Home Bias

Pacific-Basin Finance Journal 62 (2020) 101375
Contents lists available at ScienceDirect
Pacific-Basin Finance Journal

journal homepage: www.elsevier.com/locate/pacfin
The potential effect of taxes on the equity home bias in New

T
Zealand PIEs
Shaun McDowella, John B. Leeb,⁎, Alastair Marsdenc
a
School of Accounting, Finance, and Economics, The University of Waikato, Hamilton, New Zealand
b
Department of Accounting and Finance, The University of Auckland, Auckland, New Zealand
c
Department of Accounting and Finance, The University of Auckland, Auckland, New Zealand
ARTICLE INFO ABSTRACT
JEL classifications: This paper examines the effect of taxes on the international diversification benefits available to
G15 New Zealand investors who invest in equities held through a portfolio investment entity (PIE).
G11 Historical monthly index data for 34 markets denominated in New Zealand dollars from 1993 to
2018 is used with an in-sample mean-variance optimization approach to estimate the optimal
Keywords: domestic asset allocation and potential benefits of international diversification. Our findings
International financial markets
suggest that New Zealand's dividend imputation tax regime and differential taxation rates on
Portfolio choice
domestic and offshore investment can account for a 7%–11% weight shift towards the New
Home bias
Zealand domestic market compared to the case of no taxes. Monte-Carlo simulation analysis
using different return patterns also provide support for home bias in the presence of higher taxes
on offshore equity investment. Overall, our results provide a partial explanation for current in-
vestor home bias in New Zealand. In contrast, without dividend imputation and differential rates,
taxes would suggest a weight shift to offshore equities of between 1% and 6% due to high di-
vidend yields in the New Zealand market.
1. Introduction
This paper investigates the extent to which taxation may explain investor home bias in the New Zealand context. Investor home
bias refers to the propensity of investors to dis-proportionately weight their investment portfolio to stocks or bonds in their home
jurisdiction. Investor home bias still remains a puzzle where the barriers to investors holding a portfolio of well diversified inter-
national equities have substantially reduced. This follows greater integration of the world's financial markets since the beginning of
the 1990s (Kho et al., 2009) and reduced information and trading costs with the advent of the internet and electronic trading
platforms. However, despite the theoretical benefits of international diversification (Sercu, 1980; Solnik, 1974; French and Poterba,
1991),1 investors have continued to exhibit a substantial degree of home bias. For example, Levy and Levy (2014) report that the bias
by United States (US) investors to overweight investment in domestic stocks still persists and has remained remarkably steady since
1998 at around 40% over the last 15 years.
Corresponding author.
⁎
E-mail addresses: shaun.mcdowell@waikato.ac.nz (S. McDowell), byong.lee@auckland.ac.nz (J.B. Lee),

