] MICHIGAN BUSINESS SCHOOL From the Department of FINANCE Product Number wp03.27 Released November 11, 2003 rurKu31NI8Nliulml;................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................. "O O, a, 10, 'VO U............. 10....................................................................... -,- - -,- - - -,- -,............................................................................................ -, ---, ----.................................................................................................................................................................................................................................................................................................................................................................................................................................................................. I'll.................................................................................................................................................... RHI'IDVISTU D EN T GF INAN...................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................................... IC........................................... g n............................................................................................................ -, ---, ----................................................................................................... -, ---, ---l..................................... " I'll"....... mth.......................................................................................................................................................................... Iling.......... -, ---, ----................................................................................................... fo r................................................................................................................................................... me n vvvvv ion ind i's I n ign........................................................................................................................................................ More patriotic countries and more patriot regions within the U.S. hold smaller forei equity positions. Increased patriotism is negatively related to foreign holdings groi The patriotism e.ect remains after contro for endogeneity and is robust to controls the four standard explanations for the ho bias: transaction barriers, information, familiarity, and risk. Patriotism explains i additional 7% of the cross-country variat of foreign holdings. Economic, political, a social correlates do not explain patriotism e.ect on the home bias. A 10% decrease patriotism is associated with a rise in forn equity holdings of 2-4% of total country holdings, a 29-48% increase. LEADING IN THOUGHT AND ACTION

Patriotism in Your Portfolio Adair Morse and Sophie Shive* Nove:mber 11, 2003 (Veterans' Day) ABSTRACT More patriotic countries and more patriotic regions within the i.S. hold smaller ioreign equity positions. Increased patr;iot;ismr is negatively relatedi to foreigrn hold.ings growth. The patriotisim effect remains after contr'o]iirig for endogeneit:y and is robust t;o contri;ols for the fhur stoaldardi expianatkions lir thoe home bias: t;ransac-.ion barriers, in for imal on.L, [fa.miiarity, and risk. Patirloilsro explains an. addilt:io 1al 7%5 of the cross-country variation of foreign holdings. Economic, political, and social correlates do not explain [pat:[iot:ism.'s eflect o} thie 1ho 'o me bias. A 1l)% decrease in patriotism is associated with a rise in foreign equity holdings of 2-14 of total country holdings, a 29-48% increase. 'Doctoral Students, University of Michigan Business School. Department of autthor: Adair Morse, 701 a'ppan Street, Ann Arbor, MI 48109-1234. ITe 647-8133 E-mail adairm(limuich.edu. We wish to thank Nicholas Barberis, C(ar, L:auren (Coben Artyorn Durnev, -[fernan Kamil, Linda Iesar, Neja t [' ir Shiumnway, Luigi Ziigales, and seminar participants at the University of of Econormics and FPilarce for their val.able commenlts. Finance. Corresponding 734 -973-7791 Fa['x. 734 -(;eert Belaert, Anusha Sey iuni Paige )uimtet, Michilga's Departmrent.s

Why do people bet on home sports teams despite unfavorable odds? Why do employees allocate retirement savings to their employer's stock despite having their human capital invested inr the firmr? And why do investors over-weiht their portfolios with domestic stocks? In 'he Economics of Discrimznation, Becker l(957) establishes that discriminatloo bears a cost which the free-market will be unwilling to pay. iln this paper we are c:o:ncerneed not, asi Be rcker was. with discrim in ati o: n againsi peop le, but di scriinlation toward home entities. In particular, investors discriminate in favor of domestic stocks in their portfolios. The dispr"portionately large allocation of a country's aggregate equity investment to domestic assets is known as the equity home bias. French and Poterba (1991); ITesar and Wlerner (1995), a:nd Pastor (2000) empririca]ly doclument that i:nvestolr:s' domesltic holdings are quite high relative to an optimally diversified portfolio.1 'he CAPlM predicts that an investor will hold the world market portfolio of all countries in proportion to each cC)ntrLys n rarket capitalizaltio: (Lewis, 1]999). T'aking Jntao accouunt returns io humans capita.l which generally follows the country's fortunes, investors should invest even less in their homei country (Baxter and Jermallnn, 1997). Coiuntry portfolios with sm:nal domrestkic hoildin:s are, howvever, simply not observed. It its not surprisintg that Ob — stfeld and i.ogoff (2)000) categorize the home bias as one of "the six major international macroeconomrics puzzles". The final'ce and economics 'iteratures give four principal explsanationis for the home bias: the ptre-sence of transactions barriers. irnornimaltion asynrmetries between local and foreign investors, familiarity bias and unaccounted for risk or uncertainty. A simple explanation as n as been omtted. Could it be that investors' sentuie:rLntet toward th.eir coL:n:[try exp;)lai.n[s par:t of the equ(ity horne bias? Consider again the bettor choosing her team, and the employee choosing a pension allocation. Ptresu:rnaib,]' a bettor de.rives utility from.a ilhorne victory, a.nd an. ermn ployee enjoys seeing her firm succeed. if the individual is risk averse; portfolio theory would then require that the bettor put money on the opposing team, aind the emnployee short the com[pay stock. Tis hedJtg is rarely observed, and this may be due solely to lack of ireinch and Poterba calculate that U.S., Japanese and UJ. K. investors held 94%, 98% anld 82% of their portfolios in domestic assets. 1

