View
2
Download
0
Category
Preview:
Citation preview
Who are the Value and Growth Investors?∗
Sebastien Betermier, Laurent E. Calvet, and Paolo Sodini
First version: October 2013This version: May 2015
Abstract
This paper investigates value and growth investing in a large administrative panel of Swedishresidents over the 1999-2007 period. We show that over the life-cycle, households progres-sively shift from growth to value as they become older and their balance sheets improve. Weverify that households climb the value ladder by actively rebalancing their stock and fundholdings. Furthermore, investors with low human capital and low exposure to aggregate la-bor income shocks tilt their portfolios toward value. While several behavioral biases seemevident in the data, the patterns we uncover are overall strikingly consistent with risk-basedexplanations of the value premium.
∗Betermier: Desautels Faculty of Management, McGill University, 1001 Sherbrooke St West, Montreal, QC H3A1G5, Canada, sebastien.betermier@mcgill.ca. Calvet: Department of Finance, HEC Paris, 1 rue de la Libération,78351 Jouy-en-Josas Cedex, France; calvet@hec.fr. Sodini: Department of Finance, Stockholm School of Economics,Sveavägen 65, Box 6501, SE-113 83 Stockholm, Sweden, Paolo.Sodini@hhs.se. We are grateful to Kenneth Single-ton (the Editor), the Associate Editor, and an anonymous referee for many insightful comments. We thank StephanJank, Claus Munk, Per Östberg, Jonathan Parker, Sébastien Pouget, and Shaojun Zhang for helpful discussions, andacknowledge constructive comments from Laurent Barras, John Campbell, Chris Carroll, Luigi Guiso, Marcin Kacper-czyk, Bige Kahraman, Hugues Langlois, Alex Michaelides, Ben Ranish, David Robinson, Johan Walden, and seminarparticipants at the City University of Hong Kong, Copenhagen Business School, HEC Montréal, HEC Paris, Impe-rial College Business School, Lund University, McGill University, Peking University, the Securities and ExchangeCommission, the Swedish School of Economics, the Toulouse School of Economics, Université de Sherbrooke, theUniversity of Helsinki, the University of Southern Denmark, the Norges Bank Household Finance Workshop, the 2014IFM2 Math Finance Days, the 2014 China International Conference in Finance, the 2014 NBER Summer Institute As-set Pricing Workshop, the 2014 European Conference on Household Finance, and the 2015 Cologne Colloquium onFinancial Markets. We thank Statistics Sweden and the Swedish Twin Registry for providing the data. The projectbenefited from excellent research assistance by Milen Stoyanov, Tomas Thörnqvist, Pavels Berezovkis, and espe-cially Andrejs Delmans. This material is based upon work supported by Agence Nationale de la Recherche, BFI, theHEC Foundation, Riksbank, the Social Sciences and Humanities Research Council of Canada, and the Wallander andHedelius Foundation.
Who are the Value and Growth Investors?
ABSTRACT
This paper investigates value and growth investing in a large administrative panel of
Swedish residents over the 1999-2007 period. We show that over the life-cycle, house-
holds progressively shift from growth to value as they become older and their balance
sheets improve. We verify that households climb the value ladder by actively rebal-
ancing their stock and fund holdings. Furthermore, investors with low human capital
and low exposure to aggregate labor income shocks tilt their portfolios toward value.
While several behavioral biases seem evident in the data, the patterns we uncover are
overall strikingly consistent with risk-based explanations of the value premium.
JEL Classification: G11, G12.
Keywords: Asset pricing, value premium, household finance, portfolio allocation, hu-
man capital, factor-based investing.
1 Introduction
A large academic and practitioner literature documents that value stocks outperform growth stocks
on average in the United States and around the world (Basu 1977, Fama and French 1992, 1998,
Graham and Dodd 1934).1 The economic explanation of these findings is one of the central
questions of modern finance. The value premium may be a compensation for forms of system-
atic risk other than market portfolio return risk (Fama and French 1992, 1995), such as aggre-
gate labor income and consumption shocks (Cochrane 1999, Jagannathan and Wang 1996, Let-
tau and Ludvigson 2001, Petkova and Zhang 2005, Yogo 2006),2 cash-flow risk (Campbell and
Vuolteenaho 2004), long-run consumption risk (Bansal, Dittmar, and Lundblad 2005, Hansen,
Heaton, and Li 2008),3 the costly reversibility of physical capital (Zhang 2005), or displacement
risk (Garleanu, Kogan, and Panageas 2012).4 The underperformance of growth stocks relative to
value stocks may also be evidence that investors are irrationally exuberant about the prospects of in-
novative glamour companies (DeBondt and Thaler 1985, Lakonishok, Shleifer, and Vishny 1994).5
The extensive empirical literature on the value premium focuses primarily on stock returns and
their relationships to macroeconomic and corporate data. Disentangling theories of the value pre-
mium, however, has proven to be challenging on traditional data sets that do not provide individual
positions and therefore do not permit researchers to assess the determinants of investor decisions.6
The present paper proposes to use the rich information in investor portfolios to shed light on ex-
planations of the value premium. We investigate value and growth investing in a highly detailed
1See also Asness, Moskowitz, and Pedersen (2013), Ball (1978), Basu (1983), Capaul, Rowley, and Sharpe (1993),Chan, Hamao, and Lakonishok (1991), Fama and French (1993, 1996, 2012), Griffin (2003), Liew and Vassalou(2000), and Rosenberg, Reid, and Lanstein (1985). Some recent work also shows that the strength of the valuepremium can be improved by refining the sorting methodology (Asness and Frazzini 2013, Barras 2013, Hou, Karolyi,and Kho 2011).
2Eiling (2013), Jagannathan, Kubota, and Takehara (1998), Addoum, Korniotis, and Kumar (2013), and Santosand Veronesi (2006) provide further evidence on the relationship between labor income and the value premium.
3See also Bansal, Kiku, Shaliastovich, and Yaron (2014), Bansal, Dittmar, and Kiku (2009), and Gulen, Xing, andZhang (2011).
4Other forms of countercyclical risk can contribute to explaining the value premium. For instance, the variance ofidiosyncratic labor income risk is high during recessions (Storesletten, Telmer, and Yaron 2004) and value stocks tendto provide low dividends when the aggregate housing collateral is low (Lustig and van Nieuwerburgh 2005).
5See also Barberis and Thaler (2003), Daniel, Hirshleifer, and Subrahmanyam (2001), La Porta, Lakonishok,Shleifer, and Vishny (1997), and Shleifer (2000).
6See Liu, Lu, Sun, and Yan (2015) for a recent discussion.
1
administrative panel, which contains the disaggregated holdings and socioeconomic characteris-
tics of all Swedish residents between 1999 and 2007. The data set reports portfolio holdings at the
level of each stock or fund, along with other forms of wealth, debt, labor income, and employment
sector.
The paper makes four main contributions to the literature. First, we show that the value tilt
exhibits substantial heterogeneity across households. When we sort investors by the value tilt of
their risky asset portfolios, the difference in expected returns between the top and bottom deciles
is approximately equal to the value premium. Over the life-cycle, households climb the “value
ladder”, i.e. they gradually shift from growth to value investing as their investment horizons and
financial circumstances evolve. The value ladder is made possible by active rebalancing, which
allows households to mitigate the impact of realized returns and revert to their slow-moving target.
The positive relationship between age and the value loading is also evident among new participants,
whose portfolios are not passively affected by past returns.
Second, we relate the value tilt to household characteristics. We show that value investors are
substantially older, have higher financial wealth, higher real estate wealth, lower leverage, lower
income risk, lower human capital, and are also more likely to be female, than the average growth
investor. By contrast, men, entrepreneurs, and educated investors are more likely to invest in
growth stocks. These baseline patterns are evident both in stock and mutual fund holdings, and
are robust to controlling for the length of risky asset market participation and other measures of
financial sophistication. The explanatory power of socioeconomic characteristics is especially high
among households that invest directly in at least five companies, a wealthy subgroup that owns the
bulk of aggregate equity.
Third, we show that households adjust their portfolio value loadings to systematic risk in their
employment sectors. We uncover a strong factor structure in the panel of industry per-capita in-
come growth and show that a single macroeconomic factor, per-capita aggregate income growth,
explains on average 88% of the time-series variation of per-capita income in any given 2-digit SIC
industry. Households employed in sectors with high exposures to the macroeconomic factor tend
to select portfolios of stocks and funds with low value loadings. We obtain similar results when
2
we measure systematic risk by using industry exposures to the value factor.
Fourth, we verify that our results are robust to a large number of alternative hypotheses which
include financial market experience or stock characteristics other than the value loading, such as
professional proximity, the dividend yield, taxes, firm age, skewness, and size. We document that
the equities most widely held by households are a mix of growth stocks and value stocks, and
that the relationships between portfolio tilts and investor characteristics are unlikely to be driven
by these stocks. As in Calvet and Sodini (2014), we consider the subsample of Swedish twins
to control for latent investor fixed effects, such as family background, upbringing, inheritance, or
attitudes toward risk. The sensitivities of the value loading to socioeconomic characteristics are
similar in the twin subsample as in the general household population, regardless of whether or not
the twins communicate frequently or infrequently with each other.
The patterns that we uncover in the Swedish portfolio data appear remarkably consistent with
the implications of risk-based theories for the cross-section of portfolio tilts. Value stocks are held
by households that are in the best position to take systematic financial risk, for instance because
they have sound balance sheets or low background risk. We document that young households
with long investment horizons go growth and progressively migrate toward value as they get older,
which provides strong empirical support for intertemporal hedging explanations (Jurek and Viceira
2011, Larsen and Munk 2012, Lynch 2001). In addition, investors with high human capital and
high exposure to aggregate labor income shocks tilt away from value, which is in line with labor-
based theories of the value premium. These empirical regularities are stronger among households
that invest directly in at least five stocks, have high risky shares, and own the bulk of aggregate
equity.
The Swedish data set provides highly detailed information on household finances and demo-
graphics but is somewhat less informative about psychological traits. With this caveat, we find that
behavioral explanations of the value premium also help to explain the portfolio evidence. Overcon-
fidence, which is more prevalent among men than women (Barber and Odean 2001), is consistent
with the preference of male investors for growth stocks. As attention theory predicts (Barber and
Odean 2008), a majority of direct stockholders hold a small number of popular stocks. Further-
3
more, some of the empirical regularities documented in the paper can receive complementary risk-
based and psychological explanations. For instance, the tilt of entrepreneurs toward growth stocks
can be attributed both to a marked overconfidence in own decision-making skills (Busenitz and
Barney 1997) and to exposure to private business risk (Moskowitz and Vissing-Jørgensen 2002).
The evidence reported in this paper contributes to the growing body of work showing the rel-
evance of portfolio theory for explaining household financial behavior. Retail investors allocate
a high share of liquid financial wealth to risky assets if they have high financial wealth and hu-
man capital (Calvet and Sodini 2014), earn safe labor incomes (Betermier, Jansson, Parlour, and
Walden 2012, Calvet and Sodini 2014, Guiso, Jappelli, and Terlizzese 1996), and are not en-
trepreneurs (Heaton and Lucas 2000).7 Households actively rebalance their financial portfolios
in response to realized returns (Calvet, Campbell, and Sodini 2009a). Furthermore, a major-
ity of households incur small welfare losses from underdiversification (Calvet, Campbell, and
Sodini 2007). We document here that financial theory also accounts for the cross-sectional and
time-series properties of household value tilts.
Our results complement the literature showing that retail investors favor assets with certain
characteristics, such as familiar stocks,8 and adjust their investment styles to news and past ex-
perience (Kumar 2009a, Campbell, Ramadorai, and Ranish 2014). The Swedish panel contains
high-quality data on holdings and socioeconomic characteristics, which allows us to uncover new
patterns in the household demand for value and growth stocks. The paper also sheds light on the
potential influence of genes. Cronqvist, Siegel, and Yu (2015) estimate a variance decomposition
of the portfolios held by twins and conclude that value investing has a strong genetic component.
The present paper demonstrates that the so-called “genetic” component is negligible among twins
that communicate infrequently with each other, which suggests that simple variance decomposi-
tions severely overestimate the impact of genes. The value tilt is not simply encoded in the DNA
of retail investors, but is also strongly driven by financial circumstances and communication.
7See also Angerer and Lam (2009), Bonaparte, Korniotis, and Kumar (2014), and Knüpfer, Rantapuska, andSarvimäki (2013).
8Households are known to favor stocks that are familiar (Døskeland and Hvide 2011, Huberman 2001, Massa andSimonov 2006), geographically and culturally close (Grinblatt and Keloharju 2001), attention-grabbing (Barber andOdean 2008), or connected to products they consume (Keloharju, Knüpfer, and Linnainmaa 2012).
4
The rest of the paper is organized as follows. Section 2 presents the data and defines the main
variables. Section 3 reports the cross-sectional distribution of the value loading. Section 4 docu-
ments the value ladder and empirically investigates how the value tilt relates to the demographic
and financial characteristics of households. Section 5 relates the evidence to risk- and sentiment-
based explanations of the value premium. Section 6 presents robustness checks and Section 7
concludes. An Internet Appendix (Betermier, Calvet, and Sodini 2015) develops an equilibrium
model of the cross-section of portfolio tilts, discusses estimation details, and reports additional
empirical results.
2 Data and Construction of Variables
2.1 Local Fama and French Factors
Data on Nordic stock markets for the 1985 to 2009 period are available from FINBAS, a financial
database maintained by the Swedish House of Finance. The data include monthly stock returns,
market capitalizations at the semiannual frequency, and book values at the end of each year. Free-
float adjusted market shares are available from Datastream. We focus on stocks with at least two
years of available data. We exclude stocks worth less than 1 krona, which filters out very small
firms. For comparison, the Swedish krona traded at 0.1371 U.S. dollar on 30 December 2003. We
end up with a universe of approximately 1,000 stocks, out of which 743 are listed on one of the
four major Nordic exchanges in 2003.9
The return on the market portfolio is proxied by the SIX return index (SIXRX), which tracks
the value of all the shares listed on the Stockholm Stock Exchange. The risk-free rate is proxied
by the monthly average yield on the one-month Swedish Treasury bill. The market factor MKTt is
the market return minus the risk-free rate in month t. The local value, size, and momentum factors
are constructed as in Fama and French (1993) and Carhart (1997). We sort the stocks traded on
the major Nordic exchanges by book-to-market value, market size, and past performance. We use
9The major Nordic exchanges are the Stockholm Stock Exchange, the Copenhagen Stock Exchange, the HelsinkiStock Exchange, and the Oslo Stock Exchange.
5
these bins to compute the value factor HMLt , the size factor SMBt , and the momentum factor
MOMt , as is fully explained in the Internet Appendix.
We index stocks and funds by i ∈ {1, . . . , I}. For every asset i, we estimate the four-factor
model:
rei,t = ai +bi MKTt + vi HMLt + si SMBt +mi MOMt +ui,t , (1)
where rei,t denotes the excess return of asset i in month t and ui,t is a residual uncorrelated to the
factors. Estimated loadings are winsorized at -5 and +5. The value premium is substantial in
Sweden: HMLt averages to about 10% per year over the 1985 to 2009 period, which is consistent
with the Sweden estimate in Fama and French (1998) and also in the range of country estimates
reported in Liew and Vassalou (2000). In the Internet Appendix, we also show the Swedish HML
factor shares many similar properties with its U.S. equivalent.
2.2 Household Panel Data
The Swedish Wealth Registry is an administrative data set compiled by Statistics Sweden, which
has previously been used in household finance (Calvet, Campbell, and Sodini 2007, 2009a, 2009b,
Calvet and Sodini 2014). Statistics Sweden and the tax authority had until 2007 a parliamen-
tary mandate to collect highly detailed information on every resident. Income and demographic
variables, such as age, gender, marital status, nationality, birthplace, education, and municipality
of residence, are available on December 31 of each year from 1983 to 2007. The disaggregated
wealth data include the worldwide assets owned by the resident at year-end from 1999 to 2007.
Real estate, debt, bank accounts, stockholdings, and mutual fund investments are observed at the
level of each property, account, or security. Statistics Sweden provides a household identification
number for each resident, which allows us to group residents by living units. The age, gender,
education and immigration variables used in the paper refer to the household head.
We focus on households that participate in risky asset markets. Unless stated otherwise, the
results are based on representative random sample of approximately 70,000 households observed
at the yearly frequency between 1999 and 2007. The data requirements imposed on households
6
and the method used to construct the random panel are fully explained in the Internet Appendix.
We also use a twin panel from the Swedish Twin Registry, the largest twin database in the
world. The registry provides the genetic relationship (fraternal or identical) of each pair and the in-
tensity of communication between the twins. We have merged the twin data base with the Swedish
Wealth Registry, so that all financial and demographic characteristics are available for the twin
panel.
2.3 Definition of Main Variables
2.3.1 Financial Assets and Real Estate
We use the following definitions throughout the paper. Cash consists of bank account balances and
Swedish money market funds.10 Risky mutual funds refer to all funds other than Swedish money
market funds. Risky financial assets consist of directly held stocks and risky mutual funds. We
exclude assets with less than 3 months of return data.
For every household h, the risky portfolio contains risky financial assets. The risky share is
the fraction of risky financial assets in the portfolio of cash and risky financial assets. A market
participant has a strictly positive risky share.
The value loading of the risky portfolio at time t is the weighted average of individual asset
loadings:
vh,t =I
∑i=1
wh,i,tvi, (2)
where wh,i,t denotes the weight of asset i in household h’s risky portfolio at time t. We will oc-
casionally call vh,t the HML loading or the value tilt. The value loadings of the fund and stock
portfolios are similarly defined. The estimation methodology takes advantage of (i) the detailed
yearly data available for household portfolios, which permit the calculation of wh,i,t , and (ii) the
10Financial institutions are required to report the bank account balance at year-end if the account yields less than100 Swedish kronor during the year (1999 to 2005 period), or if the year-end bank account balance exceeds 10,000Swedish kronor (2006 and 2007). We impute unreported cash balances by following the method used in Calvet,Campbell, and Sodini (2007, 2009a, 2009b) and Calvet and Sodini (2014), as is explained in the Internet Appendix.
7
long monthly series available for individual assets, which permit the precise estimation of vi. Un-
der the unconditional pricing model (1), individual firms have constant value loadings, vi, so that
time variation in household portfolio loading, vh,t , in (2) are driven exclusively by time variation
in portfolio weights. Thus, in Section 4, our estimates of active management of the value tilt by
households will not be contaminated by exogenous changes in firm tilts over the 1999-2007 sample
period.
We measure the household’s financial wealth at date t as the total value of its cash holdings,
risky financial assets, directly held bonds, capital insurance, and derivatives, excluding from con-
sideration illiquid assets such as real estate or consumer durables, and defined contribution re-
tirement accounts. Also, our measure of wealth is gross financial wealth and does not subtract
mortgage or other household debt. Residential real estate consists of primary and secondary res-
idences, while commercial real estate consists of rental, industrial and agricultural property. The
leverage ratio is defined as the household’s total debt divided by the household’s financial and real
estate wealth.
2.3.2 Human Capital
We consider a labor income specification based on Carroll and Samwick (1997) that accounts for
the persistence of income shocks. Specifically, we assume that the real income of household h in
year t, denoted by Lh,t , satisfies
log(Lh,t) = ah +b′xh,t +θh,t + εh,t , (3)
where ah is a household fixed effect, xh,t is a vector of age and retirement dummies, θh,t is a persis-
tent component, and εh,t is a transitory shock distributed as N (0,σ2ε,h). The persistent component
θh,t follows the autoregressive process:
θh,t = ρh θh,t−1 +ξh,t ,
8
where ξh,t ∼ N (0,σ2ξ,h) is the persistent shock to income in period t. The Gaussian innovations
εh,t and ξh,t are white noise and are uncorrelated with each other at all leads and lags. We conduct
the estimation separately on household bins sorted by (i) immigration status, (ii) gender, and (iii)
educational attainment. We estimate the fixed-effects estimators of ah and b in each bin, and then
compute the maximum likelihood estimators of ρh, σ2ξ,h and σ2
ε,h using the Kalman filter on each
household income series.
As is customary in the portfolio-choice literature (e.g., Cocco, Gomes, and Maenhout 2005),
we assume that the household observes both the persistent and transitory components of income.
At a given date t−1, the household knows the contemporaneous component θh,t−1 and next-period
characteristics xh,t . The period-t log labor income, log(Lh,t), therefore has conditional stochastic
component
ηh,t = ξh,t + εh,t , (4)
and conditional variance
σ2h = Vart−1(ηh,t) = σ2
ξ,h +σ2ε,h.
We call σh the conditional volatility of income and use it as a measure of income risk.
We define expected human capital as
HCh,t =Th
∑n=1
Πh,t,t+n
Et(Lh,t+n)
(1+ r)n, (5)
where Th denotes the difference between 100 and the age of household h at date t, and Πh,t,t+n
denotes the probability that the household head h alive at t is still alive at date t + n. We make
the simplifying assumption that no individual lives longer than 100. The survival probability is
imputed from the life table provided by Statistics Sweden. The discount rate r is set equal to 5% per
year. We have verified that our results are robust to alternative choices of r. Detailed descriptions
of the labor income and human capital imputations are provided in the Internet Appendix.
9
2.4 Summary Statistics on Participating Households
Table I reports summary statistics on risky asset market participants (first set of columns), mutual
fund owners (second set of columns), direct stockholders (third set of columns), and direct stock-
holders sorted by the number of stocks that they own (last set of columns) at the end of 2003. To
facilitate comparison, we convert all financial variables into U.S. dollars using the exchange rate
at the end of 2003 (1 Swedish krona = $0.1371).