a.marsden@auckland.ac.nz (A. Marsden).
1
The international capital asset pricing model (CAPM) (Sercu, 1980; Solnik, 1974), which builds on the assumption of risk-sharing in the CAPM
(Lintner, 1965; Sharpe, 1964), predicts mean-variance optimizing investors can reduce exposure to country specific risk by forming a diversified
portfolio of stocks weighted according to the market capitalization of each country.
https://doi.org/10.1016/j.pacfin.2020.101375
Received 6 September 2019; Received in revised form 24 January 2020; Accepted 5 June 2020
Available online 20 June 2020
0927-538X/ © 2020 Elsevier B.V. All rights reserved.
S. McDowell, et al. Pacific-Basin Finance Journal 62 (2020) 101375
The reasons advanced to explain investor home bias include higher costs of investing in a foreign jurisdiction, greater relative
price risks, hedging exchange rate and inflation risks, cultural and language barriers, governance issues and information asymmetry
(see for example, Errunza and Losq, 1985; Cooper and Kaplanis, 1994; Lewis, 1999; and Karolyi and Stulz, 2003). Behavioural
explanations include investors' overly optimistic forecasts of stocks in their home country (Gehrig, 1993), irrational behaviour (Sercu
and Vanpee, 2007), and unfamiliarity in distant markets (Ardalan, 2019).
The explanations for the high level of home bias exhibited by most investors across different global markets still remain a subject
of much debate. Baxter and Jermann (1997) suggest investors have incentives to short domestic stocks to hedge their human capital
risk. Using data on real exchange rates and asset returns for the US compared to the rest of the world, Pesenti and Van Wincoop
(2002) report that portfolio home bias related to trade costs is close to zero. Levy and Levy (2014) argue, however, that greater global
market integration and the resulting increase in average correlations between markets offset these gains and the home bias will
persist over time. Ardalan (2019) concludes that no single explanation, by itself, can explain the full extent of home equity bias.
This paper fills a gap in the literature by empirically examining the extent to which taxation may contribute to home bias in a
global index comprising 34 markets. We examine this question in the context of the New Zealand market. The specific focus of our
study is the effect of taxes on the potential benefits from international diversification from investments held indirectly in a portfolio
investment entity (PIE). PIEs are commonly used as an investment vehicle in KiwiSaver retirement accounts and by a number of New
Zealand institutional portfolio managers, who manage individual investment portfolios. The New Zealand market is an interesting
setting for the following reasons. First, the New Zealand market is a small capital market whose investors would benefit more from
international diversification. Second, for New Zealand residents, significant differences exist between taxes on dividends and capital
gains sourced in New Zealand versus those on overseas income. New Zealand tax-resident investors benefit from imputation credits
attached to New Zealand company dividends, which reduces the tax liability on dividends for most resident investors.2 Capital gains
on New Zealand shares held by a New Zealand investor are generally not taxed. In contrast, dividends paid to New Zealand investors
on shares held overseas are often taxed at source or subject to non-resident withholding tax (NRWT) by foreign governments. In
general, the higher taxation of dividends and capital gains received by New Zealand investors on overseas income can reduce the
realized after-tax returns on international equity investments compared to investments in New Zealand equities.
To see if the observed effects of taxes are specific to our sample, we undertake a simulation analysis with the same pre-tax
expected returns and standard deviations for all markets. To further examine the impact of taxation on home bias, we also consider
the hypothetical cases where New Zealand residents were subject to taxation under the rules prior to introduction of dividend
imputation and changes to taxation of offshore income. When both domestic and offshore dividends are equally taxed and there is no
capital gains tax, the impact of taxation on home bias is far less clear. This is because differences in dividend yields between markets
matter, together with the impact of market frictions from NRWT where New Zealand investors in lower tax brackets are unable to
fully claim back the tax credits on NRWT deducted. We also examine the sensitivity of our results to an international portfolio
restricted to the New Zealand market and the major world markets of the US, Japan and the United Kingdom (UK).
The contribution of our paper is as follows. First, we showcase the importance of differential taxation in the context of inter-
national diversification utilizing the unique tax environment of New Zealand. Our findings will be of direct interest to investors in the
other markets with a full or partial imputation tax regime (for example, Australia, Canada and Malta) as well as New Zealand
domestic investors.3 Similarly, our results will be of interest to domestic investors in countries with differential taxation rules be-
tween domestic and offshore investment. An example is countries that operate a single-tier tax system where dividends paid, credited
or distributed by a resident company are exempt from tax in the hands of the shareholders (e.g., Singapore and Malaysia).4 Second,
we also consider the extent of any home bias tilt under the New Zealand taxation regime prior to the introduction of dividend
imputation and the FIF regime. These results will be of general interest to investors in other markets, where investors may face a tax
regime that treat domestic and foreign investments similarly. Even in that case, we show that investors in different tax brackets would
rationally employ different diversification strategies due to a wedge between the tax rates on dividends and capital gains. Third, we
examine home bias from both the perspective of investors who wish to maximize the return-to-risk ratio and for domestic investors
who wish to minimize the portfolio variance only. Fourth, we consider the extent to which taxation may contribute to home bias from
the viewpoint of investors in a small market such as New Zealand which is less than 1% of the world's total market capitalization. The
New Zealand market lacks exposure to large multinational companies across a range of different industries5 and the theoretical
benefits of international diversification to New Zealand are therefore be expected to be high. Prior studies (e.g., Levy and Levy, 2014;
O'Hagan-Luff and Berrill, 2015) examine the benefits of international diversification for investors domiciled in larger capital markets.
Lastly, our findings will be of interest to the New Zealand Exchange, financial advisors and managers of investment funds, who are
agents for investors, capital market participants and regulators such as the Financial Markets Authority. Ongoing allocation by
KiwiSaver funds to domestic investment may increase the depth of the New Zealand market and encourage more companies to list on
2
The New Zealand dividend imputation regime was introduced in 1988.
3
For an introduction to Australia's franking or dividend imputation regime see Officer (1994). In the case of Canada, EY (2019) state that
“Dividends paid by a Canadian corporation to a Canadian resident individual are generally taxable, but the individual also receives a tax credit because the
income has already been taxed within the corporation”. Ainsworth (2016) notes that Canada, Chile, Mexico and New Zealand, Australia were five
countries in the Organisation for Economic Co-operation and Development (OECD) that have continued to operate an imputation tax system.
4
See EY (2019).
5
Over the time period of this study the New Zealand market was dominated by exposure to agricultural or commodity products, tourism and
utility-like companies. It lacked exposure to industry sectors such as large technology companies, pharmaceutical and health-care companies.
2
the New Zealand Exchange. An investigation of factors that may explain home bias is important to help understand investor portfolio
positions and cross-border capital flows.
We summarise our findings as follows. Using historical monthly equity index data for 34 markets denominated in New Zealand
dollars from 1993 to 2018 with an in-sample mean-variance optimization approach to measure the benefits of international di-
versification, we find that the differential (higher) taxation of overseas equity investments renders domestic equity investments more
attractive on an after-tax return basis. For New Zealand domestic investors who wish to maximize the return-to-risk ratio, we find that
the differential taxation of overseas investments reduces the potential benefits of international diversification for mean-variance
optimizing investors, which calls for an increase in the weight invested in the domestic market of between 7% and 11% compared to
the weight allocation to the New Zealand market when taxes are ignored. However, under the New Zealand tax regime prior to the
introduction dividend imputation and changes to the taxation of offshore investment, we find that this tax regime potentially fa-
voured an increased weight to offshore equities of between circa 1% and 6%. This reflects the New Zealand market being a high
“dividend yield” market relative to offshore markets and where, prior to the introduction of the imputation tax regime, domestic and
offshore dividends were generally subject to the same marginal tax rates with no capital gains tax on either domestic and offshore
investments. In comparison, domestic investors who wish to minimize the portfolio variance only are generally not affected by taxes.
Our simulation results also support the presence of home bias, with an increased allocation to New Zealand domestic equities of
between roughly 8% and 11% for investors on a 28% PIE tax rate, when equity market return correlations equal 0.25. Higher return
correlations of 0.75 result in greater tilts towards domestic equities due to differential taxes. The simulation results support the
argument of Levy and Levy (2014) that strengthening correlations between markets magnify home-advantages such as higher ex-
pected domestic returns which can rationalize a greater domestic tilt.
The rest of this paper is structured as follows. Section 2 reviews the literature related to New Zealand KiwiSaver investments and
prior literature on the impact of taxes on the equity home bias. Section 3 outlines the dividend and capital gains tax regime for equity
investments held in PIEs and the assumptions adopted in this study. Section 4 describes the data and methods to investigate the
benefits of international diversification. Section 5 reports the results of this investigation. We conduct a simulation analysis with a set
of expected returns and standard deviations different to historical market returns in Section 6. Section 7 examines the optimal New
Zealand or domestic portfolio weights and allocation decisions under the tax regime prior to the introduction of dividend imputation
and provides robustness tests where the international market is restricted to the US, Japan and UK markets only. Section 8 considers
the possible impact of transaction costs on our results. Section 9 concludes with a summary of the main findings and suggestions for
future study.
2. Literature review
2.1. Overview of KiwiSaver funds in New Zealand and performance
KiwiSaver is a New Zealand retirement savings scheme introduced in 2007. There were 2,837,656 active KiwiSaver accounts in
New Zealand as at end of March 2018 (shortly before the end of our sample time period in Dec 2018)6 with investments made by
KiwiSaver funds exhibiting a high degree of home biases. There is roughly $48.6 billion in total assets under management in Ki-
wiSaver accounts as of March 2018, with circa $7.3 billion held in New Zealand and Australian equities and nearly $14.1 billion held
in overseas equities.7 This compares to the market capitalization of New Zealand's stock exchange main board of equities of ap-
proximately $140 billion as at March 2019, which was less than 0.20 of 1% of the global market capitalization in US dollars.8
Moreover, KiwiSaver accounts are predicted to show strong growth and exceed $70 billion by the year 2020 (Heuser et al., 2015).
MacDonald et al. (2019) also report typical KiwiSaver fund allocations to New Zealand and Australian equities of between 44% and
70% of the total equity fund asset allocation for different risk profiles from default to aggressive growth funds.9
The literature investigating fund performance is limited. Frijns and Tourani-Rad (2015) report that the risk-adjusted performance
of KiwiSaver funds concentrated in domestic or international equity markets generally under-performed their benchmarks from
September 2007 to April 2013. Heuser et al. (2015) present similar results, with under-performance more pronounced and prevalent
amongst international funds. Bauer et al. (2006) also report that 143 New Zealand non-KiwiSaver domestic and international equity
mutual funds from 1990 to 2003 generally under-performed their benchmarks. Funds that have positive alpha were not found to be
statistically different from zero.
MacDonald et al. (2019) employ stochastic simulations to examine asset allocations that can improve the probabilities of adequate
retirement balances for KiwiSaver investors, particularly for investors in ‘defensive’, ‘conservative’ and ‘balanced’ strategies. The
6
Source: http://www.fma.govt.nz/assets/Reports/KiwiSaver-Report-2018.pdf
This is out of a resident population of approximately 4.5 million people.
7
Ibid.
8
Australia was only circa 2% of the global world market capitalization as at March 2018 (see https://www.statista.com/statistics/710680/global-
stock-markets-by-country/).
9
Chan et al. (2005) report that New Zealand funds examined in their study allocate in aggregate circa 75% of assets to New Zealand during 1999
to 2000. This was a time period when the New Zealand market was only circa 0.07% of world market capitalization. Those New Zealand funds that
had a foreign bias tilted most of their offshore portfolio to Australia (New Zealand's nearest geographical neighbour and where close economic ties
exist between the countries), with an overall allocation of circa 14% of equity assets to that country. At this time the Australian market represented
circa 1.2% of the world market capitalization.
3
results suggest that New Zealand's legislated default KiwiSaver funds are overly conservative and that investors may achieve better
results with higher allocations to equities. However, MacDonald et al. (2019) do not examine optimal weights to New Zealand
equities for investors in different tax brackets, but instead adopt asset weights based upon the average typical KiwiSaver
asset allocation across Australasian equities, NZ bills, NZ bonds, international equities and international bonds for different risk
profiles.10
2.2. Impact of taxes on portfolio asset allocation decisions
Taxes reduce investment returns and this affects portfolio allocation decisions and performance. In many countries, such as the
US, capital gains are taxed when assets are sold. To minimize these taxes investors have an incentive to realize capital losses and defer
capital gains (see Constantinides, 1984). Thus, capital gains taxes can create a capitalization effect that reduces demand and a lock-in
effect that can reduce supply (Dai et al., 2008).11 Trading and holding investments for purely tax reasons can be sub-optimal over
time if an investor is seeking benefits associated with diversification. That is, as a result of taxes, optimal portfolio allocations are
impacted where trading frictions exist between investors who face differential tax rates on asset classes.
Taxation cost differentials between investment in home and foreign markets may also explain some degree of investor home bias.
For example, Mishra and Ratti (2013) report that investor home bias may be partially explained by overseas taxes.12 The authors
control for information asymmetry and other behavioural and governance issues and report that dividend imputation is a statistically
significant impediment to cross border equity flows and international diversification. Tax credits available to domestic investors on
foreign tax paid reduces the extent of home bias. Another study by Christoffersen et al. (2005) report that Canadian fund managers
consider the tax preferences of investors when setting investment policy.
In a New Zealand setting Lally (2003) examines arguments by the New Zealand Stock Exchange (NZSE) that a minimum 20% of
the New Zealand Superannuation Fund portfolio should be invested in New Zealand equities. Lally (2003) rejects most of the ar-
guments advanced by the NZSE in favour of tilting the equity portfolio to domestic New Zealand investment. Nevertheless, Lally
concludes a local tilt of up to a 15% weighting in New Zealand equities could be justified where the New Zealand Superannuation
Fund is tax-exempt on local but not foreign assets. As far as we are aware, however, there has been no prior paper that empirically
examines the impact of differential taxes and weight constraints on the optimal portfolio asset allocation between domestic and
offshore equities.
3. Tax
This section provides an overview of the New Zealand taxation assumptions that we adopt with respect to equities held in a PIE.
The effective tax rate on New Zealand domestic investments is lower than those on overseas investments due to 1) New Zealand's
dividend imputation tax regime, 2) foreign investment fund rules, and 3) non-resident withholding tax imposed by the source country
on offshore dividends received by New Zealand tax residents.
3.1. Portfolio investment entity (PIE)
PIEs are typically managed funds operating as unit trusts, being open-ended collective investments, constituted under a trust deed
similar to mutual funds. From October 2007 unit trusts that meet the definition of a PIE were able to elect into tax rules under which
they are not taxable on capital gains on shares in New Zealand and Australian companies listed on an approved index.13 A PIE pays
tax on investment income for each investor based on that investor's prescribed investor rate (PIR). There are four prescribed investor
rates of 0%, 10.5%, 17.5% and 28%, which we use to measure the potential effect of taxes on the equity home bias in this study.
Individual investors have only one rate that they qualify for based on their income and tax residency status.14 Income in the PIE can
be taxed at a lower PIR than the investor's marginal tax rate. For example, the 28% PIR is lower than the 33% marginal tax rate for
individual investors with income in excess of $70,000.
3.2. Taxation of New Zealand income
Dividends paid by New Zealand companies to New Zealand tax residents receive imputation credits that substantially reduce the
income tax liability on cash dividends. The regime was introduced April 1, 1988. It is designed to avoid the double taxation of
company profits. When New Zealand companies pay corporate taxes in New Zealand on profits, imputation credits recorded in a
10
While MacDonald et al. (2019) seek to account for taxes, it is not explicitly clear how investments in New Zealand equities are taxed differently
to equity investments in Australia and other international markets.
11
In the case of New Zealand investors, the annual taxation of international gains and the inability to carry forward losses would likely eliminate
these trading incentives.
12
Warren (2010) also argues the home bias by Australian investors may be partially explained by factors such as anchoring formed from legacy
investments and industry peers.
13
All KiwiSaver default schemes are PIEs. Funds that do not meet the definition of a PIE are taxed on all earnings.
14
Only certain trustees are able to choose more than one tax rate.
4
memorandum account are created. These imputation credits can then be attached to cash dividends and provide an offset against the
New Zealand tax-resident investor's personal tax liability on gross dividend income. In this study we treat cash dividend income
received by a PIE and sourced from investment in New Zealand shares as effectively tax-free.15 Capital gains from New Zealand shares
held by a PIE are also assumed not to be taxed.
3.3. Taxation of foreign dividends (other than dividends from Australian shares)
Dividends received overseas are considered to be part of total taxable foreign investment fund (FIF) income. Dividends paid from
shares held overseas generally incur non-resident withholding tax (NRWT) in the source country in accordance with Double Taxation
Agreements (DTA) between New Zealand and the offshore jurisdiction. These agreements outline the rate at which sources of income
from investments are taxed in the country they are paid. Table 1 (cols. 12 and 13) identifies the countries that have DTAs with New
Zealand and reports the tax rates applied to dividends paid in the 34 countries. Thirty of the thirty-four countries studied in this paper
have DTAs with New Zealand. These DTA countries tax dividends at a 15% NRWT rate. Argentina, Brazil, Greece and Portugal do not
have DTAs and dividends are taxed at different NRWT rates between 0% and 20%.
Taxes paid on overseas dividends provide an offset or credit to reduce the income tax on the fair dividend rate (FDR). However, if
there is no New Zealand income tax payable on the FIF, the credits for foreign tax paid cannot be claimed. In addition, unused tax
credits cannot be used to reduce tax payable on other income such as Australian dividend income.16
3.4. Foreign investment fund (FIF)
The FIF rules were introduced from April 1, 2007.17 The FIF rules apply to investments in foreign companies and unit trusts. PIEs
use the FDR method to determine the income to be taxed from their foreign investments (other than certain Australian listed shares
on an approved index).18 The FDR of 5% is multiplied by the value of assets held at the beginning of the tax year. This amount is
treated as total income to be taxed at the PIR.
PIEs are generally held by many different unit holders and units in the PIE can be purchased and sold by different investors each
day. For this reason, PIEs generally calculate a daily closing value for existing units. The FDR is pro-rated by the number of trading
days in the period being measured compared to the number of days in the tax year. The value of the PIE at the beginning of the day is
multiplied by the pro-rated FDR, and this amount is taxed at the client's PIR to determine the tax owed for the trading day (See
Appendix 1, which also provides an example on how we impute monthly investor taxes in a PIE).19 The FDR method ignores positions
bought and sold in the tax period. These are treated separately as quick sale adjustments which are then added to the FDR of total
taxable income.20
3.5. Taxation of dividends and capital gains from Australian shares
Australian shares listed on an approved index such as the ASX All Ordinaries are exempt from being an attributing interest in a
FIF. In this study we treat all Australian dividends as fully franked, with no NRWT on dividends paid by Australian shares.21 Thus,
15
For a discussion of New Zealand's dividend imputation regime see Cliffe and Marsden (1992). In practice, not all dividends receive full
imputation credits. Lally (2000, page 6) estimates the ratio of imputation credits to cash dividends at 80% of the maximum possible rate. Our
simplifying assumption of 100% fully imputed dividends will marginally over-estimate the tax disadvantages of overseas investments for New
Zealand investors on the top marginal personal tax rate. However, given the New Zealand 28% corporate tax rate which is assumed constant over
the time period of our study, an 80% dividend imputation credit would still fully provide for investors' personal taxes with a PIR lower than 28%.
16
Tax credits can be used to offset the tax payable on the FDR income associated with that attributing interest. As explained in IR 461 (Inland
Revenue (2016). A guide to foreign investment funds and the fair dividend rate.): “Such foreign tax credits can only be used to reduce the income tax
payable on your FIF income. If there is no New Zealand income tax payable on your FIF investment, no claim can be made for the foreign tax paid on
any dividends received from the FIF. You can't use foreign tax credits to get a refund or reduce tax payable on other income. This includes other
foreign income with a different nature or source, e.g., dividends from companies with the Australian exemption and credits attached to United
Kingdom dividends.” (p.25)
17
The PIE rules address earlier differences in tax treatment between investors invested directly in New Zealand shares and those invested in shares
via a New Zealand collective investment vehicles (CIV). An investor holding New Zealand shares directly was not typically taxed on capital gains.
Under the old rules, an equivalent investment in a CIV was taxed on any realized New Zealand share gains. This was done because a CIV was
considered a business trading in shares in order to make an income. The new PIE tax rules make indirect foreign investments held in a PIE taxable in
a similar way as directly held foreign investments.
18
New Zealand Income Tax Act 2007, section EX 53.
19
For a more detailed explanation also refer to section 3.2.10 in Elliffe (2015).
20
In our study we use static portfolios, with portfolio asset weights adjusted each month to reflect differential changes in asset returns. Thus, we
do not consider quick sale adjustments because all portfolio adjustments would occur on the first trading day of each month. As explained in Elliffe
(2015), section 3.2.10(2): “If a fund does daily valuations, then no quick sale adjustment is required as all changes in the value of the portfolio will
be picked up in the FDR formula.”
21
As explained in Elliffe (2015), section 2.6.5(c): “Most Australian companies pay their dividends and attach Australian franking credits. When a
fully franked dividend is paid, the Australian company does not have to deduct Australian NRWT [non-resident withholding tax]. These Australian
franking credits are not creditable under the New Zealand provisions.” This study treats all dividends paid in Australia as fully franked.
5
Table 1
1993–2018 market characteristics.
Local Currency New Zealand Dollar
Col. 1 Col. 2a Col. 2b Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 Col. 8 Col. 9 Col. 10 Col. 11 Col. 12 Col. 13
Country (i) 1993 wi 2006 wi r sd r/sd r sd r/sd cap.gain div. div./total DTA Rate
Panel A: Developed markets

Australia 1.51 2.37 0.0914 0.129 0.71 0.0813 0.153 0.53 0.0384 0.0429 0.53 yes 0.15
Austria 0.21 0.41 0.0369 0.222 0.17 0.0241 0.210 0.11 0.0011 0.0230 0.95 yes 0.15
Belgium 0.58 0.86 0.0692 0.188 0.37 0.0559 0.186 0.30 0.0203 0.0355 0.64 yes 0.15
Canada 2.41 3.68 0.0884 0.146 0.60 0.0745 0.171 0.44 0.0495 0.0250 0.34 yes 0.15
Denmark 0.31 0.50 0.1217 0.179 0.68 0.1085 0.183 0.59 0.0892 0.0193 0.18 yes 0.15
Finland 0.17 0.57 0.1302 0.291 0.45 0.1190 0.287 0.42 0.0851 0.0339 0.28 yes 0.15
France 3.37 5.25 0.0741 0.177 0.42 0.0616 0.171 0.36 0.0334 0.0281 0.46 yes 0.15
Germany 3.43 3.54 0.0778 0.205 0.38 0.0645 0.195 0.33 0.0386 0.0259 0.40 yes 0.15
Hong Kong 2.85 1.93 0.0889 0.245 0.36 0.0774 0.219 0.35 0.0433 0.0341 0.44 yes 0.15
Ireland 0.14 0.35 0.0391 0.208 0.19 0.0241 0.203 0.12 0.0011 0.0230 0.95 yes 0.15
Italy 1.01 2.22 0.0563 0.212 0.27 0.0399 0.220 0.18 0.0061 0.0338 0.85 yes 0.15
Japan 22.18 10.21 0.0203 0.183 0.11 0.0149 0.170 0.09 0.0004 0.0145 0.97 yes 0.15
Netherlands 1.34 1.68 0.0901 0.180 0.50 0.0766 0.174 0.44 0.0449 0.0317 0.41 yes 0.15
N.Z. 0.19 0.10 0.0776 0.157 0.49 0.0776 0.157 0.49 0.0235 0.0540 0.70 yes 0.15
Norway 0.20 0.61 0.0997 0.211 0.47 0.0791 0.223 0.36 0.0441 0.0350 0.44 yes 0.15
Singapore 0.98 0.60 0.0562 0.215 0.26 0.0529 0.206 0.26 0.0238 0.0291 0.55 yes 0.15
Spain 0.88 2.86 0.0976 0.209 0.47 0.0765 0.211 0.36 0.0397 0.0368 0.48 yes 0.15
Sweden 0.79 1.24 0.1213 0.212 0.57 0.1002 0.218 0.46 0.0706 0.0296 0.30 yes 0.15
Switzerland 2.01 2.62 0.0823 0.150 0.55 0.0877 0.142 0.62 0.0639 0.0238 0.27 yes 0.15
UK 8.52 8.20 0.0692 0.133 0.52 0.0512 0.137 0.37 0.0154 0.0358 0.70 yes 0.15
US 37.98 41.98 0.0918 0.144 0.64 0.0807 0.143 0.56 0.0595 0.0212 0.26 yes 0.15
Panel B: Emerging markets

