 243 (4896),
1313-1318
Malkiel, B. G. (2003). The efficient market hypothesis and its critics. The Journal of
Economic Perspectives, 17 (1), 59-82.
Osborne, M. F. M. (1959). Brownian motion in the stock market. Operations Research, 7
(2), 145-73.
Pension Markets in Focus. Annual Report of OECD. Available at:
www.oecd.org/daf/pensions/pensionmarkets. P. 7-11.
Poterba, J. M., Summers, L. H. (1988). Mean reversion in stock prices: Evidence and
implications. Journal of Financial Economics, 22 (1), 27-59.
Roberts, H. (1967). Statistical versus clinical prediction of the stock market, Unpublished
manuscript.
26
Rubin, A., Smith, D. R. (2009). Institutional ownership, volatility and dividends. Journal
of Banking and Finance, 33 (4), 627-639.
Samuelson, P. A. (1965). Proof that properly anticipated prices fluctuate randomly.
Industrial Management Review, 6 (2), 41-49.
Samuelson, P. A. (1973a). Mathematics of speculative Price. SIAM Review, 15 (1), 1-42.
Samuelson, P. A. (1973b). Proof that properly discounted present values of assets vibrate
randomly. The Bell Journal of Economics and Management Science, 4 (2), 369-374.
Sargan, J. D. (1958). The estimation of economics of relationship using instrumental
variables. Econometrica, 26, 393-415.
Scholes, M. S. (1972). The market for securities: Substitution versus price pressure and
the effects of information on share prices. The Journal of Business, 45 (2), 179-211.
Sias, R. W. (1996). Volatility and the institutional investor. Financial Analysts Journal, 52
(2), 13-20.
Shiller, R. J. (1979). The volatility of long-term interest rates and expectations models of
the term structure. Journal of Political Economy, 87 (6), 1190-1219.
Shiller, R. J. (1981). Do stock prices move too much to be justified by subsequent
changes in dividends? The American Economic Review, 71 (3), 421-436.
Shiller, R. J. (1987). The volatility of stock market prices. Science, 235 (4784), 33-37.
Shiller, R. J. (1988). The volatility debate. Agricultural and Applied Economics
Association, 70 (5), 1057-1063.
Shiller, R. J. (1990). Market volatility and investor behavior. Papers and Proceedings of
the Hundred and Second Annual Meeting of the American Economic Association.
The American Economic Review, 80 (2), 58-62.
Shiller, R. J. (2000). Irrational Exuberance. Princeton University Press, Princeton, NJ.
Stiglitz, J. E. (1981). The allocation role of the stock market: Pareto optimality and
competition. The Journal of Finance, 36 (2), 235-251.
Thomas, A., Spataro, L., Mathew, N. (2014). Pension funds and stock market volatility:
An empirical analysis of OECD countries. Journal of Financial Stability, 11, 92-103.
Walker, E., Lefort, F. (2002). Pension reform and capital markets: are there any (hard)
links? Revista ABANTE, 5 (2), 77-149.
Wermers, R. (1999). Mutual fund herding and the impact on stock prices. The Journal of
Finance, 54 (2), 581-622.
Xu, Y., Malkiel, B. G. (2003). Investigating the behavior of idiosyncratic volatility.
Journal of Business, 76, 613-644.
27