The average participating household has a 46-year old head and a yearly income of $45,000.
It owns about 1 year of income in liquid financial wealth, 3 years of income in real estate wealth,
and 21 years of income in human capital. Within the financial portfolio, the average participant
has a risky share of 40%, owns 4 different mutual funds, and directly invests in 2 or 3 firms. These
estimates are similar to the average number of stocks in U.S. household portfolios (Barber and
Odean 2000, Blume and Friend 1975). The vast majority of risky asset participants (88%) hold
mutual funds, while 59% of them own stocks directly.
About half of direct stockholders invest in 1 or 2 companies; they have modest levels of finan-
cial wealth ($35,000), and low risky shares. We classify a stock as popular if it is one of the 10
most widely held by the household sector in at least one year between 1999 and 2007. Popular
stocks, which account for 59% of the Swedish equity market, represent 79% of the direct hold-
ings of households with 1 or 2 stocks. The diversification losses of these households are modest,
however, because concentrated stock portfolios represent only a small fraction of their financial
wealth.11
By contrast, almost 30% of direct stockholders own at least 5 different stocks. This subgroup
is important for the following reasons. Households in the subgroup have high education levels
and exhibit no bias toward popular stocks. They also have substantially higher financial wealth
($125,000) and select a higher risky share (61%) than the average investor, and correspondingly
own the bulk of aggregate equity. In the bottom rows of Table I, Panel B, we report the fraction of
the aggregate portfolio held by specific subsets of investors. The aggregate portfolio is constructed
11See Calvet, Campbell, and Sodini (2007) for a detailed analysis.
10
by adding up the stock and fund holdings of all participants. Households owning 5 stocks or more,
which represent only 17% of all participants, own 36% of aggregate mutual fund holdings and
80% of aggregate direct stockholdings. They therefore account for a substantial fraction of the
household demand for risky assets. Polkovnichenko (2005) obtains very similar statistics for this
wealthy subgroup in the U.S. Survey of Consumer Finances.
Households are not heavily tilted toward stocks in their employment sector. We classify a stock
as professionally close to household h if it has the same 1-digit Standard Industrial Classification
code as the employer of one of the adults in h. The average direct stockholder allocates 16% of the
stock portfolio to professionally close companies, which is rather modest and consistent with the
evidence from Norway (Døskeland and Hvide 2011).
In Figure 1, we report the fraction of corporate equity held by Swedish households. Specif-
ically, we sort firms by market capitalization, and for each size bucket we report the fraction of
firms in the size bucket (solid line) and the fraction of equity owned directly by Swedish house-
holds (solid bars). Households directly own 30% to 50% of firms with a market capitalization up to
100 million U.S. dollars, and a smaller fraction of larger firms.12 For the majority of Swedish com-
panies, the aggregate demand from the household sector is therefore substantial and can potentially
have a sizable impact on stock prices.
3 The Cross-Section of Household Tilts
Table II reports the distribution of the value loading of individual stocks at the end of 2003. The
loadings of individual stocks range from -3.22 (10th percentile) to 0.94 (90th percentile), with a
median of -0.37. The distribution of the value loading across individual stocks is therefore highly
heterogenous and negatively skewed. The value-weighted (VW) portfolio of Swedish stocks,
which by construction coincides with the SIXRX market index, has a value loading of -0.15 in
12In the Internet Appendix, we verify that the share of equity held by the household sector is the same for value andgrowth firms.
11
2003.13 We will therefore view a value loading of -0.15 in 2003 as being neutral. The equal-
weighted (EW) average stock loading is more negative than its VW counterpart, which stems from
the large number of small growth stocks.
Like individual stocks, household portfolios exhibit substantial heterogeneity in their value
loadings. Among participants, the loading of the risky portfolio ranges from -0.94 (10th percentile)
to 0.10 (90th percentile); the implied expected return differential is therefore approximately equal
to the value premium.14 The median loading is neutral at -0.18, so the loading distribution is neg-
atively skewed. Cross-sectional heterogeneity is slightly stronger for stock portfolios, as intuition
suggests.
The aggregate risky portfolio of the household sector has a loading of -0.26, which confirms
that the household sector as a whole exhibits only a mild growth tilt. This slight tilt originates
from the aggregate stock portfolio, which has a loading of -0.36, while the aggregate fund portfo-
lio is neutral. Moreover, whether we consider stocks or funds, the EW average household has a
more negative tilt than that its VW counterpart. A natural explanation is that low-wealth house-
holds invest in growth stocks, while high-wealth households invest in value stocks. We test this
explanation in the next section.
4 What Drives the Value Tilt?
4.1 The Value Ladder
In Figure 2, we illustrate that households progressively switch from growth to value stocks over
the life-cycle, a phenomenon which we call the “value ladder.” We sort households by birthyear
into 9 cohorts, and for each cohort we plot the average VW value loading between 1997 and 2007.
The figure is based on the stock portfolios of all Swedish households that directly hold equities
13As equation (2) implies, the value loading of the SIXRX index can vary from year to year because the universe oflisted stocks changes over time and the value loadings of individual stocks are time-invariant over the period.
14In the Internet Appendix, we report standard errors for the loading percentiles and infer that their difference arehighly significant. We also show that the return differential is slightly higher for households owning 5 stock or more,which suggests that heterogeneous loadings are not just the by-product of portfolio underdiversification.
12
during the period. We weigh households by their financial wealth because this aggregation method
has the strongest implications for asset pricing. All value loadings in a given year are demeaned in
order to control for changes in the average loading of individual stocks, which are caused by the
exit of some stocks from the stockmarket and the entry of new stocks. A similar ladder also exists
with the EW average loading or for the risky portfolio, as the Internet Appendix shows.
The value ladder in Figure 2 indicates that between the ages of 30 and 70, the value loading
varies by 0.58 and the implied return differential is therefore about one half of the value premium
(5.8% per year), which is economically substantial. The relationship between the loading and age
is strikingly linear, and in every cohort there is a tendency for households to migrate toward higher
loadings as time goes by. We note that the value ladder cannot be explained by cohort effects alone,
since such effects can explain the average loading of a cohort but not the migration of each cohort
toward value stocks. A tight combination of time and cohort effects would therefore be required
to generate the value ladder in the absence of age effects. Such a structure might originate from
inertia and other mechanical effects driving the portfolio tilt, but would otherwise be economically
challenging to explain.
From a portfolio-choice perspective, the value ladder can be attributed to (i) changes in financial
conditions over the life-cycle and (ii) pure investment horizon effects. In Sections 4.2 to 4.4, we
run pooled panel regressions to quantify the respective roles of age, financial characteristics, and
income risk on the value ladder. In Section 4.5, we consider the impact of purely mechanical
effects, such as portfolio inertia, by investigating if households actively rebalance their value tilts
and by analyzing how new entrants select their initial portfolios. These sections indirectly answer
concerns about cohort and time effects. The value ladder might also originate from learning, time-
variation in overconfidence, or peer effects. In Section 6 and in the Internet Appendix, we use
measures of sophistication, financial market experience, interpersonal communication, and firm
age to control for such mechanisms.
13
4.2 Demographic and Financial Determinants
Table III maps the relationships between portfolio tilts and socioeconomic variables. We report
pooled regressions of a household’s value loading on the household’s characteristics and year,
industry, and county fixed effects. The industry fixed effect is the 2-digit Standard Identification
Code of the household head. In the first three columns, we consider the value loading at the level
of (1) the risky portfolio, (2) the stock portfolio, and (3) the fund portfolio. In column (4), we
regress the risky share on characteristics. Standard errors are clustered at the household level.
The financial wealth coefficient is positive and strongly significant for all three portfolios.
Households with more liquid financial wealth tend have a higher value tilt than other households.
The financial wealth coefficient is the highest for the stock portfolio, which suggests that wealthy
households achieve a value tilt primarily via direct stockholdings. This finding is consistent with
the fact that mutual funds tend to have fairly neutral HML loadings (see Table II).
Households with high current income Lh,t and high expected human capital HCh,t (as defined in
equation (5)) tilt their financial portfolios toward growth stocks. These relationships are significant
for all three portfolios. Measures of income risk also have strongly negative coefficients: house-
holds with high income volatility and a self-employed or unemployed head are prone to selecting
growth stocks.15 Expected human capital and the volatility of the income process therefore both
tilt household financial portfolios toward growth stocks.
Demographic characteristics are significantly related to the value tilt. The age of the household
head tends to increase the value loading in the regression, which suggests that age is a likely
contributor to the value ladder. Younger households tend to go growth and older households tend
to go value, primarily through direct stockholdings. The gender variable is strongly significant:
men have a growth tilt and women a value tilt. Immigrants and educated households both tend to
go growth, which suggests that the value loading is not just driven by sophistication.
In Table IV, we reestimate the baseline regression on five separate groups of investors: (1) mu-
15In the Internet Appendix, we verify that our findings are robust to regressing the value tilt on the persistent andtransitory components of income risk, σξ,h and σε,h, instead of the total volatility σh. We also show that the results arerobust to alternative definitions of the income process.
14
tual fund owners, (2) direct stockholders, and (3) to (5) direct stockholders sorted by the number of
firms that they own. The baseline results remain valid in all groups. Furthermore, the explanatory
power of the regression is twice as high for households with at least 3 stocks as for households
with 1 or 2 stocks. Thus, wealthier, more educated direct stockholders holding at least three stocks
are prone to selecting value tilts that are well explained by their financial and demographic char-
acteristics.
Tables III and IV raise some immediate questions about real estate wealth, which are important
for the interpretation of the results and their connections with risk-based theories. The effects
of real estate and leverage on the value loading are relatively weak in the panel regressions. In
Table III, real estate has a positive but small effect for the risky and stock portfolios, and no effect
for the fund portfolio. Likewise, leverage has a negative effect on the value loading of the stock
portfolio, but no effect for the risky and the fund portfolios. These weak results are potentially due
to the fact that real estate is both (i) a form of wealth and (ii) a source of background risk. The
strength of the two channels is likely influenced by the level of leverage.
In Table V, Panel A, we obtain stronger results by interacting demeaned real estate with de-
meaned leverage. The leverage ratio as a standalone variable has a strongly negative impact on the
value loading, which is significant for all portfolios. For households with low leverage, residen-
tial and commercial real estate tilt the risky and stock portfolios toward value stocks, whereas for
households with high leverage, both forms of real estate tilt the financial portfolio toward growth
stocks.
Like leverage, family size plays an ambiguous role in the baseline regressions of Table III.
On the one hand, households with secure jobs and sound financial prospects are more likely to
decide to have children; thus family size can be viewed as a predictor of sound financial conditions
and co-vary positively with value investing in the cross-section. On the other hand, children are a
source of random needs and other forms of background risk that can discourage value investing.
We use twins to disentangle the two effects. Our identification strategy is that while the decision
to have a child is endogenous, the arrival of twins is an exogenous financial shock that could not
be fully anticipated and should tilt the portfolio toward growth stocks. In Table V, Panel B, we
15
accordingly modify the baseline regression by including a dummy variable for having children and
a dummy variable for having twins. While the child variable has a positive coefficient, the twin
variable has a negative impact on the value loading for all three portfolios. Thus, the unexpected
birth of an additional child tilts the portfolio toward growth stocks.
The regressions in Tables III to V provide substantial evidence that the portfolio value loading
co-varies with financial and demographic characteristics. Value investors have high financial and
real estate wealth, low leverage, low income risk, and low human capital; they are also more likely
to be older and female. Conversely, young males with risky income and high human capital are
more likely to go growth. In Section 6 and in the Internet Appendix, we verify that these results
hold in a large set of investor subgroups and sub-portfolios, and are robust to a large number
of alternative hypotheses. In particular, we show that the links between the value loading and
socioeconomic characteristics are unlikely to be due to financial market experience, the latent
heterogeneity of investors, or stock characteristics other than the value loading, such as popularity,
professional proximity, dividend yield, taxes, skewness, firm age, or exposure to the size factor.
4.3 Economic Significance
We now assess the respective contributions of age and financial characteristics to the value ladder.
In Table VI, we consider a 30-year old investor, to which we assign the average wealth-weighted
characteristics of his age group in 2003. We also consider a 50-year-old and a 70-year old with
the average characteristics of their age groups. The estimates in Table III allow us to quantify how
characteristics drive the life-cycle variation in the value loading. Between 30 and 70, the value
loading of the risky portfolio increases by 0.23, out of which 0.11 is due to age. For the stock
portfolio, the value loading increases by 0.58 between 30 and 70, out of which 0.35 is due to age.
For both portfolios, age therefore explains about 50% of the life-cycle variation in the value load-
ing. Financial characteristics also have a substantial impact. The decumulation of human capital
between 30 and 70 accounts for 27% of the life-cycle variation of the risky portfolio loading, while
the accumulation of financial wealth accounts for another 10% of the migration. Other character-
16
istics, such as real estate, have more marginal impacts.16 Overall, age and financial characteristics
explain almost entirely the slope of the value ladder.
The table reveals that life-cycle changes in age and financial characteristics all tend to increase
the value loading. Households migrate to higher loadings as their investment horizons shorten,
their balance sheets improve, and their human capital decumulates. The estimates for the 50-year
old investor confirm that consistent with the value ladder, the predicted migration is approximately
linear, which we attribute to linear changes in average characteristics across age groups.
In the Internet Appendix, we obtain similar results when we consider equal-weighted averages
instead of wealth-weighted averages. We also reestimate the decomposition when the interaction
between real estate and leverage is taken into account, and verify that age continues to explain half
of the life-cycle variation in the value loading. The measured impact of real estate and leverage
is then substantially stronger, which illustrates once again that is important to account for the
interaction between debt and real estate.
4.4 Systematic Labor Income Risk
We have seen that income volatility tends to tilt the financial portfolio toward growth stocks. We
now investigate if the value loading can also be affected by forms of systematic risk to which
households in different industries are heterogeneously exposed.
For every two-digit SIC sector s, we compute per-capita income, Ls,t , in year t using all workers
in the sector, and impute the sector’s per-capita income growth,
ℓs,t = log(Ls,t)− log(Ls,t−1).
16Demographic characteristics other than age, such as immigration status or educational attainment, vary acrosscohorts but are not expected to vary over the life-cycle of a typical household. Moreover, as Table V shows, the impactof family size is not accurately measured by the regression coefficient in Table III. We include all characteristics inTable VI for completeness, but we observe that demographic characteristics other than age only have a marginal impacton the value loading and therefore have no impact the conclusions of this section.
17
We similarly compute the growth rate of per-capita income in the economy:
ℓt = log(Lt)− log(Lt−1),
where Lt is average per-capita income in year t.
Table VII, Panel A, documents that there is considerable comovement in sectoral income
growth.17 We estimate the regression:
ℓs,t = αs +ϕs ℓt + εs,t (6)
for each of the 70 sectors, and report the distribution of the sensitivity ϕs and the coefficient of
determination R2 across regressions. The R2 coefficients are generally high and equal to 0.88
on average. Thus, aggregate income growth, ℓt , is an important factor explaining the panel of
sectoral growth rates. The sensitivity, ϕs, is heterogeneous across sectors, ranging from 0.81 (10th
percentile) to 1.22 (90th percentile).
Table VII, Panel B, shows that households working in sectors with high aggregate income
exposures tend to reduce the value tilts of their risky portfolios. Specifically, we regress a house-
hold portfolio’s value tilt, vh,t , on the household sensitivity to the macro factor, ϕh,t , the conditional
volatility of household income, σh,t , and all the other characteristics in the baseline regression. The
household sensitivity, ϕh,t , is measured by the average income-weighted sensitivity of its members,
as is explained in the Internet Appendix. The value tilt of the financial portfolio is negatively related
to industry sensitivity and income volatility. These results are especially strong for the risky port-
folio, which further confirms that household tilts are not simply the by-product of a preference for
certain types of stocks. Economic significance is substantial. For instance, as Table VII shows, the
income exposures of sectors in the 10th and 90th percentiles differ by about 0.4, which corresponds
to an absolute difference in household portfolio loading of 0.2×0.4 = 0.08. As a comparison, this
estimate slightly exceeds the change in loading induced by the life-cycle decumulation of human
capital.
17We thank the referee for encouraging us to investigate income risk at the sector level.
18
We make several observations about these results. First, we impute household sensitivities
from industry data because household income growth has a large idiosyncratic component and the
direct measurement of household sensitivity entails large estimation error, as is further explained
in the Internet Appendix. Second, our approach is motivated by earlier research showing that
the value factor correlates positively with future economic growth (Liew and Vassalou 2000) and
future labor income in U.S. and international data. In the Internet Appendix, we replicate these
earlier results on Swedish data, even though the available time series are relatively short. We
also consider a direct measure of risk, the sensitivity of labor income to the lagged value factor
itself, and similarly obtain that the portfolio value loading is negatively related to the labor income
sensitivity to HML. Overall, Table VII uncovers a powerful factor structure in industry income
growth and shows that households adjust their portfolio loadings to systematic labor income risk.
4.5 Active Rebalancing and New Entrants
4.5.1 Active Rebalancing at the Yearly Frequency
In order to climb the value ladder over the life-cycle, households presumably need to rebalance
their portfolios at shorter horizons to mitigate the impact of realized returns and revert to their
slow-moving target. For this reason, we now investigate passive and active variation in the value
tilt of household portfolios.18 Consider household h with portfolio weights wh,i,t−1 (i = 1, . . . , I) at
the end of year t −1. If the household did not trade during the following year, the share of asset i
at the end of year t would be
wPh,i,t =
wh,i,t−1 (1+ ri,t)
∑Ij=1 wh, j,t−1 (1+ r j,t)
. (7)
18Calvet, Campbell, and Sodini (2009a) define active and passive changes of the risky share, and show that house-holds actively rebalance the passive variation in the risky share due to realized asset returns. We apply a similarmethodology to the value tilt of the risky, stock, and fund portfolios.
19
By equation (2), the value loading of the passive household at the end of year t would then be:
vPh,t =
I
∑i=1
wPh,i,tvi. (8)
The data set reports the actual loading vh,t . We can therefore decompose the actual change of the
portfolio value loading as the sum of active and passive changes:
vh,t − vh,t−1 = ah,t + ph,t .
where ah,t = vh,t − vPh,t denotes the active change and ph,t = vP
h,t − vh,t−1 the passive change.
Table VIII regresses the active change, ah,t , on (i) the passive change, ph,t , (ii) the lagged value
loading, vh,t−1, and (iii) either no characteristics or all other lagged characteristics. The passive
change has a negative and highly significant coefficient for all portfolios, regardless of whether or
not one controls for household characteristics. Specifically, the passive change coefficient is -0.36
for the risky portfolio, is slightly stronger for the stock portfolio, and is slightly weaker for the
fund portfolio. Thus, households actively rebalance the passive variations generated by realized
returns, which confirms that their portfolios tilts are not purely driven by inertia.
4.5.2 New Entrants
We now verify that the value ladder is not the mechanical consequence of exogenous drifts to which
stockmarket participants are passively exposed. A natural identification strategy is to consider new
participants in the year they enter risky asset markets. In Table IX, we regress the stock portfolio
value loading of new participants on their characteristics. All the results are consistent with the
baseline regression.
In Table X, we verify that the positive age coefficient is not driven by specific age groups.
Specifically, we regress the value loading on cumulative age dummies, cumulative age dummies
for new entrants, and all characteristics other than age. The cumulative age dummies common
to all participants are strictly positive and almost all significant. Moreover, the relationship be-
20
tween a participant’s age and its value loading is approximately linear, consistent with our baseline
specification and the value ladder in earlier sections.
The dummy variable for new entrants aged 30 or more is significantly negative. The other
age dummies of new entrants are insignificant. Since the age dummy coefficients are cumulative,
these results imply that (i) all new entrants have a significant bias toward growth stocks and (ii)
age does not impact the difference between the tilt of preexisting participants and the tilt of new
entrants. Thus, the value ladder of new entrants is located below and is parallel to the value ladder
of preexisting participants.
Overall, this section documents that household portfolios progressively shift from growth to
value over the life cycle. The migration is explained by the complementary impact of age and
socioeconomic characteristics on the portfolio tilt. The value ladder is made possible by active
rebalancing and is also observed in the portfolio of new entrants. In the next section, we discuss
how these results relate to theoretical explanations of the value premium.
5 Interpretation of the Portfolio Evidence
In this Section, we show that our empirical results are consistent with some of the leading expla-
nations of the value premium.
5.1 Intertemporal Hedging
5.1.1 Investment Horizon and Age
Risk-based theories of the value premium emphasize the link between the HML factor and the
dynamics of the investment opportunity set. Empirically, good realizations of the value factor pre-
dict high aggregate returns (Campbell and Vuolteenaho 2004) and high economic growth (Koijen,
Lustig, and Van Nieuwerburgh 2014, Liew and Vassalou 2000) in U.S. and international data.
Thus, the HML factor explains both the cross-section of risk premia and the distribution of future
21
returns, which is consistent with the definition of a factor in the Intertemporal Capital Asset Pric-
ing Model (ICAPM, Merton 1973). Investors can use growth stocks to hedge against low future
aggregate returns.