Argentina 0.33 0.17 0.1076 0.387 0.28 −0.0466 0.418 −0.11 −0.0679 0.0213 −0.46 no 0
Brazil 0.74 1.54 0.4561 0.385 1.18 0.1109 0.348 0.32 0.0700 0.0409 0.37 no 0
Chile 0.33 0.38 0.0983 0.185 0.53 0.0624 0.216 0.29 0.0335 0.0289 0.46 yes 0.15
Greece 0.09 0.45 0.0272 0.293 0.09 0.0040 0.289 0.01 −0.0199 0.0238 6.03 no 0.1
Indonesia 0.24 0.30 0.1456 0.313 0.47 0.0523 0.380 0.14 0.0256 0.0266 0.51 yes 0.15
Korea 1.03 1.81 0.0838 0.274 0.31 0.0584 0.310 0.19 0.0410 0.0175 0.30 yes 0.15
Malaysia 1.63 0.51 0.0658 0.238 0.28 0.0365 0.274 0.13 0.0108 0.0257 0.70 yes 0
Mexico 1.49 0.75 0.1448 0.217 0.67 0.0556 0.269 0.21 0.0361 0.0195 0.35 yes 0.15
Philippines 0.3 0.15 0.0718 0.251 0.29 0.0286 0.267 0.11 0.0087 0.0200 0.70 yes 0.15
Portugal 0.09 0.23 0.0510 0.196 0.26 0.0332 0.203 0.16 −0.0030 0.0362 1.09 no 0.20
Taiwan 1.43 1.29 0.0661 0.252 0.26 0.0476 0.255 0.19 0.0207 0.0269 0.56 yes 0.15
Thailand 0.97 0.30 0.0525 0.327 0.16 0.0317 0.325 0.10 0.0022 0.0295 0.93 yes 0.15
Turkey 0.28 0.35 0.3763 0.425 0.89 0.0638 0.473 0.13 0.0350 0.0289 0.45 yes 0.15
This table reports the percentage of market capitalization for each market as the percent of market capitalization of all 34 markets combined in US
dollars at the end of the year 1993 and 2006. The geometric annual returns (r), standard deviation of returns (sd), and the return-to-risk ratios (r/sd)
of each market in the local currency and in New Zealand dollars is also presented. The annual capital gains (cap. gain), dividends (div.) and the
percentage of gains from dividends as a percent of the total gains of these markets in New Zealand dollars (div./total) is reported. Whether a country
has a double taxation agreement (DTA) with New Zealand is reported. The withholding tax rate applied to dividends paid to non-residents of each
country is also reported.
dividend income from Australian shares are treated as income by the PIE and taxes are deducted based on the investor's PIR. Capital
gains from Australian shares that are listed on the approved index are assumed to be tax exempt.
4. Data and Methods
4.1. Data
The monthly MSCI price and total return data for 34 international markets priced in New Zealand dollars is used to investigate the
effect of taxes on dividends and capital gains on the potential benefits of international diversification for New Zealand investors. The
price data represents the capital gains realized by the markets. The difference between the price and the total monthly returns
represent the dividend portion of gains paid by these markets. The sample period covers December 31, 1992 to December 31, 2018. A
time series of monthly New Zealand treasury rates is also used to calculate market portfolio after-tax excess returns in New Zealand
dollars. The data is collected from Datastream.
Annual market capitalization data in US dollars for 1993 and 2006 is from two sources: the World Bank and the World Federation
of Exchanges. The World Bank data is available for 33 of the 34 markets. The 1993 data is incomplete for Ireland. This incomplete
data is calculated using changes to the MSCI total return data to backward fill the missing capitalization values from the first available
6
1995 market capitalization data point. The Taiwanese market capitalization data is retrieved from the World Federation of
Exchanges.
4.2. International diversification
We consider two types of portfolio optimization. Investors may choose to maximize the return-to-risk (RR) payoff of their
portfolio or choose to minimize the volatility of their portfolio. To achieve either goal, investors can decide to allocate funds in
international markets where the investment opportunities of those markets are stated as a vector of multivariate Gaussian stochastic
returns of n assets, RT = [r1, r2, …, rn]. The estimated geometric after-tax returns for these markets can be expressed as a vector μ. The
variance-covariance of estimated after-tax asset returns can be expressed as a positive definite matrix V. Let S be the set of all real
vectors w that define the weights such that wT1 = w1 + w2 + … + wn = 1, where 1 is an n vector of ones. Using the methods
developed by Markowitz (1952), the efficient frontier of global investments can be formed when the objective function and re-
strictions are combined in order to find the efficient portfolio that minimizes volatility at every level of expected return:
1 T
min = w Vw + ( µp wT µ ) + (1 wT 1)
{w , , } 2 (1)
where μp is the expected after-tax return of the portfolio, and the shadow prices ϕ and η are two positive constants. The quadratic
programming solution of assets in a portfolio spanning wp can be obtained by the first-order conditions of Eq. (1).
An investor may choose to diversify out of the domestic market and into the maximum return-to-risk portfolio (MRRP). When
there are no weight restrictions on the amount invested in any one market, the MRRP can be determined using the estimated after-tax
asset returns and the estimated variance-covariance of after-tax asset returns to find the vector w of weights that define the estimated
after-tax maximum return-to-risk (MRR) ratio achieved by the MRRP such that:
wpT µ
MRR = max T
| wpT S
{wp } (wp Vwp )1/2 (2)
where wp is the vector of weights of the assets held in the MRRP. The MRR measures the estimated after-tax return achieved by the
MRRP compared to its estimated standard deviation of returns.
Similarly, an investor may choose to diversify out of the domestic market and into the minimum volatility portfolio (MVP). When
there are no weight restrictions, the MVP can be determined using the estimated variance-covariance of asset after-tax returns to find
the vector w of weights that define the minimum volatility (MV) achieved by the MVP such that:
MV = min { (wpT Vwp )1/2| wpT S}
{wp } (3)
where wp is the vector of weights of the assets held in the MVP.
Investors may be limited in how much they can invest in overseas markets for various reasons. We consider varying degrees of
weight constraints on overseas markets in proportion to their relative market capitalizations in order to simulate these barriers as
investing in small overseas markets is in general, more costly and subject to greater information asymmetry. Furthermore, these
weight constraints can be interpreted as how far investors are allowed to deviate from the CAPM benchmark. When an investor has
constraints on the weights that can be assigned to assets held in the optimized portfolio and assuming no short sales, Eqs. (2) and (3)
can be solved such that wpT ∈ Sc where Sc = {wp ∈ S : 0 ≤ wi ≤ kw(Cap)i ≤ 1, i = 1, 2, ⋯, n}, where k is the weight constraint
multiplier greater than or equal to one, and w(Cap)i is the weight of the market value of each country i in the set of markets. We
analyse the changes to the benefits of diversification when constraints are relaxed and weighting the amount invested in international
markets can occur with the maximum weights set to 1, 2, 5 and 10 times the market value of each country. In addition, we also
consider the benefits when there are no weight (nw) constraints. The value-weighted global market portfolio where all market
weights are given by their relative market capitalization weights in our entire sample is referred to in this paper as the naive market
capitalization weighted (1/M) portfolio. The optimized portfolio that constrains all markets in the same way, except for New Zealand,
with k set to 1 is referred to as the 1/M⁎ portfolio. In the 1/M⁎ portfolio, the domestic New Zealand market allocation is not
constrained in order to measure the potential effect of taxes on the home bias in the optimal portfolios. We relax weight constraints
on overseas markets by allowing higher weight multiplier values and refer to the resulting optimized portfolios as wxk portfolios – for
example, wx10 for the case where k is set to 10. As unit trusts and PIEs are likely to restrict short sales, we do not allow short sales in
the optimization solutions.
4.3. Estimated returns
This study uses the historical monthly index returns of 34 markets denominated in New Zealand dollars from January 1993 to
December 2018 to assess the potential effect of taxes on the benefits of international diversification for equity investments held in a
PIE. The estimated optimal MVPs and MRRPs held in a PIE with different tax rates are determined using a vector μ of estimated after-
tax asset returns and the estimated variance-covariance of after-tax asset returns for these markets. The estimated monthly after-tax
market returns (in New Zealand dollars) are calculated by deducting the appropriate overseas dividend withholding taxes and New
Zealand taxes to be paid each month for each individual market.
As already noted, we assume New Zealand dividends and capital gains are not taxed. Australian equities are treated as exempt
7
from being an attributing interest in a FIF. All Australian dividends are assumed to be fully franked and capital gains on Australian
shares are not taxed. For all other markets, taxes withheld on overseas dividends are available as a tax credit to reduce the New
Zealand FDR taxes. The New Zealand taxes to be paid are equal to the monthly pro-rated FDR multiplied by the PIR (see Appendix 1).
The after-tax monthly returns for each market are used to calculate the estimated geometric annual after-tax market returns used in
the optimization procedure. The monthly after-tax returns are also converted to logarithmic monthly returns which are used to build
the variance-covariance matrix employed in the optimization solution.
4.4. Measuring the risk-adjusted performance
The monthly after-tax returns achieved by the estimated MRRP and the MVP are derived to measure the benefits from diversi-
fication. Using the same historical data as that used in the optimization stage, the after-tax monthly returns achieved by the static
portfolio are calculated.22 The geometric annual return and the standard deviation of returns for the portfolio are calculated from the
portfolio's monthly after-tax returns. These are used to determine the Sharpe and Sortino ratios of the MRRP and MVP. The Sharpe
ratio (SR) is calculated as:
r rf
SR =
where r is the annualised geometric after-tax portfolio return, rf is the annualised geometric after-tax risk-free return, and σ is
the annualised standard deviation of the after-tax portfolio return.
The Sortino ratio (ST) uses downside deviation as a measure of risk. Sortino and Meer (1991) argue downside deviation better
captures the risk–return measure where investors are concerned about the likelihood and magnitude of adverse outcomes. The ST is
defined as:
r rf
ST =
ST
where σST is the annualised standard deviation of the after-tax portfolio vector of downside returns, where the returns are the
minimum of [zero; the monthly after-tax portfolio return less the after-tax risk-free return for the same month].
The Sharpe ratios of the MRRP, MVP and the variance of returns of the MVP (VMVP) are then used to measure the benefits from
diversification.
4.5. Measuring the potential benefits of international diversification
For the New Zealand investor, adjusting their portfolio out of the domestic market and into the internationally diversified
portfolio that maximizes the return-to-risk ratio or minimizes volatility, the increment of unit-risk performance is measured as:
SRP
= 1
SRNZ (4)
where δ (delta) represents the percentage improvement that the investor achieves from the Sharpe ratio of the optimized inter-
nationally diversified portfolio (SRP) over the New Zealand market's Sharpe ratio (SRNZ).
The reduction in volatility that can result from diversification out of the domestic market and into the MVP is measured as:
1/2
VMVP
=1
VNZ (5)
where VNZ is the variance of the domestic New Zealand market portfolio and VMVP is the variance of the MVP. A positive ε (epsilon)
value represents the percentage reduction in portfolio volatility when diversifying out of the domestic portfolio and into the optimal
internationally diversified MVP. A result approaching zero represents diminishing benefits from diversification.
The results are tested for statistical significance using the Ledoit and Wolf (2008) bootstrap testing methods, which are designed
to test whether the difference between two portfolio strategies is significantly different from zero. The Ledoit and Wolf (2008)
bootstrap procedure is designed to address the non-normality of returns and fat-tail events that occur with historical financial data.
Using bootstrap techniques, regressions are performed on paired data points of a given block size between the monthly returns of two
portfolios to provide a p-value measuring the significance of the hypothesis that the difference between the two portfolios is zero.
Using a block size of 5 with 499 iterations, the Sharpe and variance bootstrap methods are used to test the significance of the MRRP
and MVP performance versus the domestic New Zealand market.
22
The classic mean-variance optimisation procedure finds the optimal static portfolio weights to fit the mean-variance distributions used as inputs
(i.e., the ex-post numbers taken from the historical data).
8
Fig. 1. 1/M versus N.Z. (1993–2018).

This figure shows the difference in untaxed annual returns between the naive global 1/M portfolio and the domestic New Zealand market portfolio
in New Zealand dollar terms for all possible holding periods greater than or equal to 60 months from January 1993 to December 2018.
5. Results
5.1. Full period analysis
Fig. 1 presents the distribution of the difference in (annualised) returns between the untaxed naive global 1/M portfolio and the
local New Zealand market over various monthly holding periods from January 1993 to December 2018. The returns from the static 1/
M portfolio are calculated using the market capitalization weights from 1993. From each month, starting at January 1993, the returns
for the domestic market portfolio and the naive 1/M portfolio are calculated for each possible monthly holding period to the end of
2018. Starting from January 1993 there are 252 measurable sixty-month holding periods and a single measurable 312-month holding
period.
Fig. 1 shows that the 1/M portfolio does not always outperform the local market portfolio. The median of the 1/M portfolio
holding period returns tends to be similar to that of the domestic portfolio. The figure shows that there is a larger distribution in the
potential difference in annual returns between the 1/M portfolio and local portfolio during shorter holding periods than compared to
longer holding periods. These results suggest that depending on the sample period used, diversification into the 1/M portfolio could
provide potential excess returns beyond the New Zealand local portfolio if there were no taxes. However, there is a great deal of
sample variation in the size of these gains across time.
Table 1 reports characteristics of the 34 markets considered in the study. The second column (Col. 2a headed 1993 wi) presents
the market capitalization of each market used in US dollars as a percent of all 34 markets combined at the end of 1993. The next
column (Col. 2b headed 2006 wi) presents the market capitalization of each market used in US dollars as a percent of all 34 markets
combined at the end of 2006. The global market portfolio we consider is dominated by the US and Japan accounting for more than
60% of the portfolio combined using 1993 weights and 52% using 2006 weights. The third largest market, UK, comprises
8.20%–8.52% of the entire portfolio. The rest are in general considerably smaller compared to the markets of the US, Japan and the
UK. New Zealand for example, makes up only 0.19% (1993 weights) and 0.10% (2006 weights) of the global market. For investors
who face weight constraints based on relative market capitalizations of the countries considered, these constraints are likely to be
more binding for small markets.
Table 1, Columns 3–8 report the geometric annual returns (r), the standard deviation of returns (sd) and the return-to-risk ratios
of these markets as measured in the local currency (Table 1, Cols. 3 to 5) and the New Zealand dollar (Table 1, Cols. 6 to 8) for the
1993 to 2018 holding period. A comparison of the market returns realized in each local currency and the returns realized in the New
Zealand currency shows that the New Zealand dollar has appreciated against foreign currencies over the sample period in this study.
As a result, market return-to-risk ratios are often lower in the New Zealand currency than they are in their own domestic currency.
For example, a US dollar denominated US market geometric return of 9.18% translates to an 8.07% New Zealand dollar denominated
return. The US dollar denominated US market RR ratio is 0.64 and in New Zealand dollar denominated terms it is 0.56. A comparison
of the return-to-risk values in New Zealand dollar terms shows that only the four markets of Australia, Denmark, Switzerland, and the
US provide a higher return-to-risk ratio than the New Zealand market during 1993–2018.
The gains derived from capital gains (Table 1, Col. 9) and dividends (Table 1, Col. 10) in New Zealand dollars are also presented.
9
Table 2
Optimal domestic allocations.
MRRP domestic market weight MVP domestic market weight
Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 Col. 8 Col. 9 Col. 10 Col. 11
Case wx1 wx2 wx5 wx10 nw wx1 wx2 wx5 wx10 nw

No Tax 0.4779 0.3894 0.3528 0.2945 0.2503 0.3443 0.3177 0.2875 0.2712 0.2699
0% PIR 0.4937 0.4075 0.3677 0.3100 0.2590 0.3443 0.3176 0.2876 0.2713 0.2696
10.5% PIR 0.5034 0.4247 0.3828 0.3210 0.2932 0.3445 0.3179 0.2880 0.2717 0.2701
17.5% PIR 0.5204 0.4401 0.4038 0.3500 0.3200 0.3447 0.3181 0.2883 0.2719 0.2704
28% PIR 0.5507 0.4787 0.4383 0.3917 0.3647 0.3450 0.3184 0.2887 0.2723 0.2709
This table reports the domestic New Zealand market allocation held in each MRRP and MVP optimized with different tax levels and allocation
weight constraints. The portfolios do not implement short sales and the domestic New Zealand market weight is unconstrained in the optimization
solutions. Each overseas market weight is constrained by the country's 1993 market capitalization weight at the one times (wx1), two times (wx2),
five times (wx5), ten times (wx10) and no weight constraint (nw) level. The portfolios are optimized with no taxes (No Tax), or with non-resident
withholding taxes on overseas dividends and the local New Zealand FDR taxes calculated with the PIR set to 0%, 10.5%, 17.5% or 28%.
All markets returned positive returns in New Zealand dollars from dividend income and most (but not all) markets also showed
capital price appreciation. Australia and New Zealand show the highest dividend yields of 4.29% and 5.40% respectively.
Table 2 reports the optimal domestic New Zealand market weights in MRRPs and MVPs optimized with no taxes (No Tax), or with
non-resident withholding taxes on overseas dividends and the local New Zealand FDR taxes calculated with the PIR set to 0%, 10.5%,
17.5% or 28%. First, for the return-to-risk maximizing investors, the baseline case, where there are no weight constraints (nw) on an
investment in any market (Table 2, Col. 6) and no taxes, the results show that the optimal portfolio weight in the New Zealand
domestic market during 1993–2018 is 25.03%. The seemingly high optimal weight on the New Zealand domestic market may be in
part due to the high New Zealand dollar denominated returns on the domestic market over our sample period seen in Table 1.
On an after-tax basis, dividend imputation credits attached to New Zealand domestic cash dividends and the FIF rules for overseas
returns favour the home market. When a 0% PIR is considered the optimal New Zealand allocation with no weight constraints
increases to 25.90% from 25.03% (Table 2, Col. 6). This is because although a 0% PIR does not result in New Zealand taxes owed on
overseas investment, foreign governments still collect NRWT on overseas dividends. Overall, the tax advantages of investing in the
New Zealand home market increase with the investor's PIR due to the FIF rules. For example, a PIR of 10.5% makes the New Zealand
home market even more attractive and increases the optimal home market allocation to 29.32%. At the 28% PIR top tax bracket, the
optimal New Zealand weight is 36.47%.
Not surprisingly, constraining overseas market weights or the amount that can be invested in any offshore markets increases New
Zealand domestic investors' optimal investment allocation further towards the home market. In the case of the 1/M⁎ or weighting
times one (wx1) portfolio with no taxes, the New Zealand allocation is 47.79% (Table 2, Col. 2). Again, the tax advantages of the
domestic market tilt the weighting further to a point where the optimal domestic allocation is over 50% for the tax brackets 10.5%,
17.5% and 28%. As one would expect, relaxing the weighting constraints on overseas markets reduces the optimal New Zealand
weight towards the baseline case with no constraints. For example, without taxes, weakening the weight constraints to a maximum of
ten times (wx10) the market capitalization weight reduces the allocation to the New Zealand market to 29.45% (Table 2, Col. 5).
More importantly, the impact of taxes for various levels of weight constraints are remarkably consistent; we see roughly a 7% to 11%
increase in the optimal domestic allocation from no taxes to a 28% PIR.
In the case of investors who minimize variance, taxes at each weight constraint level have little impact on the domestic allocation.
For example, in the case of the 1/M⁎ portfolio (wx1) with no taxes, the domestic allocation is 34.43%, which is very similar to the
allocation of 34.50% at the 28% PIR rate (Table 2, Col. 7). This is because while (known) taxes reduce overseas returns, taxes have
little impact on the variances of returns. This is particularly true where taxes for international equities (other than Australia) under
the FDR are fixed at 5% of the opening asset value. Like before, the optimal domestic allocation is reduced as the weight constraints
are relaxed. For example, the New Zealand allocation of 34.50% is with the 1/M⁎ portfolio taxed at a 28% PIR, which compares to the
New Zealand allocation for the unconstrained MVP with no taxes of 27.09% (Table 2, Col. 11). Overall, the results presented in
Table 2 suggest that taxes on overseas equity investments and weight constraints imposed on overseas allocations held in a unit-trust
are factors that can contribute towards a home bias.
In order to show the benefits of international diversification, Table 3 reports the after-tax excess geometric annual returns,
standard deviation of returns, Sharpe ratios, and Sortino ratios for the MRRPs and MVPs optimized with different weight constraints
and tax levels, along with the 1/M portfolio and the domestic New Zealand market portfolio. Recall that the 1/M portfolio is the naive
market capitalization weighted portfolio that the international CAPM suggests investors would hold in order to capture the naive
diversification benefits from sharing risk. The New Zealand market allocation in this portfolio is only 0.19% using year 1993 weights.
Panel A of Table 3 reports the annual excess returns for each optimized portfolio with different tax rates and the annual excess returns
of the domestic New Zealand market which is effectively not taxed. The untaxed 1/M portfolio achieves annual excess after-tax
returns of 1.3% versus the New Zealand untaxed market portfolio excess after-tax return of 2.4% (Table 3, Panel A, Cols. 3 and 2). As
expected, relaxing the weight constraints on the 1/M portfolio increases the returns realized by the MRRP. For example, the excess
return of the untaxed MRRP optimized with no weight constraints is 3.9% (Table 3, Panel A, Col. 8). Also reported are the increase in
excess after-tax returns at each tax level compared to the no-tax case and 0% PIR. For example, the domestic portfolio taxed at a 28%
10
Table 3
Portfolio characteristics.
MRRP overseas markets constrained by … MVP overseas markets constrained by …
Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 Col. 8 Col. 9 Col. 10 Col. 11 Col. 12 Col. 13
Tax Rate NZ 1/M wx1 wx2 wx5 wx10 nw wx1 wx2 wx5 wx10 nw
Panel A: Excess return

No Tax 0.024 0.013 0.031 0.032 0.033 0.033 0.039 0.016 0.015 0.015 0.019 0.019
0% PIR 0.024 0.009 0.029 0.030 0.032 0.032 0.037 0.014 0.013 0.013 0.016 0.016
10.5% PIR 0.029 0.013 0.034 0.034 0.035 0.035 0.040 0.018 0.017 0.017 0.020 0.020
17.5% PIR 0.033 0.013 0.036 0.036 0.037 0.037 0.042 0.020 0.018 0.018 0.021 0.022
28% PIR 0.039 0.013 0.039 0.039 0.040 0.040 0.045 0.022 0.020 0.020 0.023 0.023
Panel B: Standard deviation

No Tax 0.157 0.128 0.120 0.118 0.116 0.113 0.117 0.113 0.111 0.110 0.109 0.109
0% PIR 0.157 0.128 0.120 0.118 0.116 0.113 0.117 0.113 0.111 0.110 0.109 0.109
10.5% PIR 0.157 0.128 0.120 0.118 0.116 0.113 0.117 0.113 0.111 0.110 0.109 0.109
17.5% PIR 0.157 0.128 0.121 0.118 0.116 0.113 0.117 0.113 0.111 0.110 0.109 0.109
28% PIR 0.157 0.128 0.121 0.119 0.117 0.114 0.118 0.113 0.111 0.110 0.109 0.109
Panel C: Sharpe ratio

No Tax 0.151 0.098 0.259 0.270 0.285 0.295 0.335 0.145 0.135 0.140 0.170 0.172
0% PIR 0.151 0.072 0.245 0.255 0.272 0.279 0.319 0.127 0.114 0.118 0.148 0.151
10.5% PIR 0.187 0.099 0.283 0.291 0.304 0.312 0.347 0.164 0.152 0.156 0.183 0.186
17.5% PIR 0.212 0.101 0.298 0.306 0.318 0.326 0.361 0.177 0.165 0.168 0.196 0.199
28% PIR 0.248 0.103 0.323 0.331 0.341 0.349 0.383 0.196 0.183 0.185 0.212 0.215
Panel D: Sortino ratio