The existing portfolio-choice literature on the value factor focuses on intertemporal hedging
and its implications for investors with different horizons (Jurek and Viceira 2011, Larsen and
Munk 2012, Lynch 2001). Because the hedging motive is stronger for long-term investors than
for short-term investors, portfolio theory implies that young investors should be more tilted toward
growth stocks than old investors. A closely related mechanism is that value stocks have shorter
durations and less discount-rate risk than growth stocks (Campbell and Vuolteenaho 2004, Cornell
1999, Dechow, Sloan, and Soliman 2004, Lettau and Wachter 2007); one therefore expects young
investors to hold long-duration growth stocks, while old investors should hold short-duration value
stocks.19
The empirical evidence in Section 4 shows that age is positively and significantly related to the
value loading. This relationship is observed even when we control for real estate, debt, financial
market experience, human capital, income risk, and other socioeconomic characteristics that vary
with age. Our baseline results thus provide strong empirical support for one the main predictions
of portfolio choice models incorporating the value factor, the positive link between age and value
investing.
5.1.2 Household Tilts in Partial and General Equilibrium
In order to facilitate the theoretical interpretation of the value ladder and the panel regressions in
Section 4, we now consider an ICAPM model similar to Merton (1973) and Breeden (1979). The
economy consists of K state variables, I risky assets, and a set of investors with finite horizons and
heterogeneous lifespans, which we fully specify in the Internet Appendix. The model accommo-
dates a wide range of overlapping generations structures, which is important for the analysis of
the value ladder. When the state variables consist of the market price of risk and aggregate labor
income, the model can relate the HML portfolio to time-varying returns, as in Lynch (2001), and
19Campbell and Viceira (2001) apply a similar logic to bond investments.
22
to labor income risk, as in Jagannathan and Wang (1996).20
The following portfolios play an important role in the analysis. The tangency portfolio τtτtτt
maximizes the Sharpe ratio of a myopic (or short-lived) agent. The kth mimicking portfolio is
the portfolio with the highest absolute correlation with the kth state variable. We denote by fk,tfk,tfk,t
the zero-sum portfolio that is short the tangency portfolio and long the kth mimicking portfolio.
The long-short portfolios fk,tfk,tfk,t can be viewed as “factor portfolios” that have similar definitions and
pricing implications as HML.
The optimal portfolio of an individual investor h is determined by diversification and hedging.
The shares of risky wealth held in each risky asset, ωhtωhtωht ∈ R
I, satisfy
ωhtωhtωht = τtτtτt +
K
∑k=1
ηhk,t
wht
fk,tfk,tfk,t , (9)
where each coefficient ηhk,t quantifies the investor’s sensitivity to state variable k and wh,t denotes
the risky share. The investor’s deviation from the tangency portfolio is substantial if the ratios
ηhk,t/wh
t are large, that is if the hedging demand is strong and represents a substantial fraction of
the risky portfolio.
The portfolio choice literature implies that the following properties hold under a wide range
of state dynamics. Young investors generally have stronger sensitivities ηhk,t than old investors,
which stems from the fact that older investors are close to the terminal date and tend to behave
like myopic investors (Lynch 2001, Wachter 2002). When aggregate income is a state variable,
the sensitivity ηhk,t is strong if the household is exposed to high systematic risk in labor income
or has a large stock of human capital. The risky share wht , which has been widely studied in the
portfolio-choice literature,21 is low if the investor has high risk aversion, holds little liquid wealth,
earns a risky income, and has high debt. In the context of value and growth investing, this partial
equilibrium analysis suggests that young investors with risky incomes and weak balance sheets
should tilt their financial portfolios away from value.
20Breeden (1979) and Cochrane (2007) show that labor income risk can easily be included in the ICAPM framework.21See Campbell and Viceira (2000) and the references therein.
23
In general equilibrium, households hold the market portfolio, mtmtmt , and heterogeneous positions
in the factor portfolios:
ωhtωhtωht =mtmtmt +
K
∑k=1
(
ηhk,t
wht
−ηm
k,t
wmt
)
fk,tfk,tfk,t , (10)
where each coefficient ηmk,t/wm
t denotes the relative sensitivity of the aggregate investor to the
kth factor.22 While the aggregate investor holds the market portfolio, each investor h tilts toward
or away from the kth factor if its relative sensitivity to the state variable, ηhk,t/wh
t , differs from the
average sensitivity ηmk,t/wm
t . We refer the reader to the Internet Appendix for a full discussion of the
model. Equation (10) implies that the value ladder can arise in the equilibrium of the overlapping
generations ICAPM economy, because the young have stronger hedging needs and the old weaker
hedging needs than the average investor.
The value ladder of new entrants reported in Section 4.5 also has a natural interpretation in
a general equilibrium context. In an economy in which participants gradually sell their growth
stocks and migrate toward value stocks, the growth stocks must be absorbed by another group of
investors. The empirical evidence shows that new entrants have a growth tilt compared to other
households. Thus, new entrants absorb the growth stocks of preexisting participants. At the other
end of the ladder, the portfolios of the deceased contain value stocks that surviving investors can
purchase. New entrants and inheritances therefore permit the migration from growth stocks to
value stocks over the life-cycle. Our results also suggest that demographic changes can affect the
demand for value and growth stocks, which may have implications for the value premium.
5.2 Risk Aversion, Wealth and Background Risk
Risk-based explanations of the value premium are based on the premise that HML carries sys-
tematic risk that is unspanned by the market index. Value stocks should therefore be picked by
investors who have a strong capacity to bear risk, for instance because they have high liquid finan-
cial wealth, high real estate wealth, and low leverage. These effects are apparent in the ICAPM
model discussed in Section 5.1.
22Our derivation builds on the work of Long (1974), Ingersoll (1987), and Cochrane (2007).
24
Quite remarkably, the empirical impact of financial variables on the portfolio tilt is generally in
accordance with the predictions of risk-based theories. Liquid financial wealth is positively related
to the value loading across participants (Table III) and in all subgroups of investors, including the
wealthy group of stockholders owning 5 stocks or more (Table IV). As in earlier studies, financial
wealth is also associated with high risky shares (Table III). These results suggest that wealthier
households adopt value strategies because they are effectively more risk tolerant and therefore more
prone to bearing the systematic risk (other than market portfolio risk) embedded in value stocks.
The results on real estate wealth and leverage provide further evidence that households with sound
balance sheets tilt their portfolios toward value stocks in order to earn the value premium.
Expected utility theory also implies a link between effective risk tolerance and the level of
background risk (see, e.g., Kihlstrom, Romer and Williams 1981). The baseline results on family
size, income risk, self-employment, and immigration status all give empirical support to this view.
The unexpected birth of a child induces a growth tilt, which is consistent with the lower resources
per-capita and higher idiosyncratic needs that the arrival of a newborn entails. High volatility of
income also creates a growth tilt. Indeed, the volatility of household real disposable income is
substantial in Sweden, with an average of 16% per year (Table I),23 and is primarily idiosyncratic,
as we show in the Internet Appendix.24 Similarly, entrepreneurs and immigrants exhibit a growth
tilt, presumably because of substantial idiosyncratic risk in business assets and income.25
23The high volatility of income in Sweden is also documented in Floden and Lindé (2001) and Betermier, Jansson,Parlour, and Walden (2012). Complementary evidence from Holmlund and Storrie (2002) shows a sharp rise in fixed-term contracts following the recession of the early 1990’s, accounting for up to 70% of new hires in Sweden by thelate 1990’s.
24We verify that consistent with the baseline regression, idiosyncratic income volatility induces both a low riskyshare and growth tilt, just as theory predicts.
25The income volatility of self-employed households is estimated at 29% per year, as compared to 16% for the aver-age household. Private businesses are also characterized by high failure rates and highly risk in capital returns, whichare primarily idiosyncratic (Moskowitz and Vissing-Jørgensen 2002). For immigrants, the rate of unemployment is11% in Sweden in 2003, as compared to 7% for non-immigrants. Lemaître (2007) provides a detailed description ofthe hurdles faced by immigrants in the Swedish labor market.
25
5.3 Income and Human Capital
Risk-based theories consider that the HML factor captures forms of cash flow risk to which value
and growth stocks are heterogeneously exposed.26 While the exact nature of this risk is the subject
of some debate, an extensive literature relates HML to labor income and human capital. For
instance, Jagannathan and Wang (1996), Lettau and Ludvigson (2001), Palacios-Huerta (2003),
Petkova and Zhang (2005) and Santos and Veronesi (2006) develop conditional versions of the
CAPM and C-CAPM that incorporate aggregate income growth and can price the Fama and French
portfolios.27 Garleanu, Kogan, and Panageas (2012) consider that human capital is sensitive to
innovation shocks, and derive the implications of the resulting “displacement” risk for the cross-
section of stock returns and household portfolio tilts.
The present paper uncovers several empirical facts in support of labor-based theories of the
value premium. First, we document a strong factor structure in the industry distribution of income
growth. As Section 4.4 shows, aggregate income growth explains on average 88% of the time
series variation in sectoral income growth, and the sensitivity to the factor is heterogeneous across
sectors. Industry data therefore confirm that aggregate income growth is an appealing macroeco-
nomic factor for asset pricing.
Second, we find that households working in sectors with high exposures to the macro factor
tend to select financial portfolios with low exposures to HML, as the hedging motive implies. The
household data thus reveal a close link between aggregate income risk and the portfolio choices
of retail investors, which confirms that aggregate income growth is a good candidate for asset
pricing applications. In his Presidential Address to the American Finance Association, Cochrane
(2011) develops the following implications of the value factor: “If a mass of investors has jobs
or businesses that will be hurt especially hard by a recession, they avoid stocks that fall more
26Statistical decompositions show that value stocks have higher exposures to the market’s cash-flow risk than growthstocks (Campbell and Vuolteenaho 2004), and that the value loadings of individual stocks are primarily driven by theirown cash flows (Campbell, Polk, and Vuolteenaho 2010). Furthermore, value stocks are strongly exposed to deeprecessions and the persistent reductions in aggregate cash flows that they entail (Campbell, Giglio, and Polk 2013).
27Complementing these empirical studies, Parlour and Walden (2011) and Sylvain (2013) derive general equilibriummodels in which risky labor income drives the cross-section of book-to-market ratios and risk premia. See alsoLewellen and Nagel (2006), Nagel and Singleton (2011), Ang and Kristensen (2012) and Roussanov (2013) for recentcritiques of the CAPM and C-CAPM.
26
than average in a recession.” The present paper confirms Cochrane’s prediction by showing that
workers in exposed industries select portfolios with low HML loadings. Furthermore we document
that households with self-employed heads exhibit additional tilts toward growth stocks, presumably
because small businesses are especially sensitive to recession risk.
In addition to these results, we uncover that high expected human capital is associated with a
growth tilt in the financial portfolio. This relationship is strong in all the specifications considered
in this paper and the Internet Appendix. Intuition suggests that human capital is both of form of
wealth, which in principle might induce a value tilt, and a form of risk, which in the data induces a
growth tilt. We can offer several possible explanations for the dominance of the risk channel, which
build on the extensive literature relating the value premium to the production process.28 Since
human capital is a key complement of physical capital in production, households with high levels
of human capital may tilt away from the physical capital in value firms and invest instead in growth
firms. Sylvain (2013) accordingly develops a general equilibrium model with both human and
physical capital investment, and shows that value stocks endogenously exhibit a high sensitivity to
human capital risk.29 A complementary explanation is that human capital is highly risky because
it is exposed to tail risks and innovation shocks that are difficult to anticipate and measure ex ante,
as in the theoretical models Garleanu, Kogan, and Panageas (2012) and Kogan, Papanikolaou, and
Stoffman (2013). The strong empirical link between human capital and growth investing is a novel
empirical fact that deserves to be explored in future research.
28Production-based asset pricing models, which have had success in relating the sensitivity of a firm’s traded equityto the firm’s physical assets and growth options (Berk, Green, and Naik 1999, Gomes, Kogan, and Zhang 2003).Cutting physical capital in bad times entails more adjustment costs that expanding physical capital in good times.Assets in place are therefore riskier than growth options, especially in bad times when the price of risk is high. Asa result, value stocks are more sensitive than growth stocks to the business cycle (Zhang 2005). Related channelsinclude operational leverage (Carlson, Fisher, and Giammarino 2004), investment-specific technology (Kogan andPapanikolaou 2014), and the cyclicality of the demand for durable goods (Gomes, Kogan, and Yogo 2009).
29Baxter and Jermann (1997) show that human capital is positively correlated with aggregate physical capital at themacro level.
27
5.4 Overconfidence
Sentiment-based explanations of the value premium consider that investors exuberantly overprice
growth (“glamour”) stocks and underprice value stocks (“fallen angels”), which explains the long-
run success of value investing. Several psychological biases may account for such mispricing. In-
vestors may be overconfident and thereby overestimate the accuracy of available information. They
may also pay more attention to recent events than Bayesian updating would imply (Kahneman and
Tversky 1973). Investor with such biases tend to overprice stocks following positive news and un-
derprice stocks following negative news, so that valuation ratios can predict future returns (Daniel,
Hirshleifer, and Subrahmanyam 2001). These mechanisms are consistent with the biases in stock
analyst expectations (La Porta 1996, La Porta, Lakonishok, Shleifer, and Vishny 1997, Greenwood
and Sheifer 2013, Skinner and Sloan 2002) and the pricing impact of sentiment measures (Baker
and Wurgler 2006).
Cognitive biases have a number of potential implications for portfolio choice. Men and en-
trepreneurs are known to be especially prone to overconfidence (Barber and Odean 2001, Busenitz
and Barney 1997, Cooper, Woo, and Dunkelberg 1988) and should therefore favor growth stocks.
The evidence in Section 4.2 confirms these predictions. Women tend to select low risky shares and
invest in value stocks, while men tend to select aggressive risky shares and go growth. These pat-
terns cannot easily be explained by differences in risk aversion alone, since a risk-tolerant investor
should choose both a high risky share and a growth tilt. The positive link between entrepreneurship
and growth investing might also be explained by overconfidence.
In the Internet Appendix, we reestimate the baseline regression on the subsample of households
with a male head and on the subsample of households with a self-employed head. The baseline
results of the paper hold in both subsamples and are therefore unlikely to be driven by cross-
sectional differences in overconfidence alone.
28
6 Identification and Robustness Checks
6.1 Popular and Professionally Close Stocks
A potential concern with Swedish data is that a handful of firms dominate the domestic stock mar-
ket and therefore the stock portfolios of a majority of households (Table I). In Table XI, we consider
the 10 stocks that are most widely held by households at the end of 2003, and report for each firm
the percentage of direct stockholders owning it, the stock’s percentage of aggregate household fi-
nancial wealth, the stock’s percentage of the Swedish stock market, the stock’s percentage of the
Swedish free float, the stock’s value loading, and the percentile of the stock’s book-to-market ratio.
Popular stocks are a mix of growth stocks and value stocks, regardless of whether one classifies
stocks by their value loadings or book-to-market ratios.
In the first two sets of columns of Table XII, we reestimate the baseline regression on (1)
household portfolios of popular stocks and (2) household portfolios of non-popular stocks. For
both portfolios, characteristics have the same impact as in the baseline regression.30 In the Internet
Appendix, we verify that the baseline results also hold among households that invest either 100%
or 0% of their stock portfolios in popular firms.
We next ask if professionally close stocks, which represent 16% of household stock portfolios,
can account for the relationships between the value loading and characteristics. In columns (3)
and (4) of Table XII, we show that the baseline results are also valid in household portfolios of
professionally close stocks and in household portfolios of other stocks. In the Internet Appendix,
we reach a similar conclusion for households with extreme shares of professionally close stocks or
working in specific sectors. Thus, the relationships between the value loading and characteristics
are not driven by popular and professionally close stocks.
30One complementary result is that the baseline results holds among the wealthy subgroup of investors holding atleast 5 firms, who do not have a bias for popular stocks.
29
6.2 Other Stock Characteristics
We now verify that the baseline results are not driven by stock characteristics other than the value
loading.
Dividends and Taxes. One may ask if the value tilt picks up a retail demand for dividend-paying
or tax-advantaged stocks, which is unrelated to HML. For example, Graham and Kumar (2006) use
U.S. brokerage data to show that the demand for dividend yield increases with age and decreases
with income, which they interpret as evidence of age and tax clienteles. We note that in Sweden,
capital losses are deductible and the tax rate is 30% on both capital gains and dividends, so the tax
clientele story is not as clear in Sweden as it is in the United States. Furthermore, in Table XIII, we
reestimate the baseline regression on sub-portfolios of stocks sorted by dividend yields, and verify
that the baseline results hold in all portfolios (including the 40% of stocks that pay no dividends).
In Table XIII, we investigate the potential impact of taxes by considering two identification
methods. First, the wealth tax, which was levied on Swedish households until 2007, applied to
stocks in the A list of the Stockholm Stock Exchange but not to the smaller stocks in the O list.
The O list includes approximately 230 stocks and is updated each year. The baseline results are
observed both in the portfolio of A-listed stocks and in the portfolio of O-listed stocks. Second,
until 2004, Swedish households were levied inheritance and gift taxes at death, but these taxes
did not apply to O-listed stocks (Du Rietz, Henrekson, and Waldenström 2012). In the Internet
Appendix, we verify that the baseline results hold in the subperiod that follows the repeal of the
inheritance tax (2005-2007).
Firm Age A possible story is that young households invest in young firms and old households
invest in old firms, without consideration of HML loadings. This mechanism is unlikely to explain
our baseline results for two main reasons. First, since we use unconditional estimates of firm
loadings, our results cannot be contaminated by exogenous changes in firm value tilts between
1999 and 2007. Consequently, the value ladder in Figure 2 shows that the shares of value stocks
increase over time in household portfolios. Second, we consider all the firms traded on Nordic
exchanges in 2007. We construct the “young” portfolio as the set of stocks that have been listed
30
for less than 10 years, and the “old” portfolio as the set of firms that have listed for at least 20
years. The young and old portfolios respectively contain 50% and 30% of all stocks. The baseline
results hold in both portfolios.
Skewness. A recent literature suggests that the demand for positively-skewed “lottery” stocks
could explain the under-diversification of household portfolios (Goetzmann and Kumar 2008,
Kumar 2009b, Mitton and Vorkink 2007, Polkovnichenko 2005). While lottery stocks tend to
be small and young growth stocks, it is unlikely that the value tilt is explained by preference for
skewness. First, the demand for lottery-stocks is relatively small. Kumar (2009b) estimates that the
average share invested in lottery-stocks is less than 4% of household risky portfolios. We observe
a similar pattern in Table I. Among households that own 1 or 2 stocks directly, the amount invested
in the smaller non-popular stocks only represents $1,000 out of a financial wealth of $37,000. Sec-
ond, households choose similar value tilts in their stock and fund portfolios (Table III), which is
inconsistent with the implications of portfolio theory when investors have a preference for skew-
ness (Langlois 2013, Mitton and Vorkink 2007). Third, it is clear from Table IV that our results
are the strongest among households with more diversified portfolios. In Table XII and XIII, our re-
sults remain strong in the portfolios of popular and old stocks, which do not include typical lottery
stocks. Thus, preference for skewness alone cannot explain our main empirical results.
6.3 Financial Market Experience
A possible interpretation of the value ladder is that age proxies for financial market experience.
Naive new investors might purchase overpriced growth stocks, learn that these stocks are bad deals,
and then progressively migrate toward value stocks as time goes by.31 In Table XIV, we control
for learning effects by considering a measure of experience, the number of years since entry. We
focus on 2007 risky asset market participants and regress the 2007 value loading on the experience
measure, age, the value loading in the year of entry, and the other usual characteristics in 2007. The
coefficient on the number of years since entry is significantly negative for all portfolios, which is
31The psychology literature documents that cognitive biases attenuate with experience in sufficiently regular envi-ronments (Hogarth 1987, Kahneman 2011, Oskamp 1965). Malmendier and Nagel (2011) provide some evidence thatyounger or less experienced investors are especially likely to extrapolate from recent financial data.
31
inconsistent with the simple learning story.32 Thus, financial market experience, measured by the
number of years in risky asset markets, induces a growth tilt and, more importantly, cannot explain
away the positive link between age and the value tilt. In a recent study, Campbell, Ramadorai,
and Ranish (2014) consider an Indian brokerage data set containing highly detailed information on
individual trades, but no socioeconomic characteristics. They show that the returns experienced by
a household drive its future portfolio style. Our results indicate that the number of years spent on
financial markets cannot explain away the relationship between age and value investing.
Table XIV also sheds light on the dynamics of the portfolio tilt during the participation period.
The value loading in the entry year has a positive and strongly significant impact on the value
loading in 2007, as one might expect. Furthermore, the impact of other characteristics remain
significant and are consistent with our earlier results when we control for the initial loading. This
suggests that the value loading is not simply driven by the initial portfolio in the year of entry, but
also depends on financial and demographic characteristics in the subsequent participation period.
6.4 Latent Heterogeneity
The baseline regression includes household characteristics and yearly, industry, and county fixed
effects. We now use the twin panel to verify that the characteristics do not merely proxy for latent
traits or cohort effects. In Table XV, we estimate the specification:
vk,1,t = αk,t +b′xk,1,t + ek,1,t , (11)
vk,2,t = αk,t +b′xk,2,t + ek,2,t , (12)
where vk, j,t denote the value loading of sibling j ∈ {1,2} in pair k at date t, αk,t is a yearly pair
fixed effect, xk, j,t denotes the vector of yearly characteristics of sibling j, and ek, j,t is an orthogonal
error. The yearly twin pair fixed effect captures the impact of time, such as age or stock market
performance, as well as similarities between the twins, such as common genetic makeup, family
32In the Internet Appendix, we verify that the results of Table XIV are not driven by multicollinearity between ageand the experience variable, nor by limitations in the experience variable due to the length of the panel.
32
background, upbringing, and expected inheritance. Since twin siblings have the same age, the
twin regression naturally controls for cohort effects.33 The twin regressions are consistent with the
baseline results. The Internet Appendix reports similar results for the subsample of identical twins.
The twin regression has a substantially higher adjusted R2 coefficient than the baseline regres-
sion. For the stock portfolio, characteristics and yearly twin pair fixed effects account for 23% of
the cross-sectional variation of the stock portfolio value tilt. By contrast, in the baseline regression
(Table III), socioeconomic characteristics and year, industry, and county fixed effects explain 4%
of the cross-sectional variation in the value loading. Large increases in adjusted R2 are also ob-
tained for the risky and fund portfolios. In the next section, we discuss the possible origins of the
high explanatory power of the yearly twin pair fixed effects.
6.5 Communication and Genes
We now assess the impact of communication between twins on their portfolio tilts. The twin panel
contains detailed information on the frequency of communication between twins. In Table XVI, we
classify a twin pair as “high communication” if the frequency of mediated communication and the
frequency of unmediated communication are both above the median, and as “low communication”
otherwise. We sort twin pairs into communication bins, and reestimate in each bin the baseline
regression with year, industry and county fixed effects. The reported regressions are generally
consistent with the baseline results. In the Internet Appendix, we obtain similar coefficients when
we include yearly twin pair fixed effects. Thus, communication does not drive the relationship
between the value tilt and socioeconomic variables.
The large R2 of the twin regressions in Table XV might suggest that value investing has genetic
origins. Cronqvist, Siegel, and Yu (2015) consider a genetic model of the value loading, in which
the value loading vk,s of sibling s in pair k is assumed to be the sum of a genetic component, ak,s,
33Calvet and Sodini (2014) apply this methodology to the determinants of the risky share. Cesarini, Dawes, Johan-nesson, Lichtenstein, and Wallace (2009), Cesarini, Johannesson, Lichtenstein, Örjan Sandewall, and Wallace (2010),and Barnea, Cronqvist, and Siegel (2010) also use twins to investigate risk-taking.
33
a common component, ck, and an idiosyncratic component, εk,s :
vk,s = ak,s + ck + εk,s.
The main identification condition is that the twin correlation of the genetic component, Corr(ak,1;
ak,2), is 1 for identical twins and 1/2 for fraternal twins. Under this model, the contribution of the
genetic component to the cross-sectional variance of the value loading satisfies
Var(ak,s)
Var(vk,s)= 2(ρI −ρF), (13)
where ρI denotes the twin correlation of the value loading among identical siblings, and ρF denotes
the twin correlation among fraternal siblings. This equation serves as the basis for estimation.
Cronqvist, Siegel, and Yu (2015) attribute 30% of the value loading to the genetic component.
Table XVII confirms these earlier results, but shows that they are highly sensitive to communi-
cation. The genetic share reaches 35% for frequent communicators but disappears almost entirely
among infrequent communicators, with estimates that do not exceed 1% across specifications.34
These low estimates are especially surprising if the genetic is correctly specified, because purely
genetic effects should not depend on communication. One could argue that the communication
frequency itself has genetic origins, so that the results of Table XVII could be construed as ev-
idence that value investing is driven by genes. However, by equation (13), the genetic share is
zero if and only if the twin correlation of the value loading is the same for identical and fraternal
pairs: ρI = ρF . Thus, a genetic theory of value investing would need to explain why infrequently
communicating twins have the same loading correlations regardless of genetic makeup.
The sensitivity of the genetic decomposition is related to one of its well-known shortcomings,
namely that it neglects interactions between genetic and environmental variables. Interactions
34The estimator of the genetic share (13) is the rescaled difference between two sample correlations. It can thereforetake negative values if the estimate of ρI is lower than the estimate of ρF in a particular sample. In fact, under the nullhypothesis
H0 : ρI = ρF ,
the estimator of the genetic share converges asymptotically to a centered normal as the number of pairs goes to infinity.Negative estimates of the so-called genetic share are then asymptotically as likely as positive estimates.
34
between nature and nurture are known to be empirically important in medicine and experimental
psychology (Ridley 2003). The modern view in these fields is that genes cause a predisposition to
certain behaviors or diseases, which develop only in particular environments. Table XVII shows
that the clean dichotomy between nature and nurture is equally elusive in the context of value
investing.
7 Conclusion
An extensive asset-pricing literature relates the value premium to a wide range of macroeconomic
variables. The present paper uncovers that strong patterns exist at the microeconomic level in a
large panel of household portfolios. We document substantial heterogeneity in the value tilt across
households, which is of the same size as the value premium. Households migrate from growth
stocks to value stocks as they become older and their balance sheets improve. Households achieve
their migration up the value ladder by actively managing their stock and fund holdings, for instance
by rebalancing the passive variation in portfolio tilts induced by market returns.
The paper provides empirical support for several key explanations of the value premium. In
striking accordance with risk-based theories, value stocks are held by households that are in the
best position to take systematic financial risk, for instance because they have sound balance sheets
or low background risk. Pure horizon effects account for at least 50% of the life-cycle variation
of the value tilt, which provides strong empirical support for intertemporal hedging (Jurek and
Viceira 2011, Larsen and Munk 2012, Lynch 2001). We also find that investors with high human
capital and high exposure to aggregate labor income shocks tilt away from value, which is in
line with labor-based theories of the value premium. These empirical regularities are stronger
among households that own diversified stock portfolios, have high risky shares and own the bulk
of aggregate equity.
The evidence suggests that behavioral biases matter as well, especially among the large group
of small investors who have low risky shares and hold a small fraction of aggregate equity. A large
number of investors hold one to four stocks, which concentrate in a handful of popular companies.
35
The relationships between growth investing and variables such as gender seem consistent with
the overconfidence bias documented in the psychology literature. Thus the panel of household
portfolio tilts is explained by a mix of risk-based and behavioral explanations.
Our results provide new directions for portfolio-choice and asset-pricing theories of the value
factor. The household data reveal that growth investing is tightly linked to aggregate income risk
and human capital, which one might seek to match in a calibrated life-cycle model. Our empirical
findings also suggest that powerful general equilibrium effects are at play in the cross-sectional dis-
tribution and the dynamics of portfolio tilts. The development of overlapping generations models
matching these features are natural extensions of our work. Last but not least, the empirical pat-
terns in the demand for value and growth stocks may have major implications for equity valuation,
which will be investigated in further research.
36
References
Addoum, J. M., G. M. Korniotis, and A. Kumar, 2013, “Income Hedging, Dynamic Style Prefer-
ences, and Return Predictability,” working paper, University of Miami.
Ang, A., and D. Kristensen, 2012, “Testing Conditional Factor Models,” Journal of Financial
Economics, 106(1), 132–156.
Angerer, X., and P.-S. Lam, 2009, “Income Risk and Portfolio Choice: An Empirical Study,”
Journal of Finance, 64(2), 1037–1068.
Asness, C. S., and A. Frazzini, 2013, “The Devil in HML’s Details,” Journal of Portfolio Manage-
ment, 39(4), 49–68.
Asness, C. S., T. J. Moskowitz, and L. H. Pedersen, 2013, “Value and Momentum Everywhere,”
Journal of Finance, 63(3), 929–985.
Baker, M., and J. Wurgler, 2006, “Investor Sentiment and the Cross-Section of Stock Returns,”
Journal of Finance, 61(4), 1645–1680.
Ball, R., 1978, “Anomalies in Relationships between Securities’ Yields and Yield Surrogates,”
Journal of Financial Economics, 6(2/3), 103–126.
Bansal, R., R. Dittmar, and D. Kiku, 2009, “Cointegration and Consumption Risks in Asset Re-
turns,” Review of Financial Studies, 22(3), 1343–1375.
Bansal, R., R. F. Dittmar, and C. T. Lundblad, 2005, “Consumption, Dividends, and the Cross
Section of Equity Returns,” Journal of Finance, 60(4), 1639–1672.
Bansal, R., D. Kiku, I. Shaliastovich, and A. Yaron, 2014, “Volatility, the Macroeconomy, and
Asset Prices,” Journal of Finance, 69(6), 2471–2511.
Barber, B. M., and T. Odean, 2000, “Trading Is Hazardous to Your Wealth: The Common Stock
Investment Performance of Individual Investors,” Journal of Finance, 55(2), 773–806.
37
, 2001, “Boys Will be Boys: Gender, Overconfidence, and Common Stock Investment,”
Quarterly Journal of Economics, 116(1), 261–292.
, 2008, “All that Glitters: The Effect of Attention and News on the Buying Behavior of
Individual and Institutional Investors,” Review of Financial Studies, 21(2), 785–818.
Barberis, N., and R. Thaler, 2003, “A Survey of Behavioral Finance,” in Handbook of the Eco-
nomics of Finance, ed. by G. Constantinides, R. Stulz, and M. Harris. North Holland, Amster-
dam.
Barnea, A., H. Cronqvist, and S. Siegel, 2010, “Nature or Nurture: What Determines Investor
Behavior?,” Journal of Financial Economics, 98, 583–604.
Barras, L., 2013, “Revisiting The Value Premium using Micro Portfolios as Test Assets,” working
paper, McGill University.
Basu, S., 1977, “Investment Performance of Common Stocks in Relation to Their Price-Earnings
Ratios: A Test of the Efficient Market Hypothesis,” Journal of Finance, 32(3), 663–682.
, 1983, “The Relationship between Earnings Yield, Market Value and Return for NYSE
Common Stocks: Further Evidence,” Journal of Financial Economics, 12(1), 129–156.
Baxter, M., and U. Jermann, 1997, “The Internation Divesrsification Puzzle is Worth Than You
Think,” American Economic Review, 87(1), 170–180.
Berk, J. B., R. C. Green, and V. Naik, 1999, “Optimal Investment, Growth Options, and Security
Returns,” Journal of Finance, 54(5), 1553–1607.
Betermier, S., L. Calvet, and P. Sodini, 2015, “Internet Appendix for “Who Are the Value and
Growth Investors?”,” Available online at http://www.hec.fr/calvet.
Betermier, S., T. Jansson, C. Parlour, and J. Walden, 2012, “Hedging Labor Income Risk,” Journal
of Financial Economics, 105(3), 622–639.
Blume, M. E., and I. Friend, 1975, “The Asset Structure of Individual Portfolios and Some Impli-
cations for Utility Functions,” Journal of Finance, 30(2), 585–603.
38
Bonaparte, Y., G. Korniotis, and A. Kumar, 2014, “Income Hedging and Portfolio Decisions,”
Journal of Financial Economics, 113, 300–324.
Breeden, D., 1979, “An Intertemporal Asset Pricing Model with Stochastic Consumption and In-
vestment Opportunities,” Journal of Financial Economics, 7(3), 265–296.
Busenitz, L., and J. Barney, 1997, “Differences Between Entrepreneurs and Managers in Large
Organizations: Biases and Heuristics in Strategic Decision-Making,” Journal of Business Ven-
turing, 12(1), 9–30.
Calvet, L., J. Campbell, and P. Sodini, 2007, “Down or Out: Assessing the Welfare Costs of
Household Investment Mistakes,” Journal of Political Economy, 115(5), 707–747.
, 2009a, “Fight or Flight? Portfolio Rebalancing by Individual Investors,” Quarterly Jour-
nal of Economics, 124(1), 301–348.
, 2009b, “Measuring the Financial Sophistication of Households,” American Economic
Review, 99(2), 393–398.
Calvet, L. E., and P. Sodini, 2014, “Twin Picks: Disentangling the Determinants of Risk-Taking in
Household Portfolios,” Journal of Finance, 69(2), 869–908.
Campbell, J. Y., S. Giglio, and C. Polk, 2013, “Hard Times,” Review of Asset Pricing Studies, 3,
95–132.
Campbell, J. Y., C. Polk, and T. Vuolteenaho, 2010, “Growth or Glamour? Fundamentals and
Systematic Risk in Stock Returns,” Review of Financial Studies, 23, 305–344.
Campbell, J. Y., T. Ramadorai, and B. Ranish, 2014, “Getting Better: Learning to Invest in an
Emerging Stock Market,” working paper, Harvard University and Oxford University.
Campbell, J. Y., and L. M. Viceira, 2000, Strategic Asset Allocation: Portfolio Choice for Long-
Term Investors. Oxford University Press, New York.
, 2001, “Who Should Buy Long-Term Bonds?,” American Economic Review, 91(1), 99–
127.
39
Campbell, J. Y., and T. Vuolteenaho, 2004, “Bad Beta, Good Beta,” American Economic Review,
94(5), 1249–1275.
Capaul, C., I. Rowley, and W. F. Sharpe, 1993, “International Value and Growth Stock Returns,”
Financial Analysts Journal, 49(1), 27–36.
Carhart, M. M., 1997, “On Persistence in Mutual Fund Performance,” Journal of Finance, 52(1),
57–82.
Carlson, M., A. Fisher, and R. Giammarino, 2004, “Corporate Investment and Asset Price Dynam-
ics: Implications for the Cross-Section of Returns,” Journal of Finance, 59, 2577–2603.
Carroll, C. D., and A. Samwick, 1997, “The Nature of Precautionary Wealth,” Journal of Monetary
Economics, 40, 41–71.
Cesarini, D., C. Dawes, M. Johannesson, P. Lichtenstein, and B. Wallace, 2009, “Genetic Variation
in Preferences for Giving and Risk-Taking,” Quarterly Journal of Economics, 124, 809–842.
Cesarini, D., M. Johannesson, P. Lichtenstein, Örjan Sandewall, and B. Wallace, 2010, “Genetic
Variation in Financial Decision Making,” Journal of Finance, 65, 1725–1754.
Chan, L. K., Y. Hamao, and J. Lakonishok, 1991, “Fundamentals and Stock Returns in Japan,”
Journal of Finance, 46, 1739–1789.
Cocco, J., F. Gomes, and P. Maenhout, 2005, “Consumption and portfolio choice over the life
cycle,” Review of Financial Studies, 18, 491–533.
Cochrane, J., 1999, “New Facts in Finance,” Economic Perspectives, 23(3), 36–58.
, 2007, “Portfolio Theory,” Lecture notes, University of Chicago.
, 2011, “Presidential Address: Discount Rates,” Journal of Finance, 66(4), 1047–1108.
Cooper, A. C., C. Y. Woo, and W. C. Dunkelberg, 1988, “Entrepreneurs’ Perceived Chances for
Success,” Journal of Business Venturing, 3, 97–108.
40
Cornell, B., 1999, “Risk, Duration, and Capital Budgeting: New Evidence on Some Old Ques-
tions,” Journal of Business, 72, 183–200.
Cronqvist, H., S. Siegel, and F. Yu, 2015, “Value versus growth investing: Why do different in-
vestors have different styles?,” Working paper, China Europe International Business School,
forthcoming Journal of Financial Economics.
Daniel, K., D. Hirshleifer, and A. Subrahmanyam, 2001, “Overconfidence, Arbitrage, and Equi-
librium Asset Pricing,” Journal of Finance, 56(3), 921–965.
DeBondt, W., and R. Thaler, 1985, “Does the Stock Market Overreact?,” Journal of Finance, 40,
793–805.
Dechow, P., R. Sloan, and M. Soliman, 2004, “Implied Equity Duration: A New Measure of Equity
Risk,” Review of Accounting Studies, 9, 197–228.
Døskeland, T. M., and H. K. Hvide, 2011, “Do Individual Investors Have Asymmetric Information
Based on Work Experience?,” Journal of Finance, 46(3), 1011–1041.
Du Rietz, G., M. Henrekson, and D. Waldenström, 2012, “The Swedish Inheritance and Gift Tax-
ation, 1885-2004,” working paper no. 936, Research Institute of Industrial Economics.
Eiling, E., 2013, “Industry-Specific Human Capital, Idiosyncratic Risk, and the Cross-Section of
Expected Stock Returns,” Journal of Finance, 63(1), 43–84.
Fama, E. F., and K. R. French, 1992, “The Cross-Section of Expected Stock Returns,” Journal of
Finance, 47(2), 427–465.
, 1993, “Common Risk Factors in the Returns on Stocks and Bonds,” Journal of Financial
Economics, 33(2), 3–56.
, 1995, “Size and Book-to-Market Factors in Earnings and Returns,” Journal of Finance,
50(1), 131–155.
, 1996, “Multifactor Explanations of Asset-Pricing Anomalies,” Journal of Finance, 51(1),
55–84.
41
, 1998, “Value versus Growth: The International Evidence,” Journal of Finance, 53(6),
1975–1999.
, 2012, “Size, Value, and Momentum in International Stock Returns,” Journal of Financial
Economics, 105, 457–472.
Floden, M., and J. Lindé, 2001, “Idiosyncratic Risk in the United States and Sweden: Is There a
Role for Government Insurance?,” Review of Economic Dynamics, 4, 406–437.
Garleanu, N., L. Kogan, and S. Panageas, 2012, “Displacement Risk and Asset Returns,” Journal
of Financial Economics, 105(3), 491–510.
Goetzmann, W., and A. Kumar, 2008, “Equity Portfolio Diversification,” Review of Finance, 12(3),
433–463.
Gomes, J., L. Kogan, and M. Yogo, 2009, “Durability of Output and Expected Stock Returns,”
Journal of Political Economy, 117, 941–986.
Gomes, J., L. Kogan, and L. Zhang, 2003, “Equilibrium Cross-Section of Returns,” Journal of
Political Economy, 111, 693–732.
Graham, B., and D. L. Dodd, 1934, Security Analysis. McGraw Hill, New York.
Graham, J. R., and A. Kumar, 2006, “Do Dividend Clienteles Exist? Evidence on Dividend Pref-
erences of Retail Investors,” Journal of Finance, 61(3), 1305–1336.
Greenwood, R., and A. Sheifer, 2013, “Expectations of Returns and Expected Returns,” Working
paper, Harvard University.
Griffin, J. M., 2003, “Are the Fama and French Factors Global or Country Specific?,” Review of
Financial Studies, 15(3), 783–803.
Grinblatt, M., and M. Keloharju, 2001, “How Distance, Language, and Culture Influence Stock-
holdings and Trades,” Journal of Finance, 56, 1053–1073.
42
Guiso, L., T. Jappelli, and D. Terlizzese, 1996, “Income Risk, Borrowing Constraints, and Portfolio
Choice,” American Economic Review, 86(1), 158–172.
Gulen, H., Y. Xing, and L. Zhang, 2011, “Value versus Growth: Time-Varying Expected Stock
Returns,” Financial Management, 40(2), 381–407.
Hansen, L. P., J. C. Heaton, and N. Li, 2008, “Consumption Strikes Back? Measuring Long-Run
Risk,” Journal of Political Economy, 116, 260–302.
Heaton, J., and D. Lucas, 2000, “Portfolio Choice and Asset Prices: The Importance of En-
trepreneurial Risk,” Journal of Finance, 55, 1163–1198.
Hogarth, R., 1987, Judgement and Choice, 2nd Edition. Wiley, New York, NY.
Holmlund, B., and D. Storrie, 2002, “Temporary Work in Turbulent Times: The Swedish Experi-
ence,” The Economic Journal, 112, 245–269.
Hou, K., G. A. Karolyi, and B.-C. Kho, 2011, “What Factors Drive Global Stock Returns?,” Review
of Financial Studies, 24(8), 2527–2574.
Huberman, G., 2001, “Familiarity Breeds Investment,” Review of Financial Studies, 14(3), 659–
680.
Ingersoll, J., 1987, Theory of Financial Decision Making. Rowman and Littlefield, Savage, Mary-
land.
Jagannathan, R., K. Kubota, and H. Takehara, 1998, “Relationship Between Labor-Income Risk
and Average Return: Empirical Evidence from the Japanese Stock Market,” Journal of Business,
71(3), 319–347.
Jagannathan, R., and Z. Wang, 1996, “The Conditional CAPM and the Cross-Section of Expected
Return,” Journal of Finance, 51(1), 3–53.
Jurek, J. W., and L. M. Viceira, 2011, “Optimal Value and Growth Tilts in Long-Horizon Portfo-
lios,” Review of Finance, 15(1), 29–74.
43
Kahneman, D., 2011, Thinking Fast and Slow. Penguin Books, New York, NY.
Kahneman, D., and A. Tversky, 1973, “On the Psychology of Prediction,” Psychological Review,
80(4), 237–251.
Keloharju, M., S. Knüpfer, and J. Linnainmaa, 2012, “Do Investors Buy What They Know? Prod-
uct Market Choices and Investment Decisions,” Review of Financial Studies, 25(10), 2921–
2958.
Kihlstrom, R. E., D. Romer, and S. Williams, 1981, “Risk Aversion and Random Initial Wealth,”
Econometrica, 49(4), 911–920.
Knüpfer, S., E. H. Rantapuska, and M. Sarvimäki, 2013, “Labor Market Experiences and Portfolio
Choice: Evidence from the Finnish Great Depression,” working paper, London Business School
and Aalto University.
Kogan, L., and D. Papanikolaou, 2014, “Growth Opportunities, Technology Shocks, and Asset
Prices,” Journal of Finance, 69(2), 675–718.
Kogan, L., D. Papanikolaou, and N. Stoffman, 2013, “Technological Innovation: Winners and
Losers,” working paper, MIT, Northwestern University, and Indiana University.
Koijen, R., H. Lustig, and S. Van Nieuwerburgh, 2014, “The Cross-Section and Time-Series of
Stock and Bond Returns,” working paper, London Business School, NBER Working Paper No.
15688.
Kumar, A., 2009a, “Dymamic Style Preferences of Individual Investors and Stock Returns,” Jour-
nal of Financial and Quantitative Analysis, 44(3), 607–640.
, 2009b, “Who Gambles in the Stock Market?,” Journal of Finance, 64(4), 1889–1933.
La Porta, R., 1996, “Expectations and the Cross-Section of Stock Returns,” Journal of Finance,
51(5), 1715–1742.