No Tax 0.235 0.154 0.400 0.421 0.447 0.469 0.546 0.226 0.210 0.221 0.270 0.274
0% PIR 0.235 0.113 0.378 0.395 0.425 0.442 0.517 0.197 0.178 0.185 0.234 0.239
10.5% PIR 0.295 0.156 0.439 0.455 0.478 0.496 0.562 0.256 0.238 0.246 0.292 0.297
17.5% PIR 0.335 0.160 0.465 0.480 0.502 0.520 0.586 0.278 0.259 0.267 0.314 0.318
28% PIR 0.395 0.163 0.508 0.521 0.539 0.558 0.623 0.308 0.289 0.295 0.341 0.345
This table reports the excess geometric annual return, the standard deviation of returns, the Sharpe ratio, and the Sortino ratio for each portfolio
optimized at different tax levels and allocation weight constraints. The domestic New Zealand (NZ) market and the 1/M portfolio characteristics are
also presented. The 1/M portfolio is the portfolio with all market weights, including the New Zealand market, set by the 1993 market capitalization
weights. The MRRPs and the MVPs are optimized with no shorts and the international markets constrained by the 1993 market capitalization
weights at the one times (wx1), two times (wx2), five times (wx5), ten times (wx10) and no weight constraint (nw) level. The domestic New Zealand
market weight is unconstrained in the MRRP and MVP optimization solutions. The portfolios are optimized with no taxes (No Tax), or with
PIR has an excess after-tax return of 3.9% (Table 3, Panel A, Col. 2). This increases to 4.5% for the no weight constraint portfolio
when the PIR rate is 28% (Table 3, Panel A, Col. 8). The increase in after-tax excess returns as the tax rate increases reflects a lower
average effective tax rate on equity investments compared the tax on the risk free asset. While the excess returns for the MVPs
similarly increase as the PIR increases, the MVPs do not achieve excess returns greater than the New Zealand domestic market
portfolio (Table 3, Panel A, Cols. 9 to 13).
Table 3, Panel B, reports that the standard deviations of after-tax returns for the MRRPs and MVPs are lower than that of the New
Zealand market for all weight constraint and tax levels. Table 3, Panel C, reports the Sharpe ratios for the various portfolios. The
taxed 1/M portfolio exhibits Sharpe ratios that are smaller than the New Zealand market (Table 3, Panel C, Cols. 2 and 3). The MRRPs
with relaxed weight and nw constraint levels have Sharpe ratios that are greater than the New Zealand market at each tax level
(Table 3, Panel C, Cols. 2 and 4–8). The MVPs with relaxed weight and nw constraint levels have Sharpe ratios that are greater than
the naive 1/M portfolio (Table 3, Panel C, Cols. 3 and 9 to 13). However, only the untaxed MVP at the wx10 and nw constraint levels
have Sharpe ratios greater than the New Zealand market portfolio (Table 3, Panel C, Cols. 2, 12 and 13).
The results for Sortino ratios are reported in Table 3, Panel D. Again, the taxed 1/M portfolio exhibits Sortino ratios that are
smaller than the New Zealand market (Table 3, Panel D, Cols. 2 and 3). Similarly, the MRRPs with relaxed constraint levels have
Sortino ratios that are greater than the New Zealand market at each tax level (Table 3, Panel D, Cols. 2 and 4 to 8). Again, the MVPs
with relaxed weight constraint levels have Sortino ratios that are greater than the naive 1/M portfolio (Table 3, Panel D, Cols. 3 and 9
to 13). The MVPs for the wx10 constraint at no tax and the MVPs for the nw constraint at no tax, 0% PIR and 10.5% PIR have Sortino
ratios that are marginally greater than the New Zealand market portfolio (Table 3, Panel D, Cols. 2 and 12–13, compared to Col. 2).
To gauge the statistical and economic significance of the results, Table 4 reports the delta and epsilon results for the MRRPs and
MVPs with various weight constraints and tax rates. Table 4, Panel A reports the percentage increase in the Sharpe ratio when moving
from the domestic New Zealand market portfolio to the MRRP. The results from the Ledoit and Wolf (2008) Sharpe ratio tests are
presented in parentheses. The p-values report the significance of the hypothesis that the difference between the Sharpe ratio of the
monthly returns of the domestic portfolio and the optimized portfolio is zero. The last five columns (Table 4, Cols. 8 to12) in the table
11
Table 4
Diversification benefits.
Overseas market weights constrained by… Max. benefit captured
Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 Col. 8 Col. 9 Col. 10 Col. 11 Col. 12
Tax Rate 1/M wx1 wx2 wx5 wx10 nw 1/M wx1 wx2 wx5 wx10
Panel A: MRRP delta

No Tax −0.3487 0.7200 0.7947 0.8943 0.9608 1.2272 −0.28 0.59 0.65 0.73 0.78
(0.780) (0.510) (0.486) (0.526) (0.470) (0.344)
0% PIR −0.5216 0.6270 0.6914 0.8077 0.8560 1.1213 −0.47 0.56 0.62 0.72 0.76
(0.672) (0.546) (0.558) (0.570) (0.526) (0.384)
10.5% PIR −0.4726 0.5077 0.5556 0.6220 0.6626 0.8496 −0.56 0.60 0.65 0.73 0.78
(0.636) (0.540) (0.542) (0.582) (0.532) (0.394)
17.5% PIR −0.5222 0.4063 0.4460 0.5031 0.5408 0.7037 −0.74 0.58 0.63 0.71 0.77
(0.568) (0.574) (0.584) (0.606) (0.552) (0.414)
28% PIR −0.5849 0.3014 0.3321 0.3727 0.4055 0.5412 −1.08 0.56 0.61 0.69 0.75
(0.466) (0.612) (0.600) (0.648) (0.588) (0.436)
Panel B: MVP epsilon

No Tax 0.1861 0.2822 0.2931 0.3006 0.3039 0.3039 0.61 0.93 0.96 0.99 1.00
(0.008) (0.002) (0.002) (0.002) (0.002) (0.002)
0% PIR 0.1860 0.2821 0.2931 0.3007 0.3041 0.3041 0.61 0.93 0.96 0.99 1.00
(0.008) (0.002) (0.002) (0.002) (0.002) (0.002)
10.5% PIR 0.1857 0.2820 0.2929 0.3005 0.3039 0.3039 0.61 0.93 0.96 0.99 1.00
(0.008) (0.002) (0.002) (0.002) (0.002) (0.002)
17.5% PIR 0.1855 0.2818 0.2927 0.3003 0.3036 0.3036 0.61 0.93 0.96 0.99 1.00
(0.008) (0.002) (0.002) (0.002) (0.002) (0.002)
28% PIR 0.1851 0.2816 0.2925 0.3000 0.3033 0.3033 0.61 0.93 0.96 0.99 1.00
(0.008) (0.002) (0.002) (0.002) (0.002) (0.002)
Panel C: MVP delta

No Tax −0.3487 −0.0347 −0.1048 −0.0690 0.1272 0.1429 −2.44 −0.24 −0.73 −0.48 0.89
(0.780) (0.834) (0.778) (0.820) (0.972) (0.986)
0% PIR −0.5216 −0.1564 −0.2422 −0.2166 −0.0173 0.0032 −164.57 −49.34 −76.42 −68.34 −5.47
(0.672) (0.772) (0.672) (0.714) (0.876) (0.890)
10.5% PIR −0.4726 −0.1271 −0.1889 −0.1697 −0.0220 −0.0081 58.02 15.60 23.19 20.84 2.71
(0.636) (0.766) (0.676) (0.716) (0.864) (0.886)
17.5% PIR −0.5222 −0.1650 −0.2216 −0.2057 −0.0745 −0.0628 8.32 8.31 2.63 3.53 3.28
(0.568) (0.720) (0.624) (0.668) (0.818) (0.838)
28% PIR −0.5849 −0.2123 −0.2638 −0.2550 −0.1447 −0.1354 4.32 1.57 1.95 1.88 1.07
(0.466) (0.640) (0.542) (0.582) (0.732) (0.758)
This table reports the delta and epsilon benefits from diversification out of the New Zealand domestic market and into the MRRPs and the MVPs
optimized with no shorts and the international markets constrained by the 1993 market capitalization weights at the one times (wx1), two times
(wx2), five times (wx5), ten times (wx10) and no weight constraint (nw) level. The domestic New Zealand market weight is unconstrained in the
MRRP and MVP optimization solutions. The portfolios are optimized with no taxes (No Tax), or with withholding taxes on overseas dividends and
the local New Zealand FDR taxes calculated with the PIR set to 0%, 10.5%, 17.5% or 28%. The results for the 1/M portfolio, which is the portfolio
formed where each market weight is equal to the 1993 market capitalization weight, are also presented. The p-value results from the Ledoit and
Wolf (2008) Sharpe ratio and volatility tests using a block size of 5 and 499 iterations are presented in parentheses. The last five columns report the
percent of the unconstrained portfolio benefits that are captured by each weight constrained portfolio.
report the percent of the unconstrained portfolio gains that are captured by the constrained portfolios at each tax level. At each level
of weight constraint, taxes reduce the diversification benefits. Taxes on overseas equity investment reduce the already negative
measured benefits of diversifying into the naive untaxed 1/M portfolio, with delta being negative in all cases. The 1/M⁎ portfolio with
no taxes provides a positive delta benefit of 72.00% (Table 4, Panel A, Col. 3). This 1/M⁎ portfolio (wx1) captures the majority of
potential gains available with the unconstrained portfolio at each tax level. For example, 56% of the potential diversification benefit
provided by the unconstrained MRRP taxed with a 28% PIR can be captured by the 1/M⁎ portfolio (Table 4, Panel A, Col. 9). As the
weight constraints are relaxed the potential benefits from diversification increase. Taxes reduce the benefits from diversification,
which still remain positive at the highest PIR of 28%. Overall the results illustrate a reduction in the benefits (fall in the MRRP delta)
of investing outside New Zealand as investor PIR rates increase. The results suggest that taxes affect the expected benefit of di-
versification. We note, however, that none of the in-sample MRRP optimization results over the period of our study result in a
statistically different Sharpe ratio compared to an investment in the domestic market portfolio only.
Table 4, Panel B, reports the percentage reduction in portfolio volatility when moving from the New Zealand domestic market
portfolio to the optimized MVPs. The Ledoit and Wolf (2008) p-values indicate that reductions in volatility are statistically significant.
At each tax level, the 1/M portfolio captures circa 61% of the potential volatility reducing benefits available from the unconstrained
MVP. The 1/M⁎ portfolio can capture as much as 93% of the maximum unconstrained potential gains at each tax level. The results
show that taxes do not play a significant role in any incremental gains to investors from diversifying internationally to achieve a
12
Table 5
Time-varying diversification benefits.
1993–2005 2006–2018
Overseas market constrained by … Overseas markets constrained by …
Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 Col. 8 Col. 9 Col. 10 Col. 11 Col. 12 Col. 13
Tax Rate 1/M wx1 wx2 wx5 wx10 nw 1/M wx1 wx2 wx5 wx10 nw
Panel A: MRRP delta

No Tax −0.6498 0.4956 0.7037 1.0265 1.4357 1.9246 0.1153 1.2173 1.4455 1.5943 1.7379 2.5815
(0.720) (0.836) (0.712) (0.598) (0.416) (0.390) (0.976) (0.426) (0.510) (0.458) (0.396) (0.160)
28% PIR −0.6617 0.2406 0.3410 0.5424 0.6936 0.9638 −0.4179 0.5296 0.5838 0.6841 0.7655 1.2122
(0.560) (0.834) (0.724) (0.570) (0.460) (0.410) (0.740) (0.584) (0.670) (0.558) (0.508) (0.232)

No Tax 0.2262 0.3185 0.3365 0.3581 0.3613 0.3658 0.1005 0.2367 0.2497 0.2667 0.2808 0.2857
(0.020) (0.002) (0.002) (0.002) (0.002) (0.002) (0.194) (0.002) (0.002) (0.002) (0.002) (0.002)
28% PIR 0.2253 0.3179 0.3360 0.3576 0.3608 0.3653 0.0995 0.2361 0.2492 0.2662 0.2802 0.2851
(0.020) (0.002) (0.002) (0.002) (0.002) (0.002) (0.206) (0.002) (0.002) (0.002) (0.002) (0.002)
Panel C: MVP delta

No Tax −0.6498 −0.3101 −0.2181 −0.0973 −0.0185 0.0119 0.1153 0.6257 0.3538 0.3991 0.5274 0.6467
(0.720) (0.780) (0.806) (0.862) (0.876) (0.914) (0.976) (0.734) (0.900) (0.850) (0.836) (0.732)
28% PIR −0.6617 −0.3046 −0.2533 −0.1988 −0.1542 −0.1422 −0.4179 0.1313 −0.0537 −0.0330 0.0255 0.0918
(0.560) (0.704) (0.718) (0.750) (0.776) (0.796) (0.740) (0.946) (0.906) (0.890) (0.944) (0.998)
This table reports the delta and epsilon benefits from diversification out of the New Zealand domestic market and into the MRRPs and the MVPs
optimized with no shorts and the international markets constrained by the 1993 market capitalization weights (period 1993–2005) and 2006 market
capitalization weights (period 2006–2018) at the one times (wx1), two times (wx2), five times (wx5), ten times (wx10) and no weight constraint
(nw) level. The domestic New Zealand market weight is unconstrained in the MRRP and MVP optimization solutions. The portfolios are optimized
with no taxes (No Tax), or with withholding taxes on overseas dividends and the local New Zealand FDR taxes calculated with the PIR set to 28%.
The results for the 1/M portfolio, which is the portfolio formed where each market weight is equal to the 1993 or 2006 market capitalization weight,
are also presented. The p-value results from the Ledoit and Wolf (2008) Sharpe ratio and volatility tests using a block size of 5 and 499 iterations are
presented in parentheses
minimum volatility portfolio.

Table 4, Panel C, reports the delta results for the various weight constrained and taxed MVPs. The results from the Ledoit and Wolf
(2008) Sharpe ratio tests are presented in parentheses. As with the MRRP, taxes reduce realized returns from overseas investment
which results in lower potential risk-adjusted improvements from diversification. With no taxes, none of the optimized MVPs have
positive Sharpe ratio improvements that are statistically significant. The unconstrained portfolio with no taxes offers the greatest
potential Sharpe ratio improvement of 14.29% over the domestic market. The p-value for this portfolio is 0.986, which is not
significant (Table 4, Panel C, Col. 7). In the case of a PIE with a 28% PIR, taxes on the unconstrained portfolio reduce the delta gains
to −13.54%, which is also not significant with a p-value of 0.758. Taxes at each level of weight constraint reduce the potential
Sharpe ratio gains. While MVPs with weight constraints and taxes offer significant potential volatility reducing benefits, the resulting
Sharpe ratios are not statistically significantly larger than that of the domestic market.
Based upon the use of the historical data for the period from 1993 to 2018, these results do not lend strong support for the
improved long term performance by weighting the New Zealand equity market according to its percentage of the global market
capitalization and thereby increasing allocations to global equities. While the Sharpe ratio does increase with diversification, un-
favorable tax treatments on overseas investments reduced the potential benefit to a point where it is not statistically significant.
Furthermore, the significance of the results is further weakened when one considers that a PIE optimizing a portfolio with imperfect
market return and covariance estimates is unlikely to achieve the efficient ex-post optimized results presented in this study.
5.2. Sub-period analysis
In this sub-section, the sample is split into two sub-periods between 1993-2005, and 2006–2018. The 1993–2005 sub-period was a
time period when the New Zealand dollar appreciated in value against the currencies of most developed markets. The New Zealand
performance was mixed in the 2006–2018 sub-period, which included the period of the global financial crisis. Table 5 presents the
results in Table 4 for the no-tax and 28% PIR tax rates over these two sub-periods. For the sub-period 2006–2018, the weight
constraints are based upon the 2006 market capitalization rates for the 34 markets considered in this study.
For both sub-periods, the results (Table 5, Panel A) show that taxes reduce the MRRP delta or diversification benefits from moving
from the New Zealand market portfolio to the MRRP. In the case of the 1/M⁎ portfolio, with no taxes the positive delta benefit is
49.41% for the period 1993–2005 and 120.91% for the period 2006–2018 (Table 5, Cols 3 and 9 respectively). At a 28% PIR, the
positive delta benefit reduces to 24.21% for the 1993–2005 period and 52.96% for the 2006–2018 period. We again observe that as
weight constraints are relaxed the potential benefits from diversification increase.
13
Table 5, Panel B, shows that taxes again pay no significant role during the sub-periods 1993–2005 and 2006–2018 in any
incremental gains to investors seeking to minimize portfolio volatility, when diversifying internationally outside the New Zealand
domestic market portfolio. Lastly, Table 5, Panel C, reports the delta results for the various weight constrained and taxed MVPs.
Similar to the results in Table 4, taxes at each weight constraint reduce the MVP delta for both sub-periods, and none of the MVP delta
results are statistically significant.
6. Simulation analysis
In further analysis we undertake stochastic simulations to test if our results are robust to different return patterns. For simplicity
we assume the global equity markets comprise New Zealand, Australia and a “world” market index (excluding NZ and Australia). This
is to reflect the different tax regimes that apply to investments in New Zealand equities, Australian equities and the other global
markets. Each equity market is assumed to have an expected pre-tax geometric total return denominated in NZD of 8% p.a. and an
annualised standard deviation of returns of 20%. Expected dividend yields are 5.50%, 4.50% and 3.0% p.a. respectively for New
Zealand, Australian and the world market index. This approximately corresponds to the historical dividend returns in New Zealand
dollars as observed over the period 1993–2018 for New Zealand, Australian and the remaining markets. For New Zealand Treasury
rates, we assume a geometric return of 2.0% p.a., and a standard deviation of 4.50% p.a.23
Monthly equity total returns are simulated for a period of 26 years equal to the time period in years between 1993 and 2018.
Equity returns are simulated using a standard Geometric Brownian Motion with delta t equal to one month.24 To examine how market
weights and benefits may change under different return correlations, simulations are performed with return correlations between
each equity market of 0.25 and 0.75. Quinn and Voth (2008) report over the period 1893–2005 correlations between equity markets
were roughly between 0.0 and 0.8, with a pronounced trend towards higher market correlations from the early 1990s. Similarly, Levy
and Levy (2014) report average pairwise correlations for 45 markets of between roughly 0.3 and 0.8 over the period 1980–2012.
Monthly Treasury returns are assumed to be independent of market returns. The process above is repeated to derive 1000 simulated
monthly equity and Treasury returns over a period of 26 years.
Table 6 summarises the total returns, dividend yield, volatility of returns and market correlation assumptions used in the si-
mulation analysis.
Table 7 reports the 50th percentile and mean of the optimal domestic market weight in MRRPs and MVPs optimized for the 1000
simulations with (i) no taxes, and (ii) non-resident withholding taxes on overseas dividends and the local New Zealand FDR taxes
calculated with the PIR set to 28%. In order to simulate the impact of varying levels of investors' awareness on the benefits of
international diversification, we employ two values for the correlation coefficient values of 0.25 and 0.75 between equity markets.
In Table 7, Panels A and B, the correlation coefficient between equity market returns is assumed to equal 0.25. The results in Panel
A and B show that for risk-return maximizing investors there is roughly an 8%–11% increase in the optimal domestic allocation from
no taxes to a 28% PIR. For example, consider the 1/M⁎ (wx1) portfolio. With no taxes the 50th percentile (mean) optimal portfolio
has a 49.7% (50.7%) allocation to domestic (New Zealand) equities (Table 7, Panels A and B, Col. 2). At the 28% PIR, the 50th
percentile (mean) optimal allocation to domestic equities increases to 60.3% (59.4%).
Table 7, Panels C and D report the results when the correlation coefficient between equity markets equals 0.75. As equity market
correlations increase, the diversification benefits from investing offshore reduce in the presence of market frictions. Here, the impact
of taxation magnifies the home bias with an increase in the optimal domestic allocation from no taxes to a 28% PIR of between
roughly 18% and 43%. To illustrate, for the 1/M⁎ portfolio with no taxes the 50th percentile (mean) optimal portfolio has a 52.8%
(51.9%) allocation to domestic equities (Table 7, Panel C and D, Col. 2). This compares to the 50th percentile (mean) domestic
allocation of 96.0% (70.2%) at the 28% PIR.
Similar to our result in Table 2, for investors who seek to minimize variance, no tax or tax at the 28% PIR have little impact on the
optimal domestic allocation. For example, consider the 1/M⁎ portfolio and the correlation between equity markets equal to 0.25. The
50th percentile (mean) optimal domestic allocation is 49.1% (49.2%) for no tax and 49.2% (49.2%) with the 28% PIR (Table 7,
Panels A and B, Col. 7). Almost identical results are reported in Table 7, Panels C and D, where investors also seek to minimize
variance and equity market return correlations equal 0.75.
Table 8 reports the delta and epsilon simulation results for the MRRPs and MVPs under different weight constraints, again with no
taxes and the 28% PIR. In Table 8, Panels A, B and C (where the correlation of returns between markets is 0.25) for all weight
constrained portfolios and the no weight constrained portfolio (Cols. 3 to 7), there are positive MRRP delta and MVP epsilon and delta
benefits. However, taxes reduce the potential benefits from diversification. Similar to the results using actual historical data, the 1/M⁎
portfolio captures a large proportion of the potential diversification benefits. For example, 47.7% (41.3%) of the diversification
benefit provided by the unconstrained MRRP portfolio is captured by the 1/M⁎ portfolio at no tax (28% PIR) (Table 8, Panel A, Col.
9).
In Table 8, Panels D, E and F the results when the correlation coefficient between markets is 0.75 are presented. The delta and
epsilon simulation results for the MRRPs and MVPs are still positive for the weight constrained and no weight constraint portfolio
23
MacDonald et al. (2019) drawing upon data sourced from the Dimson, Marsh and Staunton 2015 Global Investment Returns Database report
mean real returns (standard deviation) for Australasian equities of 8.34% (17.75%) and for international equities 7.70% (18.36%) over the historical
period 1900–2014. Over the same period the authors also report mean NZ bills returns (standard deviation) of 1.77% (4.62%).
24
See Hull (2018, Chapter 14).
14
Table 6
Return and volatility assumptions in simulations.
Country (i) wi Total return (% p.a.) Dividend yield (% p.a.) Standard deviation of return
Panel A - Equity returns and volatility

Australia 1.51 8.00% 4.50% 20.00%
NZ 0.19 8.00% 5.50% 20.00%
World index 98.30 8.00% 3.00% 20.00%
Panel B - Treasury returns and volatility

NZD Treasury return n.a. 2.00% 4.50%
The correlation of total equity returns between each market is assumed to be 0.25 and 0.75.
The correlation of total equity returns between each market and the Treasury rate is assumed to be 0.0.
Table 7
Optimal domestic allocations under 1000 simulations.
Domestic market weight
Col. 1 Col. 2a Col. 2b Col. 3a Col. 3b Col. 4a Col. 4b Col. 5a Col. 5b Col. 6a Col. 6b
Case wx1 wx1 wx2 wx2 wx5 wx5 wx10 wx10 nw nw
Correlation between equity markets = 0.25

Panel A – MRRP 50th Percentile (Mean)
No Tax 0.497 (0.507) 0.490 (0.500) 0.470 (0.482) 0.444 (0.453) 0.320 (0.344)
28% PIR 0.603 (0.594) 0.597 (0.587) 0.579 (0.567) 0.546 (0.536) 0.407 (0.428)
Panel B – MVP 50th Percentile (Mean)

No Tax 0.491 (0.492) 0.483 (0.484) 0.461 (0.461) 0.423 (0.424) 0.333 (0.333)
28% PIR 0.492 (0.492) 0.484 (0.485) 0.462 (0.462) 0.424 (0.424) 0.333 (0.333)

Panel C – MRRP 50th Percentile (Mean)
No Tax 0.528 (0.519) 0.523 (0.513) 0.504 (0.500) 0.489 (0.479) 0.204 (0.349)
28% PIR 0.960 (0.702) 0.951 (0.696) 0.918 (0.684) 0.845 (0.664) 0.583 (0.541)
Panel D – MVP 50th Percentile (Mean)

No Tax 0.488 (0.491) 0.481 (0.483) 0.459 (0.460) 0.421 (0.423) 0.334 (0.332)
28% PIR 0.490 (0.493) 0.483 (0.485) 0.461 (0.463) 0.423 (0.425) 0.335 (0.334)
This table reports the 50th percentile and mean (in parentheses) results of 1000 simulations for the domestic New Zealand market allocation held in
each MRRP and MVP optimized with different tax levels and allocation weight constraints. The portfolios do not implement short sales and the
domestic New Zealand market weight is unconstrained in the optimization solutions. Each overseas market weight is constrained by the starting
market capitalization weight at the one times (wx1), two times (wx2), five times (wx5), ten times (wx10) and no weight constraint (nw) level. The
portfolios are optimized with no taxes (No Tax), or with non-resident withholding taxes on overseas dividends and the local New Zealand FDR taxes
calculated with the PIR set to 28%.
(Cols. 3 to 7), except for MVP delta benefits which are negative when the tax rate is 28% PIR (Table 8, Panel F, Cols. 3 to 7). At high
equity market correlations, the 1/M⁎ portfolio also captures a large proportion of the potential total diversification benefits (Table 8,
Panels D, E and F).
7. Further robustness tests
7.1. Impact of taxes on portfolio allocation under a no dividend imputation and no FIF tax regime
As discussed earlier, the differential tax treatment regime in New Zealand is quite unique. In order to quantify the effect of taxes
on other tax regimes, we consider a hypothetical case where domestic New Zealand and offshore investments are taxed equally.
Specifically, we assume that New Zealand dividends have no imputation credits attached, with all New Zealand dividends subject to
either no tax or tax at the investors PIR tax rate between 0% and 28%. We also assume that there is also no FIF regime, with dividends
from offshore markets (including the Australian market) also subject to either no tax or tax at the investors' PIR on dividend income
only between 0% and 28%.25 Other than the no tax case, offshore dividends still attract NRWT at the deemed rate under any
applicable DTA agreement with New Zealand. There is no capital gains tax on either New Zealand or offshore shares. Overall, the
25
Prior to the introduction of dividend imputation in New Zealand, some New Zealand companies paid dividends out of capital reserves or capital
profits. These dividends were tax-free to New Zealand shareholders. For simplicity we ignore these dividends and assume that pre-imputation all
dividends are taxable at the investors' marginal rate.
15
Table 8
Diversification benefits under 1000 simulations.
Overseas market weights constrained by… Max. benefit captured
Col. 1 Col. 2 Col. 3 Col. 4 Col. 5 Col. 6 Col. 7 Col. 8 Col. 9 Col. 10 Col. 11 Col. 12
Tax Rate 1/M wx1 wx2 wx5 wx10 nw 1/M wx1 wx2 wx5 wx10