La Porta, R., J. Lakonishok, A. Shleifer, and R. W. Vishny, 1997, “Good News for Value Stocks:
Further Evidence on Market Efficiency,” Journal of Finance, 52(2), 859–874.
44
Lakonishok, J., A. Shleifer, and R. W. Vishny, 1994, “Contrarian Investment, Extrapolation, and
Risk,” Journal of Finance, 49, 1541–1578.
Langlois, H., 2013, “Asset Pricing with Return Asymmetries: Theory and Tests,” working paper,
McGill University.
Larsen, L., and C. Munk, 2012, “The Costs of Suboptimal Dynamic Asset Allocation: General
Results and Applications to Interest Rate Risk, Stock Volatility Risk, and Growth/Value Tilts,”
Journal of Economic Dynamics and Control, 36(2), 266–293.
Lemaître, G., 2007, “The Integration of Immigrants into the Labour Market: The Case of Sweden,”
working paper, London Business School, OECD Social, Employmeny and Migration Working
Paper No. 48.
Lettau, M., and S. Ludvigson, 2001, “Resurrecting the (C)CAPM: A Cross-Sectional Test When
Risk Premia are Time-Varying,” Journal of Political Economy, 109(61), 1238–1287.
Lettau, M., and J. Wachter, 2007, “Why Is Long-Horizon Equity Less Risky? A Duration-Based
Explanation of the Value Premium,” Journal of Finance, 62(1), 55–92.
Lewellen, J., and S. Nagel, 2006, “The Conditional CAPM Does Not Explain Asset-Pricing
Anomalies,” Journal of Financial Economics, 82(2), 289–314.
Liew, J., and M. Vassalou, 2000, “Can Book-to-Market, Size and Momentum Be Risk Factors that
Predict Economic Growth?,” Journal of Financial Economics, 57, 221–245.
Liu, Q., L. Lu, B. Sun, and H. Yan, 2015, “A Model of Anomaly Discovery,” working paper, Yale
University.
Long, J. B., 1974, “Stock Prices, Inflation, and the Term Structure of Interest Rates,” Journal of
Financial Economics, 1, 131–170.
Lustig, H., and S. van Nieuwerburgh, 2005, “Housing Collateral, Consumption Insurance, and
Risk Premia: An Empirical Perspective,” Journal of Finance, 60(3), 1167–1219.
45
Lynch, A. W., 2001, “Portfolio Choice and Equity Characteristics: Characterizing the Hedging
Demands Induced by Return Predictability,” Journal of Financial Economics, 62, 67–130.
Malmendier, U., and S. Nagel, 2011, “Learning from Inflation Experiences,” working paper, Uni-
versity of California at Berkeley and Stanford University.
Massa, M., and A. Simonov, 2006, “Hedging, Familiarity, and Portfolio Choice,” Review of Finan-
cial Studies, 19(2), 633–685.
Merton, R., 1973, “An Intertemporal Capital Asset Pricing Model,” Econometrica, 59, 867–887.
Mitton, T., and K. Vorkink, 2007, “Equilibrium Underdiversification and the Preference for Skew-
ness,” Review of Financial Studies, 20(4), 1255–1288.
Moskowitz, T., and A. Vissing-Jørgensen, 2002, “The Returns to Entrepreneurial Investment: A
Private Equity Premium Puzzle?,” American Economic Review, 92(4), 745–778.
Nagel, S., and K. Singleton, 2011, “Estimation and Evaluation of Conditional Asset-Pricing Mod-
els,” Journal of Finance, 66(3), 873–909.
Oskamp, S., 1965, “Overconfidence in Case-Study Judgments,” Journal of Consulting Psychology,
29, 261–265.
Palacios-Huerta, I., 2003, “The Robustness of the Conditional CAPM with Human Capital,” Jour-
nal of Financial Econometrics, 1(2), 272–289.
Parlour, C., and J. Walden, 2011, “General Equilibrium Returns to Human and Investment Capital
under Moral Hazard,” Review of Economic Studies, 78, 394–428.
Petkova, R., and L. Zhang, 2005, “Is Value Riskier than Growth?,” Journal of Financial Eco-
nomics, 78, 187–202.
Polkovnichenko, V., 2005, “Household Portfolio Diversification: A Case for Rank-Dependent
Preferences,” Review of Financial Studies, 18(4), 1467–1502.
46
Ridley, M., 2003, Nature via Nurture: Genes, Experience, and What Makes Us Human. Harper
Collins, New York, NY.
Rosenberg, B., K. Reid, and R. Lanstein, 1985, “Persuasive Evidence of Market Inefficiency,”
Journal of Portfolio Management, 11, 9–11.
Roussanov, N., 2013, “Composition of Wealth, Conditioning Information, and the Cross-Section
of Returns,” Journal of Financial Economics, 111(1), 352–380.
Santos, T., and P. Veronesi, 2006, “Labor Income and Predictable Stocks Returns,” Review of
Financial Studies, 19(1), 1–44.
Shleifer, A., 2000, Inefficient Markets: An Introduction to Behavioral Finance. Oxford University
Press, Oxford, UK.
Skinner, D. J., and R. G. Sloan, 2002, “Earnings Surprises, Growth Expectations, and Stock Re-
turns or DonâAZt Let an Earnings Torpedo Sink Your Portfolio,” Review of Accounting Studies,
7, 289–312.
Storesletten, K., C. I. Telmer, and A. Yaron, 2004, “Cyclical Dynamics in Idiosyncratic Labor
Market Risk,” Journal of Political Economy, 112(3), 695–717.
Sylvain, S., 2013, “Does Human Capital Risk Explain the Value Premium Puzzle?,” working paper,
University of Chicago.
Wachter, J., 2002, “Portfolio and Consumption Decisions under Mean-Reverting Returns: An
Exact Solution for Complete Markets,” Journal of Financial and Quantitative Analysis, 37(1),
63–91.
Yogo, M., 2006, “A Consumption-Based Explanation of Expected Stock Returns,” Journal of Fi-
nance, 61, 539–580.
Zhang, L., 2005, “The Value Premium,” Journal of Finance, 60(1), 267–284.
47
Tabl
e I
Sum
mar
y St
atis
tics
The
tabl
ere
ports
sum
mar
yst
atis
tics
onth
efin
anci
alan
dde
mog
raph
icch
arac
teris
tics
(Pan
elA)
and
portf
olio
char
acte
ristic
s(P
anel
B)of
parti
cipa
ting
Swed
ish
hous
ehol
dsat
the
end
of20
03.W
eco
nsid
erris
kyas
setm
arke
tpar
ticip
ants
(firs
tset
ofco
lum
ns),
mut
ualf
und
hold
ers
(sec
ond
seto
fcol
umns
),di
rect
stoc
khol
ders
(third
seto
fcol
umns
)an
ddi
rect
stoc
khol
ders
sorte
dby
the
num
ber
ofst
ocks
that
they
own
(last
seto
fthr
eeco
lum
ns)
For
each
dire
ctst
ockh
olde
rs(th
irdse
tofc
olum
ns),
and
dire
ctst
ockh
olde
rsso
rted
byth
enu
mbe
rof
stoc
ksth
atth
eyow
n(la
stse
toft
hree
colu
mns
).Fo
rea
chch
arac
teris
tic,
we
repo
rtth
ecr
oss-
sect
iona
lmea
nan
dst
anda
rdde
viat
ion
inea
chsa
mpl
e.Th
ebo
ttom
row
sof
Pane
lBta
bula
teth
efra
ctio
nof
the
aggr
egat
ew
ealth
ofris
kyas
set
mar
ket
parti
cipa
nts
held
bysp
ecifi
cgr
oups
ofin
vest
ors.
The
calc
ulat
ions
are
base
don
the
repr
esen
tativ
epa
nelo
fho
useh
olds
over
the
1999
to20
07pe
riod
defin
edin
Sec
tion
2.2.
Allv
aria
bles
are
desc
ribed
inTa
ble
A.
Pane
l A: F
inan
cial
and
Dem
ogra
phic
Cha
ract
eris
tics
AllP
artic
ipan
tsFu
ndho
lder
sSt
ockh
olde
rsSt
ockh
olde
rsSo
rted
Mea
nSt
anda
rdM
ean
Stan
dard
Mea
nSt
anda
rdde
viatio
nde
viatio
nde
viatio
nM
ean
Mea
nM
ean
Fina
ncia
l Cha
ract
eris
tics
Fina
ncia
l wea
lth ($
)48
,849
121,
578
50,6
1412
1,09
966
,478
152,
690
37,1
2360
,091
126,
493
Rid
til
lt
tlth
($)
137
108
184
525
138
327
179
024
165
020
215
680
129
854
169
241
229
107
1-2
3-4
5+
All P
artic
ipan
tsFu
ndho
lder
sSt
ockh
olde
rsBy
Num
ber o
f Sto
cks
Own
edSt
ockh
olde
rs S
orte
d
Resi
dent
ial r
eal e
stat
e we
alth
($)
137,
108
184,
525
138,
327
179,
024
165,
020
215,
680
129,
854
169,
241
229,
107
Com
mer
cial
real
est
ate
weal
th ($
)19
,581
112,
626
19,5
2011
1,89
027
,255
135,
585
21,5
9830
,115
36,1
31Le
vera
ge ra
tio0.
661.
130.
651.
090.
530.
910.
650.
460.
34H
uman
Cap
ital a
nd In
com
e R
isk
Hum
an c
apita
l ($)
955,
680
515,
879
972,
402
513,
389
993,
114
545,
932
929,
517
1,03
0,77
01,
089,
285
Inco
me
($)
46,1
8431
,316
46,7
8530
,687
50,0
6637
,029
44,9
0251
,133
59,1
83Se
lf-em
ploy
men
t dum
my
0.04
0.20
0.04
0.19
0.05
0.22
0.05
0.05
0.05
Unem
ploy
men
t dum
my
0.08
0.27
0.07
0.26
0.07
0.25
0.08
0.06
0.05
Con
ditio
nal i
ncom
e vo
latili
ty0.
160.
120.
160.
110.
170.
120.
170.
170.
18D
emog
raph
ic C
hara
cter
istic
sAg
e46
.27
10.7
346
.06
10.6
947
.60
10.5
846
.82
47.5
549
.12
Mal
e ho
useh
old
head
dum
my
0.64
0.48
0.63
0.48
0.69
0.46
0.66
0.70
0.73
High
sch
ool d
umm
y0.
850.
360.
850.
350.
860.
350.
840.
860.
90Po
st-h
igh
scho
ol d
umm
y0.
370.
480.
370.
480.
420.
490.
350.
420.
53Ec
onom
ics
educ
atio
n du
mm
y0.
120.
320.
120.
320.
130.
340.
120.
140.
16Im
mig
ratio
n du
mm
y0.
080.
270.
080.
260.
080.
270.
080.
090.
07Fa
mily
siz
e2.
531.
402.
611.
402.
521.
372.
422.
562.
69Nu
mbe
r of o
bser
vatio
ns71
,639
71,6
3962
,972
62,9
7242
,153
42,1
5322
,522
7,78
611
,845
,,
,,
,,
,,
,
Tabl
e I
Sum
mar
y St
atis
tics
-Con
tinue
d
Pane
l B: P
ortfo
lio C
hara
cter
istic
sAl
lPti
it
Fdh
ldSt
khld
Stkh
ldS
td
Mea
nSt
anda
rdM
ean
Stan
dard
Mea
nSt
anda
rdde
viatio
nde
viatio
nde
viatio
nM
ean
Mea
nM
ean
Port
folio
Cha
ract
eris
tics
Risk
y sh
are
0.40
0.27
0.42
0.26
0.46
0.27
0.37
0.49
0.61
By N
umbe
r of S
tock
s O
wned
1-2
3-4
5+
All P
artic
ipan
tsFu
ndho
lder
sSt
ockh
olde
rsSt
ockh
olde
rs S
orte
d
Shar
e of
dire
ct s
tock
hold
ings
in ri
sky
portf
olio
0.29
0.37
0.19
0.28
0.49
0.37
0.44
0.48
0.58
Shar
e of
pop
ular
sto
cks
0.71
0.37
0.71
0.36
0.71
0.37
0.79
0.71
0.57
Shar
e of
pro
fess
iona
lly c
lose
sto
cks
0.16
0.32
0.16
0.31
0.16
0.32
0.15
0.17
0.18
Num
ber o
f sto
cks
2.59
5.15
2.53
5.30
4.40
6.10
1.35
3.42
10.8
5Nu
mbe
r of f
unds
4.11
4.51
4.68
4.53
4.55
5.19
3.49
4.90
6.34
Shar
e of
Agg
rega
te W
ealth
ggg
Risk
y po
rtfol
io1.
000.
940.
860.
180.
130.
54St
ock
portf
olio
1.00
0.85
1.00
0.09
0.11
0.80
Fund
por
tfolio
1.00
1.00
0.75
0.25
0.14
0.36
Num
ber o
f obs
erva
tions
71,6
3971
,639
62,9
7262
,972
42,1
5342
,153
22,5
227,
786
11,8
45
Tabl
e II
Cro
ss-S
ectio
nal D
istr
ibut
ion
of th
e Va
lue
Load
ing
The
tabl
ere
ports
sum
mar
yst
atis
tics
onth
ecr
oss-
sect
iona
ldis
tribu
tion
ofth
eva
lue
load
ing
atth
een
dof
2003
fors
ome
ofth
em
ain
fam
ilies
ofas
sets
and
hous
ehol
dpo
rtfol
ios
used
inth
epa
per.
The
colu
mns
repo
rt(i)
the
valu
elo
adin
gof
the
aggr
egat
epo
rtfol
io(ii
)th
ecr
oss
sect
iona
ldis
tribu
tion
ofth
eva
lue
load
ing
and
(iii)
the
valu
elo
adin
gsp
read
betw
een
the
aggr
egat
epo
rtfol
io,(
ii)th
ecr
oss-
sect
iona
ldis
tribu
tion
ofth
eva
lue
load
ing,
and
(iii)
the
valu
elo
adin
gsp
read
betw
een
the
top
and
botto
mde
cile
s.Th
efir
stro
wco
nsid
ers
stoc
kslis
ted
onth
eSt
ockh
olm
Stoc
kEx
chan
gean
dth
ese
cond
row
cons
ider
sal
lSw
edis
hris
kym
utua
lfun
ds.H
ouse
hold
aggr
egat
epo
rtfol
ios
are
cons
truct
edby
addi
ngup
the
mar
ketv
alue
sof
indi
vidu
alho
useh
old
hold
ings
.
Valu
e Lo
adin
g
Aggr
egat
eSp
read
Portf
olio
Mea
n10
th25
th50
th75
th90
th(9
0th
- 10t
h)A
sset
sSt
ocks
list
ed o
n St
ockh
olm
Sto
ck E
xcha
nge
-0.1
5-0
.87
-3.2
2-1
.57
-0.3
70.
090.
944.
16Fu
nds
-0.1
0-0
.15
-0.4
1-0
.26
-0.1
00.
010.
200.
61H
ouse
hold
s
Cro
ss-S
ectio
nal D
istri
butio
n
All p
artic
ipan
ts -
Risk
y po
rtfol
io-0
.26
-0.3
0-0
.94
-0.4
6-0
.18
0.00
0.10
1.04
- St
ock
portf
olio
-0.3
6-0
.58
-1.2
0-1
.09
-0.5
30.
110.
391.
58 -
Fund
por
tfolio
-0.1
8-0
.20
-0.5
7-0
.30
-0.1
40.
000.
080.
65Fu
ndho
lder
s- R
isky
portf
olio
-0.2
5-0
.25
-0.7
1-0
.40
-0.1
7-0
.01
0.09
0.80
Ri
sky
portf
olio
0.25
0.25
0.71
0.40
0.17
0.01
0.09
0.80
- St
ock
portf
olio
-0.3
5-0
.57
-1.1
7-1
.06
-0.5
20.
100.
381.
55 -
Fund
por
tfolio
-0.1
8-0
.20
-0.5
7-0
.30
-0.1
40.
000.
080.
65D
irect
sto
ckho
lder
s -
Risk
y po
rtfol
io-0
.28
-0.3
8-1
.07
-0.6
1-0
.24
-0.0
20.
111.
18 -
Stoc
k po
rtfol
io-0
.36
-0.5
8-1
.20
-1.0
9-0
.53
0.11
0.39
1.58
-Fun
dpo
rtfol
io-0
19-0
22-0
58-0
33-0
16-0
030
070
65 -
Fund
por
tfolio
-0.1
9-0
.22
-0.5
8-0
.33
-0.1
6-0
.03
0.07
0.65
Tabl
e III
Pane
l Reg
ress
ion
of th
e Va
lue
Load
ing
on C
hara
cter
istic
s
This
tabl
ere
ports
pool
edre
gres
sion
sof
the
valu
elo
adin
gon
hous
ehol
dch
arac
teris
tics
and
year
,in
dust
ry,
and
coun
tyfix
edef
fect
s.Th
eva
lue
load
ing
isco
mpu
ted
atth
ele
velo
fthe
risky
portf
olio
inco
lum
n(1
),th
est
ock
portf
olio
inco
lum
n(2
),an
dth
efu
ndpo
rtfol
ioin
colu
mn
(3)
We
regr
ess
the
risky
shar
eon
the
sam
ech
arac
teris
tics
and
fixed
effe
cts
inco
lum
n(4
)Th
eco
mpu
tatio
nsar
eba
sed
onth
e(3
).W
ere
gres
sth
eris
kysh
are
onth
esa
me
char
acte
ristic
san
dfix
edef
fect
sin
colu
mn
(4).
The
com
puta
tions
are
base
don
the
repr
esen
tativ
epa
nelo
fhou
seho
lds
over
the
1999
to20
07pe
riod
defin
edin
Sect
ion
2.2.
Allv
aria
bles
are
desc
ribed
inTa
ble
A.St
anda
rder
rors
are
clus
tere
dat
the
hous
ehol
dle
vel.
(1)
(2)
(3)
(4)
Risk
y Sh
are
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ngRi
sky
Portf
olio
Stoc
k Po
rtfol
ioFu
nd P
ortfo
lioD
epen
dent
Var
iabl
e:
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tEs
timat
et-s
tat
Fina
ncia
l Cha
ract
eris
tics
Log
finan
cial
wea
lth
0.01
712
.44
0.05
016
.15
0.01
214
.57
0.09
513
5.95
Log
resi
dent
ial r
eal e
stat
e0.
001
1.75
0.00
34.
550.
000
-0.2
70.
000
3.32
Log
com
mer
cial
real
est
ate
0.00
13.
970.
007
12.3
60.
000
0.43
-0.0
02-1
1.89
(1)
(2)
(3)
(4)
Leve
rage
ratio
0.00
00.
30-0
.008
-1.7
3-0
.001
-0.9
8-0
.008
-14.
46H
uman
Cap
ital a
nd In
com
e R
isk
Log
hum
an c
apita
l -0
.052
-9.5
0-0
.103
-9.5
0-0
.021
-6.6
30.
016
5.92
Log
inco
me
-0.0
46-1
1.35
-0.0
44-5
.75
-0.0
29-1
2.87
-0.0
62-2
9.50
Self-
empl
oym
ent d
umm
y-0
.034
-4.4
1-0
.037
-2.6
6-0
.011
-2.6
2-0
.047
-13.
49Un
empl
oym
ent d
umm
y-0
.017
-3.9
9-0
.021
-2.0
3-0
.005
-1.9
7-0
.012
-5.9
2C
ondi
tiona
l inc
ome
vola
tility
-0.3
53-2
1.84
-0.3
38-1
0.98
-0.1
16-1
3.28
-0.0
62-9
.24
Dem
ogra
phic
Cha
ract
eris
tics
Age
0.00
316
.02
0.00
923
.50
0.00
15.
53-0
.002
-26.
14M
ale
hous
ehol
d he
ad d
umm
y-0
.062
-18.
48-0
.106
-13.
57-0
.013
-5.8
50.
014
8.62
High
sch
ool d
umm
y-0
.014
-3.3
8-0
.035
-3.4
3-0
.006
-2.1
60.
023
11.2
0Po
st-h
igh
scho
ol d
umm
y-0
.016
-4.6
40.
016
2.00
-0.0
15-6
.89
0.03
419
.95
gy
Econ
omic
s ed
ucat
ion
dum
my
-0.0
27-5
.94
-0.0
11-1
.09
-0.0
14-4
.76
0.01
14.
69Im
mig
ratio
n du
mm
y-0
.066
-11.
13-0
.135
-10.
33-0
.003
-0.9
5-0
.007
-2.6
1Fa
mily
siz
e0.
036
24.6
00.
024
7.42
0.01
719
.23
-0.0
07-1
0.44
Adju
sted
R2
2.37
%3.
95%
0.94
%16
.57%
Num
ber o
f obs
erva
tions
589,
561
331,
693
523,
798
589,
561
Tabl
e IV
Valu
e Lo
adin
gs o
f Inv
esto
r Sub
grou
ps
This
tabl
ere
ports
pool
edre
gres
sion
sof
the
valu
elo
adin
gof
the
risky
portf
olio
onho
useh
old
char
acte
ristic
san
dye
ar,
indu
stry
,an
dco
unty
fixed
effe
cts
estim
ated
over
diffe
rent
subs
ets
ofin
vest
ors.
The
regr
essi
ons
are
sim
ilart
oth
eba
selin
ere
gres
sion
sin
Tabl
eIII
,but
we
cons
ider
subg
roup
sof
risky
asse
tmar
ketp
artic
ipan
ts:f
und
hold
ers
inco
lum
n(1
)di
rect
stoc
khol
ders
inco
lum
n(2
)an
ddi
rect
stoc
khol
ders
sorte
dby
the
num
bero
fow
ned
risky
asse
tmar
ketp
artic
ipan
ts:f
und
hold
ers
inco
lum
n(1
),di
rect
stoc
khol
ders
inco
lum
n(2
),an
ddi
rect
stoc
khol
ders
sorte
dby
the
num
bero
fow
ned
stoc
ksin
colu
mns
(3)t
o(5
).Th
esu
bgro
ups
are
obta
ined
from
the
repr
esen
tativ
epa
nelo
fhou
seho
lds
over
the
1999
to20
07pe
riod
defin
edin
Sect
ion
2.2.
All
varia
bles
are
desc
ribed
inTa
ble
A.St
anda
rder
rors
are
clus
tere
dat
the
hous
ehol
dle
vel.
Five
orM
ore
Stoc
ksSt
ockh
olde
rs S
orte
d by
Num
ber o
f Sto
cks
Own
edO
neor
Two
Stoc
ksTh
ree
orFo
urSt
ocks
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ng o
f Ris
ky P
ortfo
lio
Fund
hold
ers
Stoc
khol
ders
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tFi
nanc
ial C
hara
cter
istic
sLo
g fin
anci
al w
ealth
0.
010
9.27
0.04
719
.97
0.04
011
.82
0.09
116
.70
0.06
718
.18
Log
resi
dent
ial r
eal e
stat
e0.
000
-0.4
50.
002
4.48
0.00
22.
650.
002
2.01
0.00
44.
92
(1)
(2)
(3)
(4)
(5)
Five
or M
ore
Stoc
ksO
ne o
r Two
Sto
cks
Thre
e or
Fou
r Sto
cks
Fund
hold
ers
Stoc
khol
ders
Log
com
mer
cial
real
est
ate
0.00
12.
340.
003
7.50
0.00
48.
400.
002
3.38
0.00
10.
99Le
vera
ge ra
tio-0
.001
-0.9
6-0
.010
-3.0
3-0
.005
-1.2
8-0
.028
-3.4
2-0
.041
-5.1
5H
uman
Cap
ital a
nd In
com
e R
isk
Log
hum
an c
apita
l -0
.039
-9.2
9-0
.073
-9.1
5-0
.068
-5.8
4-0
.060
-3.5
4-0
.067
-5.8
7Lo
g in
com
e-0
.047
-15.
12-0
.043
-7.3
8-0
.047
-5.6
3-0
.036
-2.7
3-0
.044
-5.2
9Se
lf-em
ploy
men
t dum
my
-0.0
25-4
.51
-0.0
24-2
.29
-0.0
24-1
.48
-0.0
14-0
.65
-0.0
27-1
.90
py
yUn
empl
oym
ent d
umm
y-0
.009
-2.8
3-0
.031
-3.9
3-0
.042
-3.9
1-0
.012
-0.7
5-0
.017
-1.4
7C
ondi
tiona
l inc
ome
vola
tility
-0.2
47-2
0.98
-0.4
03-1
7.01
-0.3
79-1
0.78
-0.4
44-9
.81
-0.4
13-1
2.88
Dem
ogra
phic
Cha
ract
eris
tics
Age
0.00
216
.78
0.00
517
.40
0.00
511
.92
0.00
58.
880.
005
11.6
5M
ale
hous
ehol
d he
ad d
umm
y-0
.037
-14.
08-0
.085
-16.
28-0
.077
-10.
47-0
.113
-11.
20-0
.076
-9.9
6Hi
ghsc
hool
dum
my
-000
9-2
76-0
024
-346
-002
9-3
20-0
008
-057
-001
3-1
15Hi
gh s
choo
l dum
my
0.00
92.
760.
024
3.46
0.02
93.
200.
008
0.57
0.01
31.
15Po
st-h
igh
scho
ol d
umm
y-0
.019
-6.7
50.
005
0.90
-0.0
03-0
.38
0.01
61.
620.
021
2.71
Econ
omic
s ed
ucat
ion
dum
my
-0.0
20-5
.47
-0.0
18-2
.75
-0.0
35-3
.54
-0.0
14-1
.06
0.01
11.
19Im
mig
ratio
n du
mm
y-0
.031
-6.9
3-0
.120
-12.
39-0
.115
-8.6
5-0
.108
-5.7
6-0
.138
-9.4
6Fa
mily
siz
e0.
025
22.2
40.
040
17.7
40.
046
14.5
40.
038
8.36
0.03
09.
00Ad
just
ed R
22.
02%
4.45
%3.
50%
7.54
%7.
22%
Nb
fb
ti52
379
833
169
317
570
759
697
9628
9Nu
mbe
r of o
bser
vatio
ns52
3,79
833
1,69
317
5,70
759
,697
96,2
89
Tabl
e V
Alte
rnat
ive
Ris
k M
easu
res
This
tabl
ere
ports
the
effe
cts
ofad
ditio
nalr
eale
stat
e,le
vera
ge,a
ndfa
mily
size
varia
bles
onth
eva
lue
load
ing
inth
epr
esen
ceof
year
,ind
ustry
,and
coun
tyfix
edef
fect
s.Pa
nelA
incl
udes
mea
sure
sof
dem
eane
dre
ales
tate
wea
lthin
tera
cted
with
dem
eane
dle
vera
geW
eco
nduc
tthe
estim
atio
non
the
repr
esen
tativ
epa
nelo
fhou
seho
lds
over
the
1999
to20
07pe
riod
defin
edin
Sect
ion
leve
rage
.We
cond
uctt
hees
timat
ion
onth
ere
pres
enta
tive
pane
lofh
ouse
hold
sov
erth
e19
99to
2007
perio
dde
fined
inSe
ctio
n2.
2.Pa
nelB
incl
udes
adu
mm
yva
riabl
efo
rhav
ing
ach
ilddu
ring
the
year
and
adu
mm
yva
riabl
efo
rhav
ing
twin
sdu
ring
the
year
.Th
ees
timat
ion
isco
nduc
ted
ona
sam
ple
ofho
useh
olds
that
incl
udes
allh
ouse
hold
sw
ithne
wbo
rntw
ins.
The
regr
essi
ons
are
othe
rwis
esi
mila
rto
the
base
line
regr
essi
onin
Tabl
eIII
,and
the
full
estim
atio
nde
tails
and
resu
ltsar
eav
aila
ble
inth
eIn
tern
etA
ppen
dix.
All
varia
bles
are
desc
ribed
inTa
ble
A.St
anda
rder
rors
are
clus
tere
dat
the
hous
ehol
dle
vel.
Pane
lA:R
ealE
stat
eIn
tera
cted
with
Leve
rage
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tLo
g re
side
ntia
l rea
l est
ate
0.00
01.
370.
003
3.79
0.00
0-0
.44
Risk
y Po
rtfol
ioSt
ock
Portf
olio
Fund
Por
tfolio
Pane
l A: R
eal E
stat
e In
tera
cted
with
Lev
erag
e
(1)
(2)
(3)
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ng
gLo
g co
mm
erci
al re
al e
stat
e0.
001
2.01
0.00
79.
870.
000
-0.8
8Lo
g re
side
ntia
l rea
l est
ate
× Le
vera
ge ra
tio-0
.001
-4.2
8-0
.004
-4.8
80.
000
-1.4
0Lo
g co
mm
erci
al re
al e
stat
e ×
Leve
rage
ratio
-0.0
01-3
.13
0.00
0-0
.45
-0.0
01-3
.48
Leve
rage
ratio
-0.0
12-4
.11
-0.0
40-5
.10
-0.0
04-2
.29
Dd
tVi
blV
lL
diPa
nel B
: Chi
ldre
n
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tD
umm
y fo
r hav
ing
child
ren
0.08
717
.21
0.02
82.
170.
038.
20D
umm
y fo
r hav
ing
twin
s-0
.020
-2.6
3-0
.039
-1.8
3-0
.01
-1.1
5
(1)
(2)
(3)
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ngRi
sky
Portf
olio
Stoc
k Po
rtfol
ioFu
nd P
ortfo
lio
Tabl
e VI
Econ
omic
Sig
nific
ance
This
tabl
ere
ports
the
impa
cton
the
valu
elo
adin
gof
life-
cycl
eva
riatio
nin
age
and
finan
cial
char
acte
ristic
s.W
eus
eas
benc
hmar
ksa
30-y
ear
old
hous
ehol
dhe
ad,a
50-y
ear-o
ldho
useh
old
head
,and
a70
-yea
rol
dho
useh
old
head
,to
whi
chw
eas
sign
the
aver
age
char
acte
ristic
sof
hous
ehol
dsin
thei
rre
spec
tive
coho
rtsin
2003
The
impa
ctof
chan
ges
inw
eas
sign
the
aver
age
char
acte
ristic
sof
hous
ehol
dsin
thei
rre
spec
tive
coho
rtsin
2003
.Th
eim
pact
ofch
ange
sin
char
acte
ristic
sis
asse
ssed
usin
gth
eba
selin
ere
gres
sion
coef
ficie
nts
inTa
ble
III.A
llva
riabl
esar
ede
scrib
edin
Tabl
eA.
30→
5050→
7030→
5050→
7030→
5050→
70O
bser
ved
chan
ge in
val
ue lo
adin
g0.
090.
140.
230.
250.
020.
04
Risk
y Po
rtfol
ioSt
ock
Portf
olio
Fund
Por
tfolio
gg
Pred
icte
d ch
ange
Fina
ncia
l Cha
ract
eris
tics
Log
finan
cial
wea
lth
0.01
50.
008
0.04
20.
022
0.01
00.
005
Log
resi
dent
ial r
eal e
stat
e0.
001
0.00
00.
005
-0.0
030.
000
0.00
0Lo
g co
mm
erci
al re
al e
stat
e0.
001
0.00
40.
010
0.02
50.
000
0.00
0Le
vera
gera
tio0.
000
0.00
00.
002
0.00
10.
000
0.00
0Le
vera
ge ra
tio0.
000
0.00
00.
002
0.00
10.
000
0.00
0H
uman
Cap
ital a
nd In
com
e R
isk
Log
hum
an c
apita
l 0.
024
0.03
70.
048
0.07
30.
010
0.01
5Lo
g in
com
e-0
.006
-0.0
09-0
.006
-0.0
08-0
.004
-0.0
06Se
lf-em
ploy
men
t dum
my
0.00
0-0
.009
0.00
0-0
.009
0.00
0-0
.003
Unem
ploy
men
t dum
my
0.00
00.
001
0.00
00.
001
0.00
00.
000
Con
ditio
nali
ncom
evo
latili
ty0
004
000
00
003
000
00
001
000
0C
ondi
tiona
l inc
ome
vola
tility
-0.0
040.
000
-0.0
030.
000
-0.0
010.
000
Dem
ogra
phic
Cha
ract
eris
tics
Age
0.05
50.
055
0.17
70.
177
0.01
20.
012
Mal
e ho
useh
old
head
dum
my
-0.0
01-0
.018
-0.0
01-0
.031
0.00
0-0
.004
High
sch
ool d
umm
y0.
001
0.00
20.
003
0.00
60.
001
0.00
1Po
st-h
igh
scho
ol d
umm
y0.
001
0.00
7-0
.001
-0.0
070.
001
0.00
6E
id
tid
000
20
001
000
10
000
000
10
001
Econ
omic
s ed
ucat
ion
dum
my
0.00
2-0
.001
0.00
10.
000
0.00
1-0
.001
Imm
igra
tion
dum
my
0.00
00.
003
0.00
00.
007
0.00
00.
000
Fam
ily s
ize
-0.0
35-0
.015
-0.0
24-0
.011
-0.0
16-0
.007
Cha
nge
due
to a
ge a
nd fi
nanc
ial c
hara
cter
istic
s0.
090
0.09
40.
278
0.28
70.
028
0.02
7Fr
actio
n du
e to
age
61
.12%
58.1
0%63
.61%
61.5
4%41
.14%
42.9
0%
Tabl
e VI
ISy
stem
atic
Lab
or In
com
e R
isk
This
tabl
ein
vest
igat
esth
efa
ctor
stru
ctur
eof
indu
stry
-leve
linc
ome
grow
than
dits
impl
icat
ions
forh
ouse
hold
finan
cial
portf
olio
s.Fo
rea
chof
the
70tw
o-di
git
indu
strie
s,w
ere
gres
sse
ctor
alin
com
egr
owth
onag
greg
ate
inco
me
grow
than
dre
port
inPa
nel
Ath
edi
strib
utio
nof
the
corre
spon
ding
slop
esan
dR
2co
effic
ient
sPa
nelB
repo
rtspo
oled
regr
essi
ons
ofth
eho
useh
old
portf
olio
valu
edi
strib
utio
nof
the
corre
spon
ding
slop
esan
dR
coef
ficie
nts.
Pane
lBre
ports
pool
edre
gres
sion
sof
the
hous
ehol
dpo
rtfol
iova
lue
load
ing
on(i)
the
load
ing
ofho
useh
old
inco
me
onag
greg
ate
inco
me,
(ii)s
tand
ard
char
acte
ristic
s,an
d(ii
i)ye
ar,i
ndus
try,a
ndco
unty
fixed
effe
cts.
The
hous
ehol
din
com
elo
adin
gis
defin
edas
the
wei
ghte
dav
erag
elo
adin
gof
the
sect
ors
inw
hich
the
adul
tsin
the
hous
ehol
dar
eem
ploy
ed.T
hefu
llre
sults
are
repo
rted
inth
eIn
tern
etAp
pend
ix.T
heco
mpu
tatio
nsar
eba
sed
onth
ere
pres
enta
tive
pane
lofh
ouse
hold
sov
erth
e19
99to
2007
perio
dde
fined
inS
ectio
n2.
2.St
anda
rder
rors
are
clus
tere
dat
the
hous
ehol
dle
vel.
Pane
lA:C
ross
Sect
iona
lDis
tribu
tion
ofIn
com
eEx
posu
reto
Aggr
egat
eIn
com
eSh
ocks
Mea
n10
th25
th50
th75
th90
thLo
adin
g of
sec
tora
l inc
ome
on a
ggre
gate
inco
me
1.03
0.81
0.95
1.05
1.15
1.22
R2
0.88
0.74
0.83
0.92
0.95
0.96
Pane
l B: I
ncom
e Ex
posu
re to
Agg
rega
te In
com
e Sh
ocks
Pane
l A: C
ross
-Sec
tiona
l Dis
tribu
tion
of In
com
e Ex
posu
re to
Agg
rega
te In
com
e Sh
ocks
Dep
ende
ntVa
riabl
e:Va
lue
Load
ing
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tLo
adin
g of
sec
tora
l inc
ome
on a
ggre
gate
inco
me
-0.2
05-1
0.05
-0.2
00-3
.86
-0.0
77-5
.54
Con
ditio
nal i
ncom
e vo
latili
ty-0
.342
-20.
45-0
.330
-10.
30-0
.111
-12.
23
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ngRi
sky
Portf
olio
Stoc
k Po
rtfol
ioFu
nd P
ortfo
lio(1
)(2
)(3
)
Tabl
e VI
IIA
ctiv
e R
ebal
anci
ng o
f the
Val
ue L
oadi
ng
This
tabl
ere
ports
pool
edre
gres
sion
sof
the
activ
ech
ange
inth
eva
lue
load
ing
on(i)
the
pass
ive
chan
gein
the
valu
elo
adin
gan
d(ii
)the
lagg
edva
lue
load
ing.
We
cond
uctt
hean
alys
isat
the
leve
loft
heris
kypo
rtfol
ioin
colu
mns
(1)a
nd(2
),th
est
ock
portf
olio
inco
lum
ns(3
)and
(4),
and
the
fund
portf
olio
inco
lum
ns(5
)an
d(6
)Fo
rea
chpo
rtfol
iow
ere
port
the
regr
essi
onw
ithan
dw
ithou
tla
gged
hous
ehol
dsch
arac
teris
tics
All
varia
bles
are
dem
eane
dea
chye
arTh
ean
d(6
).Fo
rea
chpo
rtfol
io,
we
repo
rtth
ere
gres
sion
with
and
with
out
lagg
edho
useh
olds
char
acte
ristic
s.A
llva
riabl
esar
ede
mea
ned
each
year
.Th
eco
mpu
tatio
nsar
eba
sed
onth
ere
pres
enta
tive
pane
lofh
ouse
hold
sov
erth
e19
99to
2007
perio
dde
fined
inSe
ctio
n2.
2.St
anda
rder
rors
are
clus
tere
dat
the
hous
ehol
dle
vel.
Dep
ende
nt V
aria
ble:
Act
ive C
hang
e of
Val
ue L
oadi
ng
(6)
Fund
Por
tfolio
(1)
(2)
(3)
(4)
Risk
y Po
rtfol
ioSt
ock
Portf
olio
(5)
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tEs
timat
et-s
tat
Valu
e Lo
adin
g Va
riabl
esPa
ssive
cha
nge
in th
e va
lue
load
ing
-0.3
56-2
7.63
-0.3
56-2
7.61
-0.3
72-2
7.30
-0.3
75-2
7.40
-0.2
83-2
7.95
-0.2
84-2
7.98
Lagg
ed v
alue
load
ing
-0.1
16-4
1.95
-0.1
19-4
2.55
-0.0
78-3
8.24
-0.0
82-3
9.15
-0.1
10-5
4.30
-0.1
11-5
4.41
Lagg
ed F
inan
cial
Cha
ract
eris
tics
(6)
(1)
(2)
(3)
(4)
(5)
Log
finan
cial
wea
lth
0.00
24.
810.
005
6.08
0.00
00.
90Lo
g re
side
ntia
l rea
l est
ate
0.00
01.
970.
001
3.71
0.00
0-1
.96
Log
com
mer
cial
real
est
ate
0.00
01.
640.
001
5.09
0.00
01.
65Le
vera
ge ra
tio0.
001
2.30
0.00
0-0
.04
0.00
00.
22La
gged
Inco
me
Log
hum
an c
apita
l -0
.020
-14.
49-0
.031
-12.
31-0
.009
-11.
79Lo
g in
com
e-0
.002
-1.5
80.
008
3.21
0.00
00.
36Se
lf-em
ploy
men
t dum
my
-0.0
06-2
.79
-0.0
06-1
.51
-0.0
01-0
.69
Unem
ploy
men
t dum
my
-0.0
04-2
.28
-0.0
04-1
.27
0.00
00.
00C
ondi
tiona
l inc
ome
vola
tility
-0.0
51-1
1.21
-0.0
28-3
.61
-0.0
11-4
.92
Lagg
ed D
emog
raph
ic C
hara
cter
istic
Fam
ily s
ize
0.00
614
.40
0.00
11.
830.
002
9.10
yAd
just
ed R
26.
85%
0.07
05.
27%
0.05
47.
06%
0.07
1Nu
mbe
r of o
bser
vatio
ns40
6,56
140
6,56
122
1,14
322
1,14
335
5,44
335
5,44
3
Tabl
e IX
Valu
e Lo
adin
gs o
f New
Par
ticip
ants
This
tabl
ere
ports
pool
edre
gres
sion
sof
ane
wpa
rtici
pant
’sva
lue
load
ing
onso
cioe
cono
mic
char
acte
ristic
san
dye
ar,i
ndus
try,a
ndco
unty
fixed
effe
cts.
We
cond
uctt
hees
timat
ion
onho
useh
olds
that
ente
rth
est
ock
mar
ket
betw
een
1999
to20
07in
the
repr
esen
tativ
esa
mpl
ede
fined
inSe
ctio
n2
2Th
een
ter
the
stoc
km
arke
tbe
twee
n19
99to
2007
inth
ere
pres
enta
tive
sam
ple
defin
edin
Sect
ion
2.2.
The
anal
ysis
isba
sed
onda
tain
the
year
ofen
try.A
llva
riabl
esar
ede
scrib
edin
Tabl
eA.
(1)
(2)
(3)
Risk
y Po
rtfol
ioSt
ock
Portf
olio
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ng Fund
Por
tfolio
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tFi
nanc
ial C
hara
cter
istic
sLo
g fin
anci
al w
ealth
0.
015
2.25
0.10
66.
820.
014
3.78
Log
resi
dent
ial r
eal e
stat
e-0
.002
-1.4
6-0
.001
-0.4
3-0
.001
-0.9
9Lo
g co
mm
erci
al re
al e
stat
e0.
000
0.13
0.00
92.
270.
000
0.39
Leve
rage
ratio
0.01
54.
980.
037
3.79
0.00
0-0
.22
Leve
rage
ratio
0.01
54.
980.
037
3.79
0.00
00.
22H
uman
Cap
ital a
nd In
com
e R
isk
Log
hum
an c
apita
l -0
.036
-1.4
9-0
.059
-1.2
6-0
.005
-0.3
8Lo
g in
com
e-0
.048
-2.1
2-0
.041
-0.8
7-0
.018
-1.9
5Se
lf-em
ploy
men
t dum
my
-0.1
89-4
.51
-0.2
43-3
.43
0.00
10.
08Un
empl
oym
ent d
umm
y-0
.029
-1.4
2-0
.126
-2.2
00.
012
1.10
Con
ditio
nali
ncom
evo
latili
ty-0
328
-599
-025
3-2
21-0
065
-211
Con
ditio
nal i
ncom
e vo
latili
ty-0
.328
-5.9
9-0
.253
-2.2
1-0
.065
-2.1
1D
emog
raph
ic C
hara
cter
istic
sAg
e0.
001
2.08
0.00
74.
380.