Panel A: MRRP delta
No Tax −0.236 0.287 0.307 0.368 0.459 0.601 −0.392 0.477 0.510 0.612 0.764
28% PIR −0.437 0.171 0.186 0.238 0.317 0.414 −1.054 0.413 0.448 0.575 0.765

No Tax 0.015 0.218 0.225 0.243 0.269 0.295 0.049 0.739 0.763 0.825 0.912
28% PIR 0.013 0.217 0.224 0.243 0.268 0.294 0.046 0.738 0.762 0.825 0.911
Panel C: MVP delta

No Tax −0.236 0.277 0.291 0.347 0.415 0.464 −0.508 0.597 0.627 0.747 0.894
28% PIR −0.437 0.148 0.151 0.187 0.219 0.248 −1.762 0.597 0.610 0.753 0.884

Panel D: MRRP delta
No Tax −0.098 0.064 0.078 0.090 0.121 0.166 −0.590 0.386 0.472 0.542 0.730
28% PIR −0.298 0.010 0.016 0.029 0.039 0.047 −6.386 0.204 0.342 0.622 0.833
Panel E: MVP epsilon

No Tax 0.006 0.068 0.070 0.075 0.081 0.089 0.069 0.761 0.782 0.843 0.910
28% PIR 0.005 0.067 0.069 0.074 0.080 0.089 0.056 0.759 0.780 0.841 0.909
Panel F: MVP delta

No Tax −0.098 0.071 0.070 0.070 0.080 0.093 −1.054 0.761 0.757 0.751 0.864
28% PIR −0.298 −0.055 −0.054 −0.048 −0.051 −0.054 5.490 1.015 0.989 0.894 0.948
This table reports summary results of 1000 simulations of the 50th percentile delta and epsilon benefits from diversification out of the New Zealand
domestic market and into the MRRPs and the MVPs optimized with no shorts and the international markets constrained by the starting capitalization
weights at the one times (wx1), two times (wx2), five times (wx5), ten times (wx10) and no weight constraint (nw) level. The domestic New Zealand
market weight is unconstrained in the MRRP and MVP optimization solutions. The portfolios are optimized with no taxes (No Tax), or with
withholding taxes on overseas dividends and the local New Zealand FDR taxes calculated with the PIR set to 28%. The results for the 1/M portfolio,
which is the portfolio formed where each market weight is equal to the 1993 market capitalization weight, are also presented. The last five columns
report the percent of the unconstrained portfolio benefits that are captured by each weight constrained portfolio.
Table 9
Optimal domestic allocations under NZ taxation regime pre-dividend imputation and pre-the FIF regime.
No Tax 0.4779 0.3894 0.3528 0.2945 0.2503 0.3443 0.3177 0.2875 0.2712 0.2699
0% PIR 0.4926 0.4073 0.3678 0.3292 0.2939 0.3439 0.3171 0.2874 0.2711 0.2694
10.5% PIR 0.4826 0.3738 0.3385 0.2833 0.2454 0.3438 0.3169 0.2871 0.2707 0.2690
17.5% PIR 0.4729 0.3516 0.3172 0.2612 0.2191 0.3437 0.3166 0.2868 0.2705 0.2686
28% PIR 0.4660 0.3281 0.2959 0.2414 0.1974 0.3435 0.3162 0.2866 0.2703 0.2681
withholding taxes on overseas dividends. The local New Zealand taxes on dividends from both New Zealand and offshore markets calculated at rates
0%, 10.5%, 17.5% or 28%. New Zealand dividends are assumed to have no attached imputation credits.
impact of taxation on optimal portfolio allocation decisions under this tax environment is of interest where apart from tax frictions
that may arise from NRWT, the taxation regime for domestic and offshore equities is similar.
Table 9 reports the optimal domestic New Zealand market weights in MRRPs and MVPs optimized with no taxes (No Tax), and
16
Table 10
Optimal domestic allocations with markets of New Zealand United States, United Kingdom and Japan only.
No Tax 0.4485 0.4210 0.4210 0.4210 0.4210 0.3606 0.3434 0.3341 0.3341 0.3341
0% PIR 0.4485 0.4388 0.4388 0.4388 0.4388 0.3606 0.3434 0.3342 0.3342 0.3342
10.5% PIR 0.4531 0.4531 0.4531 0.4531 0.4531 0.3609 0.3436 0.3344 0.3344 0.3344
17.5% PIR 0.4758 0.4758 0.4758 0.4758 0.4758 0.3610 0.3438 0.3346 0.3346 0.3346
28% PIR 0.5123 0.5123 0.5123 0.5123 0.5123 0.3613 0.3441 0.3349 0.3349 0.3349
with NRWT on overseas dividends and the PIR on dividend income only set to 0%, 10.5%, 17.5% or 28%.26 Overall, the tax
advantages of investing in the New Zealand home market now decrease with the investor's PIR under a tax regime with no dividend
imputation and the FIF regime. For example, at the weighting times two (wx2) portfolio (Table 9, Col. 3) and a PIR of 0% the optimal
home market allocation is 40.73%. However, at the PIR of 28%, the optimal home market allocation decreases to 32.81%. A similar
trend is observed when weighting constraints are relaxed and for the wx1 and nw constraint.
The explanation for this shift or tilt to offshore investment in the presence to taxation is as follows. Over the time period of the study
New Zealand was a “high dividend” yield market. More specifically, the dividend yield on the New Zealand market was 5.40% per annum
in New Zealand dollar terms, which was the highest dividend yield of all 34 markets in this study (see Table 1). Thus, subject to being able
to utilize NRWT deducted on offshore dividends to offset any New Zealand tax payable, under the New Zealand tax regime prior to
dividend imputation investors would favour markets with low dividend yields and a higher component of capital gains in total returns.
Overall, these results suggest that taxes are still an important factor to consider for portfolio optimization even when domestic and
offshore investments are taxed similarly. For example, under the New Zealand tax regime prior to the introduction dividend im-
putation and differential taxation, investors in high tax brackets would rationally shift up to 6% of their investments towards the
overseas market with lower dividend yields.
7.2. Impact of taxes on portfolio allocation when offshore markets restricted to US, Japan and UK only
We also examine the sensitivity of our results to an international portfolio restricted to the New Zealand market and the major
world markets of the US, Japan and the UK. We assume the dividend imputation and the FIF tax regime applies. Table 10 reports the
optimal domestic New Zealand market weights in MRRPs and MVPs optimized with no taxes (No Tax), or with non-resident with-
holding taxes on overseas dividends and the local New Zealand FDR taxes calculated with the PIR set to 0%, 10.5%, 17.5% or 28%.27
The results are qualitatively similar to those presented in Table 2, where New Zealand and offshore market weights were optimized
across 34 markets. The results broadly suggest under the New Zealand tax regime with dividend imputation and the FIF regime,
differential taxation can still account for a circa 6% - 9% shift to the domestic market compared to the offshore market. For example,
at the weighting times one (wx1) portfolio (Table 10, Col. 2) and a PIR of 0% the optimal home market allocation is 44.85%.
However, at the PIR of 28%, the optimal home market allocation increases to 51.22%. With the investment set constrained to the New
Zealand market and three largest offshore markets only, the offshore weight constraints proportional to their relative sizes are not
binding beyond the wx2 portfolios resulting in the same optimized portfolios as the nw case.
8. Impact of transaction costs
While the focus of this study is the impact of taxes, there are other costs to international diversification that investors should
26
For the sake of brevity, we only report these results. The results for (i) the after-tax excess geometric annual returns, standard deviation of
returns, Sharpe ratios, and Sortino ratios for the MRRPs and MVPs optimized with different weight constraints and tax levels, along with the 1/M
portfolio and the domestic New Zealand market portfolio; and (ii) the delta and epsilon results for the MRRPs and MVPs with various weight
constraints and tax rates are provided in an Online Appendix or available from the corresponding author on request. These results generally show
increase in after-tax excess returns and the Sharpe and Sortino ratios as the tax rate increases, which reflects a lower average effective tax rate on
equity investments compared to tax on the risk-free asset. Also taxes do not play a significant role in any incremental gains to investors from
diversifying offshore to achieve a minimum volatility portfolio.
27
The Online Appendix reports the after-tax excess geometric annual returns, standard deviation of returns, Sharpe ratios, and Sortino ratios for
the MRRPs and MVPs optimized with different weight constraints and tax levels and delta and epsilon results for the MRRPs and MVPs with various
weight constraints and tax rates. The results are qualitatively similar to the results already reported in Tables 3 and 4.
17
consider. The incremental costs to investment offshore can include transactions costs, exposure to exchange rate risk, cost of in-
formation asymmetries, restrictions on cross-border remittances and other costs such as custodial costs to hold investments offshore.
However, with greater access to the internet, real-time information and the adoption of electronic trading systems, transaction costs
may be expected to continue to reduce over time.28 For example, Cooper and Kaplanis (1994) estimate that the extra fees for foreign
investments by US funds in 1989 were 0.68% per year while Levy and Levy (2014) estimated the extra foreign investment fees at the
time of their study at about 0.22% only. Widespread adoption of international financial reporting standards should also improve the
quality of information and reduce information asymmetry (Iatridis, 2010; Daske and Gebhardt, 2006).
Early research by Cooper and Kaplanis (1994) finds home bias cannot be explained by inflation hedging or by the direct ob-
servable costs of offshore investment unless investors have very low risk aversion. Similarly, Coën (2001) find that deadweight costs
to explain home bias are too high compared to actual existing costs. A related study by Chan et al. (2005) find less home bias when
the domestic market is more developed, has a greater market capitalization and lower transaction costs. This is due to greater market
participation by foreign investors and less investment by domestic investors. Oehler et al. (2017) also note that the findings of greater
market integration may contradict any assumption that transaction costs contribute to any significant home-biased investment be-
haviour. On the other hand, Levy and Levy (2014) argue that with greater market integration and increased correlations between
markets, lower domestic transaction costs and exchange-rate risk for domestic investment can still justify the observed home bias.
This is despite the observation by Levy and Levy (2014) that transaction costs have now substantially reduced.
Apart from the optimization constraints based on relative market caps that indirectly reflect higher costs of investing in small
markets, we do not attempt to directly measure and deduct transaction costs from portfolio returns.29 Liquidity constraints and
bid–ask spreads are likely to vary amongst the 21 developed and 13 emerging markets used in this study. Overall the impact of
transaction and other related costs on our results are unclear from the perspective of investors in a small developed market such as
New Zealand. For instance, while domestic investors in small capital markets may still face greater information asymmetries and
exchange rate risk to invest offshore, transaction costs such as bid-ask spreads and market impact costs are likely to be lower in larger
and more liquid offshore markets. In summary, the effect of varying transaction, exchange rate and other related costs on home bias
in a small capital market will be of interest in a future study.
9. Conclusion
This paper investigates the potential effect of taxes on the benefits from international diversification and the efficiency of an
equity home bias for New Zealand investors with indirect equity investments held in a PIE under the current dividend imputation and
FDR tax regime. We find that taxes increase the equity home bias for mean-variance optimizing investors as they favour domestic
investments and reduce the potential benefits of international diversification. For various weight constraints imposed on the max-
imum level of investment in any offshore market, the allocation to New Zealand domestic equities would increase by 7%–11% for the
MRRP optimized with a PIR of 28% compared to the no assumed taxes case using historical data between 1993 and 2018. In
comparison, investors who wish to minimize the portfolio variance are not affected by taxes.
The findings presented in this paper contribute to the literature investigating the benefits of international diversification for New
Zealand investors. First, the results provide evidence of the importance of taxes in reducing the benefits from diversifying away from
New Zealand equities to international equities for KiwiSaver investors which prior literature has not specifically examined. This study
reports that the differential taxation rates applied to international equity investment returns are an important factor to consider and
demonstrated that internationally diversified KiwiSaver portfolios may not provide statistically significant positive Sharpe ratio
improvements versus the untaxed domestic New Zealand market portfolio on an after-tax basis. Second, the empirical results con-
tribute to the prior literature that has identified tax as an explanatory factor in the documented home bias that exists with inter-
national investment funds in New Zealand (e.g., Chan, Covrig and Ng). We estimate that about a 7%–11% tilt towards the domestic
New Zealand market can be justified purely due to taxes. The results presented here find that the tax effect is more pronounced when
optimizing a portfolio to maximize the excess after-tax return-to-risk performance than when minimizing portfolio variance.
The results of our simulation analysis also provide evidence of a similar home bias tilt to New Zealand equities of between 8% and
11% when equity market correlations are relatively low at 0.25. At higher New Zealand equity market correlations an even greater
tilt to New Zealand domestic equities could be justified. The results support the evidence of Levy and Levy (2014) that market
frictions such as taxation will have a greater impact on home bias propensity when market correlations increase.
While we conclude that taxes can rationally contribute to investor home bias for New Zealand tax residents under the current
dividend imputation and FDR regime, some caution must still be exercised in the interpretation of our results. We find that under a
tax regime with no capital gains tax and no significant differential between the taxation of dividends on domestic or offshore equities,
the presence of taxes over our sample period would in fact decrease any equity home bias for domestic investment in New Zealand by
1%–6% for the MRRP optimized with a PIR of 28% compared to the no taxes case. Over the time period of this study the New Zealand
28
Pradhan et al. (2014) finds evidence that for G-20 countries greater access to information technologies is associated with higher economic
growth.
29
To incorporate transaction costs into performance, Jacobs et al. (2014) measures performance from a European investor's perspective and uses a
40 basis point per transaction bid–ask spread to calculate out-of-sample Sharpe ratios after transaction costs. The focus of their study is to examine if
simple heuristic portfolio allocation rules provide similar diversification benefits to more theoretical approaches to portfolio optimisation. The
authors find that transaction costs tend to favour the simpler heuristic models.
18
market exhibited the highest dividend yield across the 34 markets examined and dividends contributed to a significant component of
the total domestic return. Thus, New Zealand investors subject to tax on domestic and offshore dividends at the same marginal tax
rate were attracted to offshore markets that have a lower dividend yield and higher capital gain yield versus the New Zealand market.
This is to maximize total after-tax returns from dividends and capital gains.
It should also be noted that the results presented in this paper reflect the historical benefits from diversification, when the New
Zealand currency strengthened against other currencies over the sample period. This study does not consider the effect of the large
forecast increase in managed funds in KiwiSaver accounts (e.g., Heuser et al., 2015) and the impact on New Zealand financial markets
if all these funds were invested domestically. Further research that addresses these points and considers the impact of differential
taxation of domestic and offshore investment on home bias is warranted.
Acknowledgements
We acknowledge the helpful comments of Craig Elliffe, Stephen Taylor, Cameron Truong, Nhut Nguyen, Tom Smith, Robert
Bianchi, and participants at the Financial Markets and Corporate Governance Conference in Sydney 2019, the AFAANZ conference in
Brisbane 2019 and the FMA International Asia-Pacific Finance conference in Vietnam 2019. We thank the anonymous referee and
editor for their insightful comments and constructive suggestions. This paper is based upon a chapter from the PhD thesis by Shaun
McDowell. Any errors are the responsibility of the authors.
Appendix 1
Calculating the FDR for a PIE
The New Zealand Income Tax Act 2007 sets out the rules governing tax obligations in New Zealand. Section EX 53 defines the
rules used by a unit-valuing fund, such as a unit trust, to calculate tax owed by unit holders using the FDR method. The unit trust
assigns each investor an interest (the unit) in a proportion of the net returns from the investments30 and determines the value of the
investors' units for each of a number of periods making up the income year.31 The FDR amount is pro-rated by the number of trading
days in the period being measured compared to the number of days in the tax year. The value of the PIE unit at the beginning of the
period is multiplied by the pro-rated FDR, and this amount is treated as income to be taxed at the clients' PIR to determine the tax
owed.
PIEs generally calculate a daily closing value for existing units. The pro-rated annual FDR of 5% for a single trading day is 1/365
of the FDR. If the unit price at the end of the previous closing is $10.00, and a client has a PIR of 28% and holds 1000 units, the unit-
trust can calculate the tax owed by the client at the end of the trading day as follows:
1 /365 × 0.05 × 1000 × $10.00 × 0.28 = $0.383562 (A.1)
For this study, the monthly after-tax portfolio returns are determined using a monthly pro-rated FDR equal to 1/12 × 0.05.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.pacfin.2020.101375.
References
EY, 2019. Worldwide Corporate Tax Guide 2019. https://www.ey.com/Publication/vwLUAssets/ey-worldwide-corporate-tax-guide-2019/$FILE/ey-worldwide-

corporate-tax-guide-2019.pdf.
Ainsworth, A., 2016. Dividend imputation: the international experience. JASSA 1, 58–63.
Ardalan, K., 2019. Equity home bias: a review essay. J. Econ. Surv. 33 (3), 949–967.
Bauer, R., Otten, R., Tourani-Rad, A., 2006. New Zealand mutual funds: measuring performance and persistence in performance. Account. Financ. 46 (3), 347–363.
Baxter, M., Jermann, U.J., 1997. The international diversification puzzle is worse than you think. Am. Econ. Rev. 87, 170–180.
Chan, K., Covrig, V., Ng, L., 2005. What determines the domestic bias and foreign bias? Evidence for mutual fund equity allocations worldwide. J. Financ. 60 (3),
1495–1534.
Christoffersen, S., Geczy, C., Musto, D., Reed, A., 2005. Crossborder dividend taxation and the preferences of taxable and nontaxable investors: evidence from Canada.
J. Financ. Econ. 78 (1), 121–144.
Cliffe, C., Marsden, A., 1992. The effect of dividend imputation on company financing decisions and cost of capital in New Zealand. Pac. Account. Rev. 4, 1–30.
Coën, A., 2001. Home bias and international capital asset pricing model with human capital. Journal multinational. Financ. Manag. 11 (4), 497–513.
Constantinides, G.M., 1984. Optimal stock trading with personal taxes: implications for prices and the abnormal January returns. J. Financ. Econ. 13 (1), 65–89.
Cooper, I., Kaplanis, E., 1994. Home bias in equity portfolios, inflation hedging, and international capital market equilibrium. Rev. Financ. Stud. 7 (1), 45–60.
Dai, Z., Maydew, E., Shackelford, D., Zhang, H., 2008. Capital gains taxes and asset prices: capitalization or lock-in? J. Financ. 63 (2), 709–742.
Daske, H., Gebhardt, G., 2006. 2006, International Financial Reporting Standards and Experts’ Perceptions of Disclosure Quality, Abacus, 42, Nos 3/4.
Elliffe, C., 2015. International and Cross-Border Taxation in New Zealand. Thomson Reuters, Wellington New Zealand.
Errunza, V., Losq, E., 1985. International asset pricing under mild segmentation: theory and test. J. Financ. 40, 105–124.
30
EX53, 1(b)(ii) of the New Zealand Income Tax Act 2007.
31
EX53, 1(b)(iii) of the New Zealand Income Tax Act 2007.
19
French, K., Poterba, J., 1991. Investor diversification and international equity markets. American. Econ. Rev. 81 (2), 222–226.
Frijns, B., Tourani-Rad, A., 2015. On the performance of KiwiSaver funds. Pac. Account. Rev. 27 (3), 266–281.
Gehrig, T., 1993. An information based explanation of the domestic bias in international equity investment. Scand. J. Econ. 95 (1), 97–109.
Heuser, A., Kwok, J., Snethlage, D., Watts, D., 2015. Review of the KiwiSaver fund manager market dynamics and allocation of assets, The Treasury, new Government.
http://www.treasury.govt.nz/publications/research-policy/tp/kiwisaver.
Hull, J., 2018. Options, Futures and Other Derivatives. Ninth Global Edition, Pearson Education.
Iatridis, G., 2010. International Financial Reporting Standards and the quality of financial statement information International Review of Financial Analysis. 19. pp.
193–204.
Jacobs, H., Müller, S., Weber, M., 2014. How should individual investors diversify? An empirical evaluation of alternative asset allocation policies. J. Financ. Markets
19, 62–85.
Karolyi, A., Stulz, R., 2003. Are financial assets priced locally or globally? In: Handbook of the Economics of Finance, pp. 975–1020.
Kho, B.-C., Stulz, R.M., Warnock, F.E., 2009. Financial globalization, governance, and the evolution of the home bias. J. Account. Res. 47, 597–635.
Lally, M., 2000. The Real Cost of Capital in New Zealand: Is it Too High? New Zealand Business Roundtable, Wellington.
Lally, M., 2003. Investing in New Zealand: A Review. Victoria University of Wellington, Working paper.
Ledoit, O., Wolf, M., 2008. Robust performance hypothesis testing with the Sharpe ratio. J. Empir. Financ. 15 (5), 850–859.
Levy, H., Levy, L., 2014. The home bias is here to stay. J. Bank. Financ. 47, 29–40.
Lewis, K., 1999. Trying to explain home bias in equities and consumption. J. Econ. Lit. 37 (2), 571–608.
Lintner, J., 1965. The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. Rev. Econ. Stat. 47 (1), 13–37.
MacDonald, K.L., Bianchi, R.J., Drew, M.E., 2019. Equity risk versus retirement adequacy: asset allocation solutions for KiwiSaver. Account. Finance. https://doi.org/
10.1111/acfi.12481.
Markowitz, H., 1952. Mean-variance analysis in portfolio choice and capital markets. J. Financ. 7 (1), 77–91.
Mishra, A., Ratti, R., 2013. Home bias and cross border taxation. J. Int. Money Financ. 32, 169–193.
Oehler, A., Wendt, S., Horn, M., 2017. Are investors really home-biased when investing at home? Res. Int. Bus. Financ. 40, 52–60.
Officer, R., 1994. The cost of capital of a company under an imputation tax system. Account. Financ. 34 (1), 1–17.
O’Hagan-Luff, M., Berrill, J., 2015. Why stay-at-home investing makes sense. Int. Rev. Financ. Anal. 38, 1–14.
Pesenti, P.A., Van Wincoop, E., 2002. Can nontradables generate substantial home bias? J. Money Credit Banking 34 (1), 25–50. Project MUSE. https://doi.org/10.
1353/mcb.2002.0034.
Pradhan, R., Arvin, M., Norman, N., Bele, S., 2014. Economic growth and the development of telecommunications infrastructure in the G-20 countries: a panel-VAR
approach. Telecommun. Policy 38 (7), 634–649.
Quinn, D.P., Voth, H., 2008. A century of global equity market correlations. Am. Econ. Rev. 98 (2), 535–540.
Sercu, P., 1980. A generalization of the international asset pricing model. Revue de l'Association Française de Finance 1 (1), 91–135.
Sercu, P. and R. Vanpee, 2007, Home bias in international equity portfolios: a review, Available at SRN: https://ssrn.com/abstract=1025806 or doi:https://doi.org/
10.2139/ssrn.1025806.
Sharpe, W., 1964. Capital asset prices: a theory of market equilibrium under the condition of risk. J. Financ. 19, 425–442.
Solnik, B., 1974. An equilibrium model of the international capital market. J. Econ. Theory 8 (4), 500–524.
Sortino, F., Meer, R.V.D., 1991. Downside risk: capturing what’s at stake. J. Portfolio Manag. Summer, 27–31.
Warren, G., 2010. Equity home bias in Australian superannuation funds. Aust. J. Manag. 35 (1), 69–93.
20

Home Bias

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Home Bias

Uploaded by

Copyright:

Available Formats

Pacific-Basin Finance Journal 62 (2020) 101375

Contents lists available at ScienceDirect

Pacific-Basin Finance Journal

The potential effect of taxes on the equity home bias in New

ARTICLE INFO ABSTRACT

E-mail addresses: shaun.mcdowell@waikato.ac.nz (S. McDowell), byong.lee@auckland.ac.nz (J.B. Lee),

2.1. Overview of KiwiSaver funds in New Zealand and performance

2.2. Impact of taxes on portfolio asset allocation decisions

3.1. Portfolio investment entity (PIE)

3.2. Taxation of New Zealand income

3.4. Foreign investment fund (FIF)

3.5. Taxation of dividends and capital gains from Australian shares

Panel A: Developed markets

Panel B: Emerging markets

4. Data and Methods

4.2. International diversification

4.3. Estimated returns

4.4. Measuring the risk-adjusted performance

4.5. Measuring the potential benefits of international diversification

Fig. 1. 1/M versus N.Z. (1993–2018).

5.1. Full period analysis

Case wx1 wx2 wx5 wx10 nw wx1 wx2 wx5 wx10 nw

Panel A: Excess return

Panel B: Standard deviation

Panel C: Sharpe ratio

Panel D: Sortino ratio

Panel A: MRRP delta

Panel B: MVP epsilon

Panel C: MVP delta

Overseas market constrained by … Overseas markets constrained by …

Panel A: MRRP delta

Panel B: MVP epsilon

Panel C: MVP delta

minimum volatility portfolio.

5.2. Sub-period analysis

Panel A - Equity returns and volatility

Panel B - Treasury returns and volatility

Case wx1 wx1 wx2 wx2 wx5 wx5 wx10 wx10 nw nw

Correlation between equity markets = 0.25

Panel B – MVP 50th Percentile (Mean)

Correlation between equity markets = 0.75

Panel D – MVP 50th Percentile (Mean)

7. Further robustness tests

Correlation between equity markets = 0.25

Panel B: MVP epsilon

Panel C: MVP delta

Correlation between equity markets = 0.75

Panel E: MVP epsilon

Panel F: MVP delta

Case wx1 wx2 wx5 wx10 nw wx1 wx2 wx5 wx10 nw

Case wx1 wx2 wx5 wx10 nw wx1 wx2 wx5 wx10 nw

8. Impact of transaction costs

Calculating the FDR for a PIE

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.pacfin.2020.101375.

EY, 2019. Worldwide Corporate Tax Guide 2019. https://www.ey.com/Publication/vwLUAssets/ey-worldwide-corporate-tax-guide-2019/$FILE/ey-worldwide-

You might also like