000
1.08
Mal
e ho
useh
old
head
dum
my
-0.1
09-7
.88
-0.1
26-3
.57
-0.0
20-2
.37
High
sch
ool d
umm
y0.
005
0.30
0.02
10.
490.
007
0.74
Post
-hig
h sc
hool
dum
my
-0.0
58-3
.72
-0.0
89-2
.46
-0.0
12-1
.33
Econ
omic
sed
ucat
ion
dum
my
003
61
870
000
000
001
61
33Ec
onom
ics
educ
atio
n du
mm
y-0
.036
-1.8
70.
000
0.00
-0.0
16-1
.33
Imm
igra
tion
dum
my
-0.0
43-2
.31
-0.0
61-1
.37
0.01
71.
64Fa
mily
siz
e0.
030
5.07
0.01
00.
690.
006
1.88
Adju
sted
R2
2.06
%3.
73%
0.44
%Nu
mbe
r of o
bser
vatio
ns13
,927
4,77
910
,472
Tabl
e X
Exis
ting
vs. N
ew P
artic
ipan
ts
This
tabl
ere
ports
apo
oled
regr
essi
onof
the
valu
elo
adin
gon
(i)ag
edu
mm
ies
forp
artic
ipat
ing
hous
ehol
ds,
(ii)
age
dum
mie
sfo
rho
useh
olds
that
ente
rris
kyas
set
mar
kets
durin
gth
eye
ar,
(iii)
all
the
othe
rch
arac
teris
tics
ofth
eba
selin
ere
gres
sion
and
year
indu
stry
and
coun
tyfix
edef
fect
sAg
edu
mm
ies
are
char
acte
ristic
sof
the
base
line
regr
essi
on,
and
year
,in
dust
ry,
and
coun
tyfix
edef
fect
s.Ag
edu
mm
ies
are
cum
ulat
ive.
The
com
puta
tions
are
base
don
the
repr
esen
tativ
esa
mpl
eof
hous
ehol
dsov
erth
e19
99to
2007
perio
dde
fined
inSe
ctio
n2.
2.Al
lvar
iabl
esar
edi
scus
sed
inTa
ble
A.St
anda
rder
rors
are
clus
tere
dat
the
hous
ehol
dle
vel.
The
full
resu
ltsar
eav
aila
ble
inth
eIn
tern
etAp
pend
ix.
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ngRi
sky
Portf
olio
Stoc
kPo
rtfol
ioFu
ndPo
rtfol
io
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tA
ge D
umm
ies
30 a
nd a
bove
0.00
30.
450.
052
3.29
0.00
0-0
.07
35 a
nd a
bove
0.00
51.
230.
047
4.28
0.00
0-0
.05
40d
b0
010
287
003
94
140
005
211
Risk
y Po
rtfol
ioSt
ock
Portf
olio
Fund
Por
tfolio
(1)
(2)
(3)
40 a
nd a
bove
0.01
02.
870.
039
4.14
0.00
52.
1145
and
abo
ve0.
012
3.55
0.05
86.
620.
002
0.91
50 a
nd a
bove
0.01
54.
490.
040
5.17
0.00
00.
1555
and
abo
ve0.
019
5.98
0.03
75.
600.
002
1.07
60 a
nd a
bove
0.01
13.
490.
022
3.36
0.00
31.
5965
and
abo
ve0.
019
3.52
0.02
11.
920.
006
1.66
70 a
nd a
bove
0.05
85.
620.
139
6.47
0.02
33.
46N
ew E
ntra
nt D
umm
ies
New
entra
nt a
ged
30+
-0.0
91-5
.99
-0.2
52-6
.14
-0.0
07-0
.83
New
entra
nt a
ged
35+
-0.0
07-0
.35
-0.0
07-0
.12
-0.0
08-0
.62
New
entra
nt a
ged
40+
-0.0
32-1
.42
-0.0
90-1
.48
0.00
40.
29Ne
w en
trant
age
d 45
+-0
.007
-0.2
8-0
.025
-0.4
10.
000
-0.0
2Ne
w en
trant
age
d 50
+0.
001
0.07
0.03
70.
690.
005
0.40
New
entra
nt a
ged
55+
-0.0
20-0
.93
-0.0
29-0
.61
-0.0
09-0
.65
New
entra
nt a
ged
60+
-0.0
01-0
.04
-0.0
68-1
.23
-0.0
02-0
.12
New
entra
nt a
ged
65+
-0.1
19-2
.96
-0.0
28-0
.39
-0.0
10-0
.43
New
entra
nt a
ged
70+
-0.0
10-0
.45
-0.0
13-0
.26
0.03
22.
58
Tabl
e XI
Stoc
ks M
ost W
idel
y H
eld
by S
wed
ish
Hou
seho
lds
The
tabl
ere
ports
the
ten
stoc
ksth
atar
em
ost
wid
ely
held
bySw
edis
hho
useh
olds
atth
een
dof
2003
.St
ocks
are
sorte
dby
the
prop
ortio
nof
hous
ehol
dsth
atho
ldth
emdi
rect
ly(fi
rstc
olum
n).W
eal
sore
port
the
stoc
k’s
perc
enta
gein
aggr
egat
eho
useh
old
finan
cial
wea
lth(s
econ
dco
lum
n),
the
stoc
k’s
perc
enta
geof
the
tota
lmar
ketc
apita
lizat
ion
ofal
lfirm
slis
ted
onSw
edis
hex
chan
ges
(third
colu
mn)
the
stoc
k’s
perc
enta
geof
the
free
the
stoc
ks
perc
enta
geof
the
tota
lmar
ketc
apita
lizat
ion
ofal
lfirm
slis
ted
onSw
edis
hex
chan
ges
(third
colu
mn)
,the
stoc
ks
perc
enta
geof
the
free
float
-adj
uste
dm
arke
tca
pita
lizat
ion
ofal
lfir
ms
liste
don
Swed
ish
exch
ange
s(fo
urth
colu
mn)
,th
est
ock’
sva
lue
load
ing
(fifth
colu
mn)
,an
dth
epe
rcen
tile
ofth
est
ock’
sbo
ok-to
-mar
ket
ratio
(six
thco
lum
n).
The
anal
ysis
isco
nduc
ted
onth
ere
pres
enta
tive
pane
ldef
ined
inSe
ctio
n2.
2.Th
eag
greg
ate
hous
ehol
dfin
anci
alw
ealth
used
inth
ese
cond
colu
mn
isth
eto
tala
mou
ntof
wea
lthow
ned
byris
kyas
setm
arke
tpar
ticip
ants
.In
the
last
row
we
cons
ider
the
aggr
egat
epo
rtfol
ioof
the
top
ten
popu
lar
stoc
ks.T
heva
lue
load
ings
and
book
-to-m
arke
trat
iope
rcen
tiles
are
base
don
stoc
kav
erag
es,w
here
stoc
ksar
ew
eigh
ted
byth
eirs
hare
sof
aggr
egat
eho
useh
old
portf
olio
repo
rted
inth
ese
cond
colu
mn.
Eric
sson
60.4
6%21
.69%
7.49
%8.
70%
-1.2
225
.41%
Telia
46.5
0%4.
02%
6.48
%4.
16%
-1.0
044
.19%
Swed
bank
24.5
4%3.
76%
2.74
%2.
75%
0.11
46.8
5%SE
B23
.57%
5.52
%2.
69%
3.14
%0.
7456
.21%
% o
f Hou
seho
ld
Stoc
k W
ealth
% o
f Sto
ckho
lder
s O
wnin
g C
ompa
ny%
of S
wedi
sh
Stoc
kmar
ket
Valu
e Lo
adin
gB/
M Q
uant
ile%
of S
wedi
sh
Free
Flo
at
SEB
23.5
7%5.
52%
2.69
%3.
14%
0.74
56.2
1%Vo
lvo14
.58%
5.00
%3.
18%
3.36
%0.
4168
.94%
H&M
11.3
9%4.
75%
5.21
%3.
76%
-0.0
74.
29%
Bille
rud
10.7
8%1.
11%
0.22
%0.
25%
-0.0
646
.26%
Astra
Zene
ca9.
66%
5.38
%4.
81%
3.79
%0.
0968
.23%
Noki
a8.
71%
3.78
%23
.77%
31.1
4%-0
.08
14.6
9%In
vest
or8
61%
248
%1
95%
159
%0
2780
77%
Inve
stor
8.61
%2.
48%
1.95
%1.
59%
0.27
80.7
7%57
.49%
58.5
3%62
.64%
-0.4
139
.21%
Aggr
egat
e po
rtfol
io o
f pop
ular
sto
cks
Tabl
e XI
IPo
rtfo
lios
of P
opul
ar a
nd P
rofe
ssio
nally
Clo
se S
tock
s
This
tabl
ere
ports
pool
edre
gres
sion
sof
the
valu
elo
adin
gof
hous
ehol
dsu
bpor
tfolio
son
hous
ehol
dch
arac
teris
tics
inth
epr
esen
ceof
year
,in
dust
ry,
and
coun
tyfix
edef
fect
s.Ev
ery
subp
ortfo
lioin
the
tabl
eis
asu
bset
ofth
est
ock
portf
olio
.W
eco
nsid
erth
epo
rtfol
ioof
popu
lar
stoc
ksin
colu
mn
(1)
the
portf
olio
ofst
ocks
othe
rth
anpo
pula
rst
ocks
inco
lum
n(2
)th
eco
nsid
erth
epo
rtfol
ioof
popu
lar
stoc
ksin
colu
mn
(1),
the
portf
olio
ofst
ocks
othe
rth
anpo
pula
rst
ocks
inco
lum
n(2
),th
epo
rtfol
ioof
prof
essi
onal
lycl
ose
stoc
ksin
colu
mn
(3),
and
the
portf
olio
ofst
ocks
othe
rth
anpr
ofes
sion
ally
clos
est
ocks
inco
lum
n(4
).Th
eco
mpu
tatio
nsar
eba
sed
onth
ere
pres
enta
tive
pane
lofh
ouse
hold
sov
erth
e19
99to
2007
perio
dde
fined
inS
ectio
n2.
2.Al
lvar
iabl
esar
ede
scrib
edin
Tabl
eA.
Sta
ndar
der
rors
are
clus
tere
dat
the
hous
ehol
dle
vel.
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ng o
f Sto
ck S
ubpo
rtfol
ioPr
ofes
sC
lose
NotP
rofe
ssC
lose
Popu
lar
NotP
opul
ar
Estim
ate
t-sta
tEs
timat
et-s
tatE
stim
ate
t-sta
tEs
timat
et-s
tat
Fina
ncia
l Cha
ract
eris
tics
Log
finan
cial
wea
lth
0.01
98.
390.
144
22.0
10.
073
12.8
70.
053
16.4
8Lo
g re
side
ntia
l rea
l est
ate
0.00
24.
080.
004
2.45
0.00
31.
900.
004
5.01
Li
ll
tt
000
715
300
002
181
000
32
290
007
1170
(3)
(4)
(1)
(2)
Prof
ess.
Clo
seNo
t Pro
fess
. Clo
sePo
pula
rNo
t Pop
ular
Log
com
mer
cial
real
est
ate
0.00
715
.30
0.00
21.
810.
003
2.29
0.00
711
.70
Leve
rage
ratio
0.00
51.
79-0
.029
-3.0
1-0
.033
-3.7
0-0
.002
-0.4
9H
uman
cap
ital a
nd In
com
e R
isk
Log
hum
an c
apita
l -0
.085
-10.
53-0
.101
-4.3
8-0
.109
-5.2
0-0
.115
-9.4
7Lo
g in
com
e-0
.021
-3.8
6-0
.077
-4.8
9-0
.040
-2.6
1-0
.034
-3.8
5Se
lf-em
ploy
men
t dum
my
-0.0
20-1
.95
-0.1
39-4
.29
-0.0
98-3
.52
-0.0
23-1
.52
Unem
ploy
men
t dum
my
-0.0
09-1
.25
-0.0
69-2
.84
-0.0
20-0
.97
-0.0
17-1
.45
Con
ditio
nal i
ncom
e vo
latili
ty-0
.047
-2.2
1-0
.704
-11.
01-0
.335
-6.3
5-0
.312
-9.5
4D
emog
raph
ic C
hara
cter
istic
sAg
e0.
005
17.2
00.
016
19.2
10.
008
10.3
30.
008
17.8
0M
ale
hous
ehol
d he
ad d
umm
y-0
.061
-10.
21-0
.169
-9.7
4-0
.080
-5.4
9-0
.113
-13.
77Hi
gh s
choo
l dum
my
-0.0
29-3
.79
-0.0
32-1
.33
-0.0
35-1
.85
-0.0
25-2
.37
gy
Post
-hig
h sc
hool
dum
my
0.01
72.
810.
061
3.55
0.03
22.
170.
022
2.65
Econ
omic
s ed
ucat
ion
dum
my
-0.0
10-1
.28
-0.0
04-0
.18
0.02
81.
70-0
.021
-1.9
5Im
mig
ratio
n du
mm
y-0
.086
-9.2
0-0
.321
-10.
59-0
.148
-5.8
6-0
.126
-9.0
9Fa
mily
siz
e0.
017
7.14
0.02
13.
030.
015
2.54
0.02
26.
58Ad
just
edR
22.
95%
5.35
%3.
08%
3.33
%N u
mbe
r of o
bser
vatio
ns28
7,57
418
8,44
998
,916
288,
409
ube
oob
seat
os
8,5
88,
998
,96
88,
09
Tabl
e XI
IID
ivid
ends
, Tax
es a
nd F
irm A
ge
This
tabl
ere
ports
pool
edre
gres
sion
sof
the
valu
elo
adin
gon
char
acte
ristic
ses
timat
edon
hous
ehol
dpo
rtfol
ios
of(1
)st
ocks
payi
ngno
divi
dend
s,(2
)st
ocks
with
aver
age
annu
aldi
vide
ndyi
elds
abov
e2%
,(3)
com
pani
esth
atha
vebe
enlis
ted
for
nom
ore
than
10ye
ars,
(4)
com
pani
esth
atha
vebe
enlis
ted
fora
tlea
st20
year
s(5
)tax
adva
ntag
edO
stoc
ksan
d(6
)tax
able
Ast
ocks
The
regr
essi
ons
incl
ude
year
indu
stry
and
coun
tyfix
edef
fect
san
dlis
ted
fora
tlea
st20
year
s,(5
)tax
-adv
anta
ged
O-s
tock
s,an
d(6
)tax
able
A-st
ocks
.The
regr
essi
ons
incl
ude
year
,ind
ustry
and
coun
tyfix
edef
fect
s,an
dar
ees
timat
edon
the
repr
esen
tativ
epa
nelo
fhou
seho
lds
over
the
1999
to20
07pe
riod
defin
edin
Sec
tion
2.2.
Allv
aria
bles
are
desc
ribed
inTa
ble
A.
(3)
(4)
(5)
(6)
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ngNo
-Div.
Sto
cks
High
-Div.
Sto
cks
Youn
g St
ocks
Old
Sto
cks
O S
tock
sA
Stoc
ks(1
)(2
)Es
timat
et-s
tat
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tFi
nanc
ial C
hara
cter
istic
sLo
g fin
anci
al w
ealth
0.
133
13.3
70.
027
10.9
10.
146
28.0
00.
048
15.0
20.
155
20.8
70.
045
16.0
2Lo
g re
side
ntia
l rea
l est
ate
0.00
41.
450.
002
3.88
0.00
21.
940.
002
3.31
0.00
52.
520.
003
4.50
Log
com
mer
cial
real
est
ate
0.00
20.
930.
003
7.30
0.00
22.
350.
002
3.27
0.00
74.
840.
007
12.0
2Le
vera
ge ra
tio-0
.058
-0.0
42.
007
6.10
-4.1
94-5
.09
-1.6
43-4
.02
-0.0
01-4
.80
0.00
01.
38H
uman
Cap
ital a
nd In
com
e R
isk
Log
hum
an c
apita
l -0
.092
-2.5
9-0
.068
-8.0
0-0
.067
-3.8
1-0
.071
-6.4
5-0
.067
-2.5
9-0
.102
-10.
31Lo
g in
com
e-0
.095
-4.2
40.
005
0.76
-0.0
43-3
.41
-0.0
38-5
.03
-0.0
96-5
.14
-0.0
25-3
.62
Self-
empl
oym
ent d
umm
y-0
.048
-1.0
3-0
.019
-1.6
7-0
.023
-0.9
9-0
.059
-3.9
5-0
.095
-2.6
6-0
.047
-3.7
0Un
empl
oym
ent d
umm
y0.
031
0.86
-0.0
06-0
.65
0.00
30.
15-0
.031
-2.8
0-0
.033
-1.2
1-0
.019
-1.9
2C
ondi
tiona
l inc
ome
vola
tilit y
-0.3
05-3
.33
-0.0
02-0
.10
-0.4
36-8
.95
-0.2
25-7
.71
-0.6
35-9
.32
-0.1
99-7
.42
yD
emog
raph
ic C
hara
cter
istic
sAg
e0.
010
8.09
0.00
27.
250.
009
13.9
10.
007
19.2
90.
014
15.6
20.
007
21.0
3M
ale
hous
ehol
d he
ad d
umm
y-0
.133
-4.8
1-0
.044
-6.6
9-0
.106
-8.4
6-0
.078
-9.5
1-0
.180
-9.3
2-0
.086
-12.
07Hi
gh s
choo
l dum
my
-0.0
54-1
.37
-0.0
07-0
.88
-0.0
09-0
.50
-0.0
01-0
.06
-0.0
34-1
.27
-0.0
29-3
.18
Post
-hig
h sc
hool
dum
my
-0.0
67-2
.49
0.01
92.
770.
016
1.19
0.04
55.
510.
021
1.09
0.02
93.
98Ec
onom
ics
educ
atio
ndu
mm
y0.
001
0.02
-0.0
07-0
.81
0.00
70.
460.
017
1.66
0.03
51.
45-0
.009
-1.0
2Ec
onom
ics
educ
atio
n du
mm
y0.
001
0.02
0.00
70.
810.
007
0.46
0.01
71.
660.
035
1.45
0.00
91.
02Im
mig
ratio
n du
mm
y0.
127
2.98
-0.0
68-6
.03
0.00
00.
01-0
.109
-8.7
9-0
.355
-10.
10-0
.106
-9.5
3Fa
mily
siz
e0.
018
1.68
0.01
35.
030.
006
1.14
0.00
82.
52-0
.002
-0.3
10.
021
7.23
Adju
sted
R2
1.88
%1.
05%
4.37
%3.
91%
5.57
%3.
32%
Num
ber o
f obs
erva
tions
110,
369
273,
442
207,
099
221,
963
117,
823
315,
207
Tabl
e XI
VFi
nanc
ial M
arke
t Exp
erie
nce
This
tabl
ere
ports
pool
edre
gres
sion
sof
the
valu
elo
adin
gin
2007
on(i)
the
num
ber
ofye
ars
inth
epa
nelw
hen
the
hous
ehol
dpa
rtici
pate
sin
risky
asse
tmar
kets
,(ii)
the
earli
estv
alue
load
ing
inth
epa
nel,
and
(iii)
all
the
othe
rho
useh
old
char
acte
ristic
san
dye
arin
dust
ryan
dco
unty
fixed
effe
cts
The
and
(iii)
all
the
othe
rho
useh
old
char
acte
ristic
san
dye
ar,
indu
stry
,an
dco
unty
fixed
effe
cts.
The
com
puta
tions
are
base
don
the
repr
esen
tativ
epa
nel
ofho
useh
olds
over
the
1999
to20
07pe
riod
defin
edin
Sec
tion
2.2.
Allv
aria
bles
are
desc
ribed
inTa
ble
A.
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ng
(1)
(2)
(3)
Risk
y Po
rtfol
ioSt
ock
Portf
olio
Fund
Por
tfolio
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tIn
itial
val
ue lo
adin
g0.
351
35.7
70.
470
34.0
10.
126
29.0
9Ex
perie
nce
Num
ber o
f par
ticip
atio
n ye
ars
-0.0
06-4
.07
-0.0
15-2
.46
-0.0
11-1
1.42
Fina
ncia
l Cha
ract
eris
tics
(1)
(2)
(3)
Log
finan
cial
wea
lth0.
018
7.24
0.07
512
.57
0.00
21.
29Lo
g re
side
ntia
l rea
l est
ate
weal
th0.
000
-0.2
40.
003
2.07
-0.0
01-4
.79
Log
com
mer
cial
real
est
ate
weal
th0.
001
1.73
0.00
77.
110.
000
0.10
Leve
rage
ratio
0.00
41.
230.
010
0.93
0.00
00.
23H
uman
Cap
ital a
nd In
com
e R
isk
Log
hum
an c
apita
l-0
.079
-6.9
6-0
.140
-6.1
8-0
.021
-3.9
8Lo
g in
com
e-0
.004
-0.4
60.
018
1.00
-0.0
14-3
.40
Self-
empl
oym
ent d
umm
y-0
.052
-3.9
3-0
.064
-2.4
5-0
.007
-1.1
8Un
empl
oym
ent d
umm
y-0
.026
-2.2
7-0
.032
-1.1
3-0
.001
-0.1
6C
ondi
tiona
l inc
ome
vola
tility
-0.1
72-6
.09
-0.0
46-0
.83
-0.0
23-1
.75
Dem
ogra
phic
Cha
ract
eris
tics
Age
0.00
11.
490.
005
5.90
0.00
14.
13g
000
90
005
590
000
3M
ale
hous
ehol
d he
ad d
umm
y-0
.037
-7.3
6-0
.055
-4.4
5-0
.003
-0.9
1Hi
gh s
choo
l dum
my
-0.0
22-3
.50
-0.0
77-4
.63
-0.0
03-0
.78
Post
-hig
h sc
hool
dum
my
-0.0
10-2
.02
0.03
73.
00-0
.015
-5.2
7Ec
onom
ics
educ
atio
n du
mm
y-0
.036
-5.0
4-0
.009
-0.5
5-0
.019
-4.8
0Im
mig
ratio
n du
mm
y-0
.073
-7.6
3-0
.133
-6.1
70.
007
1.55
Fam
ilysi
ze0
029
1172
000
50
890
012
917
Fam
ily s
ize
0.02
911
.72
0.00
50.
890.
012
9.17
Adju
sted
R2
15.1
5%13
.25%
6.12
%Nu
mbe
r of o
bser
vatio
ns50
,818
27,7
0145
,257
Tabl
e XV
Twin
s
This
tabl
ere
ports
pool
edre
gres
sion
sof
the
valu
elo
adin
gon
hous
ehol
dch
arac
teris
tics
and
year
lytw
in-p
air
fixed
effe
cts
estim
ated
onth
e19
99to
2007
pane
lofp
artic
ipat
ing
hous
ehol
dsw
ithan
adul
ttw
in.T
heva
lue
load
ing
isco
mpu
ted
atth
ele
velo
f(1)
the
risky
portf
olio
(2)t
hest
ock
portf
olio
and
(3)t
hefu
ndpo
rtfol
ioA
lllo
adin
gis
com
pute
dat
the
leve
lof(
1)th
eris
kypo
rtfol
io,(
2)th
est
ock
portf
olio
,and
(3)t
hefu
ndpo
rtfol
io.A
llva
riabl
esar
ede
scrib
edin
Tabl
eA.
Stan
dard
erro
rsar
ecl
uste
red
atth
eho
useh
old
leve
l.
Year
ly T
win
Pair
Fixe
d Ef
fect
s
Risk
y Po
rtfol
ioSt
ock
Portf
olio
Fund
Por
tfolio
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ng
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tFi
nanc
ial C
hara
cter
istic
sLo
g fin
anci
al w
ealth
0.00
92.
000.
030
2.79
0.01
03.
32Lo
g re
side
ntia
l rea
l est
ate
weal
th0.
002
1.59
0.00
51.
580.
000
0.14
Log
com
mer
cial
real
esta
tewe
alth
000
10
720
006
269
000
00
31
y (1)
(2)
(3)
Log
com
mer
cial
real
est
ate
weal
th0.
001
0.72
0.00
62.
690.
000
0.31
Leve
rage
ratio
-0.0
01-0
.18
0.03
31.
550.
000
0.08
Hum
an C
apita
l and
Inco
me
risk
Log
hum
an c
apita
l-0
.070
-4.0
0-0
.083
-1.9
3-0
.025
-1.8
7Lo
g in
com
e-0
.060
-4.5
3-0
.087
-3.0
2-0
.023
-2.4
1Se
lf-em
ploy
men
t dum
my
-0.0
01-0
.05
0.02
20.
55-0
.019
-1.4
4Un
empl
oym
entd
umm
y0
038
271
001
70
470
021
225
Unem
ploy
men
t dum
my
-0.0
38-2
.71
-0.0
17-0
.47
-0.0
21-2
.25
Con
ditio
nal i
ncom
e vo
latili
ty-0
.365
-7.7
5-0
.459
-3.8
1-0
.126
-3.6
9D
emog
raph
ic C
hara
cter
istic
sM
ale
hous
ehol
d he
ad d
umm
y-0
.027
-2.6
9-0
.025
-0.9
7-0
.019
-2.8
8Hi
gh s
choo
l dum
my
-0.0
35-2
.27
-0.0
71-1
.83
-0.0
05-0
.50
Post
-hig
h sc
hool
dum
my
0.00
40.
300.
040
1.33
-0.0
12-1
.47
Ei
dti
d0
010
073
001
90
550
008
075
Econ
omic
s ed
ucat
ion
dum
my
-0.0
10-0
.73
0.01
90.
55-0
.008
-0.7
5Fa
mily
siz
e0.
040
8.26
0.03
52.
760.
015
4.75
Adju
sted
R2
12.5
6%22
.79%
10.7
1%Nu
mbe
r of o
bser
vatio
ns10
4,52
243
,906
87,9
72
Tabl
e XV
IC
omm
unic
atio
n
This
tabl
ere
ports
pool
edre
gres
sion
sof
the
valu
elo
adin
gon
year
fixed
effe
cts
and
char
acte
ristic
s,es
timat
edon
(a)
hous
ehol
dsw
ithtw
ins
com
mun
icat
ing
frequ
ently
with
each
othe
r(“H
igh
Com
mun
icat
ion”
),an
d(b
)hou
seho
lds
with
twin
sco
mm
unic
atin
gin
frequ
ently
with
each
othe
r(“L
owC
omm
unic
atio
n”).
The
valu
elo
adin
gis
com
pute
dat
the
leve
loft
heris
kypo
rtfol
ioin
colu
mns
(1)a
nd(2
)th
est
ock
portf
olio
inco
lum
ns(3
)and
(4)
and
the
fund
portf
olio
inco
lum
nsva
lue
load
ing
isco
mpu
ted
atth
ele
velo
fthe
risky
portf
olio
inco
lum
ns(1
)and
(2),
the
stoc
kpo
rtfol
ioin
colu
mns
(3)a
nd(4
),an
dth
efu
ndpo
rtfol
ioin
colu
mns
(5)a
nd(6
).A
twin
pair
iscl
assi
fied
as“H
igh
Com
mun
icat
ion”
ifth
efre
quen
cyof
med
iate
dco
mm
unic
atio
nan
dth
efre
quen
cyof
unm
edia
ted
com
mun
icat
ion
are
both
abov
eth
em
edia
n,an
das
“Low
Com
mun
icat
ion”
othe
rwis
e.Th
eco
mm
unic
atio
nsu
bsam
ples
are
obta
ined
from
the
1999
to20
07pa
nel
ofpa
rtici
patin
gho
useh
olds
with
anad
ultt
win
.All
varia
bles
are
desc
ribed
inTa
ble
A.St
anda
rder
rors
are
clus
tere
dat
the
hous
ehol
dle
vel.
Risk
yPo
rtfol
ioSt
ock
Portf
olio
Fund
Portf
olio
Dep
ende
nt V
aria
ble:
Val
ue L
oadi
ng
Com
mun
icat
ion
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tEs
timat
et-s
tat
Estim
ate
t-sta
tEs
timat
et-s
tat
Fina
ncia
l Cha
ract
eris
tics
(1)
Risk
y Po
rtfol
ioSt
ock
Portf
olio
Fund
Por
tfolio
High
Low
High
Low
High
Low
(2)
(3)
(4)
(5)
(6)
Com
mun
icat
ion
Com
mun
icat
ion
Com
mun
icat
ion
Com
mun
icat
ion
Com
mun
icat
ion
Log
finan
cial
wea
lth0.
012
2.32
0.01
02.
150.
039
2.81
0.03
82.
970.
014
4.39
0.01
02.
94Lo
g re
side
ntia
l rea
l est
ate
weal
th0.
000
-0.0
10.
002
1.46
0.00
0-0
.01
0.01
02.
71-0
.001
-0.6
5-0
.001
-1.8
0Lo
g co
mm
erci
al re
al e
stat
e we
alth
0.00
33.
110.
002
1.83
0.01
35.
550.
009
3.88
0.00
10.
840.
001
0.71
Leve
rage
ratio
0.00
20.
25-0
.003
-0.3
7-0
.002
-0.1
10.
018
0.89
0.00
51.
11-0
.005
-1.0
8H
uman
Cap
ital a
nd In
com
e R
isk
Log
hum
an c
apita
l-0
.097
-4.5
6-0
.094
-4.7
9-0
.186
-3.7
2-0
.167
-3.3
9-0
.037
-2.4
9-0
.035
-2.3
8g
pLo
g in
com
e-0
.051
-3.2
0-0
.047
-3.1
2-0
.064
-1.8
8-0
.053
-1.5
1-0
.033
-3.1
5-0
.033
-2.9
1Se
lf-em
ploy
men
t dum
my
-0.0
16-0
.58
-0.0
11-0
.49
-0.0
06-0
.09
0.00
90.
18-0
.033
-1.8
60.
000
0.02
Unem
ploy
men
t dum
my
-0.0
33-1
.97
-0.0
35-2
.11
-0.0
26-0
.56
-0.0
53-1
.00
-0.0
23-2
.12
-0.0
13-1
.25
Con
ditio
nal i
ncom
e vo
latili
ty-0
.423
-7.0
5-0
.375
-6.4
6-0
.385
-2.8
6-0
.435
-3.4
1-0
.161
-4.0
3-0
.163
-4.5
2D
emog
raph
ic C
hara
cter
istic
sAg
e0.
002
2.58
0.00
21.
640.
009
3.37
0.00
62.
630.
000
0.36
0.00
00.
49Ag
e0.
002
2.58
0.00
21.
640.
009
3.37
0.00
62.
630.
000
0.36
0.00
00.
49M
ale
hous
ehol
d he
ad d
umm
y-0
.026
-2.2
1-0
.020
-1.9
6-0
.018
-0.5
6-0
.054
-2.0
70.
007
1.01
-0.0
11-1
.62
High
sch
ool d
umm
y-0
.025
-1.6
9-0
.001
-0.0
70.
005
0.12
0.02
60.
63-0
.007
-0.7
00.
004
0.34
Post
-hig
h sc
hool
dum
my
-0.0
04-0
.38
0.00
50.
450.
041
1.11
0.09
93.
42-0
.022
-2.5
9-0
.018
-2.4
2Ec
onom
ics
educ
atio
n du
mm
y-0
.047
-2.8
3-0
.019
-1.1
1-0
.037
-0.8
30.
029
0.77
-0.0
34-2
.62
-0.0
11-1
.03
Fam
ily s
ize
0.04
68.
590.
038
7.27
0.05
23.
360.
055
3.91
0.02
16.
140.
017
4.89
Adju
sted
R2
315
%2
13%
564
%4
08%
228
%1
64%
Adju
sted
R3.
15%
2.13
%5.
64%
4.08
%2.
28%
1.64
%Nu
mbe
r of o
bser
vatio
ns36
,230
42,5
8815
,462
17,4
4830
,572
36,0
08
Tabl
e XV
IIA
CE
Dec
ompo
sitio
n
This
tabl
ere
ports
anAC
Ede
com
posi
tion
ofth
eva
lue
load
ing
ofho
useh
olds
with
twin
sov
erth
e19
99to
2007
perio
d.W
ere
port
the
resu
ltsfo
r(i)
allt
win
s,(ii
)tw
ins
who
com
mun
icat
efre
quen
tlyw
ithea
chot
her
(“Hig
hC
omm
unic
atio
n”)
and
(iii)
twin
sw
hoco
mm
unic
ate
infre
quen
tlyw
ithea
chot
her
each
othe
r(H
igh
Com
mun
icat
ion
),an
d(ii
i)tw
ins
who
com
mun
icat
ein
frequ
ently
with
each
othe
r(“
Low
Com
mun
icat
ion”
).Th
ese
tofc
olum
nsla
bele
d“N
oC
ontro
ls”p
rese
nts
the
ACE
deco
mpo
sitio
nfo
rth
eva
lue
load
ing
itsel
f.Th
ese
tof
colu
mns
labe
led
“With
Con
trols
”pr
esen
tsth
eAC
Ede
com
posi
tion
fort
here
sidu
alof
the
valu
elo
adin
g,ob
tain
edfro
ma
regr
essi
onof
the
load
ing
onth
est
anda
rdch
arac
teris
tics
and
year
fixed
effe
cts.
Pane
lAco
nduc
tsth
ean
alys
isat
the
leve
loft
heris
kypo
rtfol
io,P
anel
Bat
the
leve
loft
hest
ock
portf
olio
,and
Pane
lCat
the
leve
loft
hefu
ndpo
rtfol
io.A
twin
pair
iscl
assi
fied
as“H
igh
Com
mun
icat
ion”
ifth
efre
quen
cyof
med
iate
dco
mm
unic
atio
nan
dth
ep
gq
yfre
quen
cyof
unm
edia
ted
com
mun
icat
ion
are
both
abov
eth
em
edia
n,an
das
“Low
Com
mun
icat
ion”
othe
rwis
e.
Gen
etic
Com
mon
Gen
etic
Com
mon
Pane
l A: V
alue
Loa
ding
of R
isky
Por
tfolio
No C
ontro
lsW
ith C
ontro
lsG
enet
icC
omm
onG
enet
icC
omm
onC
ompo
nent
Com
pone
ntC
ompo
nent
Com
pone
ntAl
l twi
n pa
irs17
.0%
0.1%
16.1
%-0
.6%
High
-com
mun
icat
ion
pairs
35.3
9%-6
.13%
33.0
5%-6
.27%
Low-
com
mun
icat
ion
pairs
0.87
%4.
74%
-0.2
3%4.
87%
Pane
l B: V
alue
Loa
ding
of S
tock
Por
tfolio
Gen
etic
Com
mon
Gen
etic
Com
mon
Com
pone
ntC
ompo
nent
Com
pone
ntC
ompo
nent
All t
win
pairs
12.3
%13
.7%
10.8
%11
.4%
High
-com
mun
icat
ion
pairs
37.5
8%7.
55%
37.6
4%3.
74%
Low-
com
mun
icat
ion
pairs
-0.2
4%11
.65%
-3.9
1%11
.15%
No C
ontro
lsW
ith C
ontro
ls
p
Gen
etic
Com
mon
Gen
etic
Com
mon
Com
pone
ntC
ompo
nent
Com
pone
ntC
ompo
nent
All t
win
pairs
8.7%
4.4%
7.3%
3.8%
Hih
ii
i33
80%
1021
%32
24%
10%
Pane
l C: V
alue
Loa
ding
of F
und
Portf
olio
No C
ontro
lsW
ith C
ontro
ls
High
-com
mun
icat
ion
pairs
33.8
0%-1
0.21
%32
.24%
-10.
75%
Low-
com
mun
icat
ion
pairs
-8.2
5%12
.74%
-9.5
0%12
.69%
Tabl
e A
Def
initi
on o
f Hou
seho
ld V
aria
bles
This
tabl
esu
mm
ariz
esth
em
ain
hous
ehol
dva
riabl
esus
edin
the
pape
r.
Vi
blD
iti
Varia
ble
Des
crip
tion
Cas
hB
ank
acco
unt b
alan
ces
and
Sw
edis
h m
oney
mar
ket f
unds
.Fu
nd p
ortfo
lioP
ortfo
lio o
f mut
ual f
unds
oth
er th
an S
wed
ish
mon
ey m
arke
t fun
ds.
Sto
ck p
ortfo
lioP
ortfo
lio o
f dire
ctly
hel
d st
ocks
.R
isky
por
tfolio
Com
bina
tion
of th
e st
ock
and
fund
por
tfolio
s.R
isky
sha
reP
ropo
rtion
of r
isky
ass
ets
in th
e po
rtfol
io o
f cas
h an
d ris
ky fi
nanc
ial a
sset
s.y
py
py
Fina
ncia
l wea
lthVa
lue
ofho
ldin
gsin
cash
,ris
kyfin
anci
alas
sets
,cap
itali
nsur
ance
prod
ucts
,der
ivat
ives
,an
ddi
rect
ly h
eld
bond
s, e
xclu
ding
def
ined
-con
tribu
tion
retir
emen
t acc
ount
s.S
hare
of p
opul
ar s
tock
sFr
actio
n of
the
stoc
k po
rtfol
io in
vest
ed in
pub
lic fi
rms
whi
ch w
ere
one
of th
e te
n m
ost w
idel
y he
ld in
at l
east
one
yea
r bet
wee
n 19
99 a
nd 2
007.
Sha
re o
f pro
fess
iona
lly c
lose
sto
cks
Frac
tion
of th
e st
ock
portf
olio
inve
sted
in fi
rms
with
the
sam
e 1-
digi
t ind
ustry
cod
e as
an
adul
t ho
useh
old
mem
ber's
cur
rent
em
ploy
er.
py
Num
ber o
f sto
cks
Num
ber o
f ass
ets
in th
e st
ock
portf
olio
.N
umbe
r of f
unds
Num
ber o
f ass
ets
in th
e fu
nd p
ortfo
lio.
Res
iden
tial r
eal e
stat
e w
ealth
Valu
e of
prim
ary
and
seco
ndar
y re
side
nces
.C
omm
erci
al re
al e
stat
e w
ealth
Valu
e of
rent
al, i
ndus
trial
, and
agr
icul
tura
l pro
perty
.Le
vera
ge ra
tioTo
tal d
ebt d
ivid
ed b
y th
e su
m o
f fin
anci
al a
nd re
al e
stat
e w
ealth
. H
uman
capi
tal
Exp
ecte
dpr
esen
tval
ueof
futu
reno
nfin
anci
aldi
spos
able
real
inco
me
Hum
an c
apita
lE
xpec
ted
pres
ent v
alue
of f
utur
e no
n-fin
anci
al d
ispo
sabl
e re
al in
com
e.In
com
eTo
tal h
ouse
hold
dis
posa
ble
inco
me.
Sel
f-em
ploy
men
t dum
my
Dum
my
varia
ble
equa
l to
one
if th
e ho
useh
old
head
is s
elf-e
mpl
oyed
.U
nem
ploy
men
t dum
my
Dum
my
varia
ble
equa
l to
one
if th
e ho
useh
old
head
is u
nem
ploy
ed.
Con
ditio
nal in
com
e vo
latil
ityS
tand
ard
devi
atio
nof
the
tota
lin
com
esh
ock,
defin
edas
the
sum
ofth
epe
rsis
tent
and
trans
itory
inco
me
shoc
ks in
a g
iven
yea
r.Ag
eAg
e of
the
hous
ehol
d he
ad.
Mal
e ho
useh
old
head
dum
my
Dum
my
varia
ble
equa
l to
one
if th
e ho
useh
old
head
is m
ale.
Hig
h sc
hool
dum
my
Dum
my
varia
ble
equa
l to
one
if th
e ho
useh
old
head
has
a h
igh
scho
ol d
egre
e.P
ost-h
igh
scho
ol d
umm
yD
umm
y va
riabl
e eq
ual t
o on
e if
the
hous
ehol
d he
ad h
as h
ad s
ome
post
-hig
h sc
hool
edu
catio
n.E
cono
mic
s ed
ucat
ion
dum
my
Dum
my
varia
ble
equa
lto
one
ifth
eho
useh
old
head
rece
ived
educ
atio
nin
anfie
ldre
late
dto
econ
omic
s an
d m
anag
emen
t.Fa
mily
siz
eN
umbe
r of p
eopl
e liv
ing
in th
e ho
useh
old.
Figu
re 1
Perc
enta
ge o
f Pub
lic E
quity
Dire
ctly
Hel
d by
Hou
seho
lds
This
figur
eillu
stra
tes
(i)th
epe
rcen
tage
offir
mm
arke
tcap
italiz
atio
nsow
ned
dire
ctly
bySw
edis
hho
useh
olds
atth
een
dof
2003
asfu
nctio
nof
firm
size
(sol
idba
rsan
dle
ftax
is),
and
(ii)t
hedi
strib
utio
nof
firm
size
(sol
idlin
ean
drig
htax
is).
The
calc
ulat
ions
are
base
don
allt
he35
2fir
ms
liste
don
Sw
edis
hex
chan
ges
(SS
EA
TON
GM
INO
)and
allS
wed
ish
hous
ehol
dsth
atow
nst
ocks
atth
een
dof
2003
30%
60%
exch
ange
s(S
SE
,ATO
,NG
M,I
NO
)and
allS
wed
ish
hous
ehol
dsth
atow
nst
ocks
atth
een
dof
2003
.
20%
25%
30%
40%
50%
60%
HouseholdsDistribu
10%
15%
20%
30%
quityHeldby Htion of firms (
0%5%
0%10%
0 -1
1 -5
5 -1
010
-50
50 -
100
100
-500
500
-1K
1K -
5K5K
-10
K10
K+
% of Eq(%)
Firm
Mar
ketC
apita
lizat
ion
($ M
illio
n)
Figu
re 2
The
Valu
e La
dder
This
figur
eillu
stra
tes
the
valu
elo
adin
gof
the
stoc
kpo
rtfol
iofo
rdiff
eren
tcoh
orts
ofho
useh
olds
.Eac
hso
lidlin
eco
rresp
onds
toa
give
nco
hort,
defin
edas
a5-
year
age
bin.
The
first
coho
rtco
ntai
nsho
useh
olds
with
ahe
adag
edbe
twee
n30
and
34in
1999
,whi
leth
eol
dest
hh
hd
db
70d
74i
1999
Thl
dif
llh
hld
it
dd
lfh
ih
coho
rtha
sa
head
aged
betw
een
70an
d74
in19
99.T
helo
adin
gsof
allh
ouse
hold
sin
year
tare
dem
eane
dto
cont
rolf
orch
ange
sin
the
com
posi
tion
ofth
eSw
edis
hst
ock
mar
ket.
Aco
hort’
slo
adin
gin
year
tis
the
wea
lth-w
eigh
ted
aver
age
year
-tlo
adin
gof
hous
ehol
dsin
the
coho
rt.Th
efig
ure
isba
sed
onth
epa
nelo
fall
Sw
edis
hdi
rect
stoc
khol
ders
over
the
1999
to20
07pe
riod.
04
02
0.3
0.4
oading
0
0.1
0.2
Value Lo
02
-0.10
04
-0.3
- 0.2
Age
-0.4
3035
4045
5055
6065
7075
Recommended