fam.iiaiity with the opposing team or other firms, which generates the ambiguity aversion described in the psychology literature. Loyalty, however, seems like a more plausible cause: to some, the tiho Ihti 0l of betting against a holm e teanm or shorting emlployyer stock may seem disloyal, and to others, the home team and the company stock are simply perceived as the best bet. The same patriotism-induced loyalty can be considered for inter:a ational diver si:fi cat:io dne.isions. The economic significance of patriotism has not previously been examined, even though anecdotal evidence shows tia't atriotic investing is not a new phenomenon. In the United States, the promoti:tion of govern:aiment boinlds uinJ]g patr iotic:rhetoric dates back at least to the Civil War, when war bonds served to finance the Union effort. W'ar bonds are still used tod ay. After the Septemrlbe[r I] ith, 2001 terrorist atta(ks, Se ie-s I arnd EE Treasury Bonds were renanmed 'Patriot Bonds', and their sales rose to $6.6 billion, a 43% increase over the previous year (Sulon 2001). Patriotic investing is not restricted to governme:n]t secu ities. Af:ter September I 1., 2001, Ithe.S. marketss remaine]ed closed for a a week and mc speculation ensued as to how far the indices would drop. MVass mnedia encouraged patriots of the country to hold on to their stocks.2 The notion that investors mnight heed calls to- p:latriotic: behavior has a clear ri:p ic ation; it seems thai patriotism may affect markets. This paper explores the role of patriotism in explaining the equity home bias. Our hypothesis is that more patriotkic invelstors choose to invest more of their po-,rtfolio at home. Since the effect of patriotism is not in conflict with the existing explanations of the home hias, our aim is to assess wlether patriotismn cean signiificant]y explain variat)on in foreign equity holdings after controlling for thle effects of existing theories. VW use a patriotism score from the University of Michigam's WVorld Values Survey. The survey asks i:adividuals in 53 cou:lltries whether they are proud to be a national 2 The Boston Globe related a story of an individual who "wonderesd what would happen 'if eiceru red blooded Amrerzcan... borught a?few sha?rs of their favrorite stock o; Monday" (September 18, 2001). A Novemtber 2001 article in Monzey criticizes the call for a 'patriot rally' to prop up the market on the day it re-opened after September 11, 2001 (Frederick, 2001). In a November, 2001 %Spectremn survey of affluent, investors, 52% of respondents sa. t hey,would show their pat, otism by making investments in U.S. colnmparnies.

of their country. Using the country average of responses to this question as a measure of patriotism, we show that patriotism is significant in explailing the weigjht' of foreign equit]ies in tihe cotuntry s C)tota equity holdingi]s. Iin fu:rt hier testsa fi.rit difference -cross survey years shows that greater increases in patriotism are accompanied by lower positive changes in foreign holdings. We find that higher patriotism scores are associated with lower fore: n holdi:ings in: the nine U.S. census regio ns as well. The significance of patriotism remains after inchlsion of controls for the th ther explanations for home bias. Wie find some support for information a-d famniliarity explanations in ou? palnej, We nd stron.g sipport for the transaction barriers exlana.tion. After controlling for the standard explanations for home bias, patriotism explains 7%i of the v:ari ati-on i:n ftoreig-n eq uity hoCdings. A:10% decrease in p.atlriot isr is associ. at ith a rise in foreign equity holdings of 2 —4% of total country equity, which in the U.S. represented 260-440 billion dollars in 1996. W )e perform severaal rol ustness tests. To control for poss ible enco]-Celeity, we instrument patriotism witl a. religion score from the lorid Values Survey and document that instrumen;ted patriotism is signiflicant in egxplaiinng foreign holdinogs. WVe lind no evidence that i nves t ors e xt a[ )polate past coolnt ry ret s or r ret;i f rse ore t.n... e explore variables capturing economic; poiiticai and social correlations with patriotism. Xe find that patriots are,-vernlmentoa mninhialists. haver less income and, education. and ihelong to a soc)ia identity. Of the correlates explored, GDP per capila, te percent.ge of the population with a secondary education, and the average propensity of residents to In Cderta;k ite hrit e ctivties can explain part Co th-e variation in flreligrn e(q]ity hold — ings. None of these correlates negates the significance of patriotism in explaining foreign holdings. Finally, we present evidence that countries with higher investor protection invest a larger per centage of their equity p ositioIn a3broad. Investor protectio vari ables do not diminisl, the significance of paatriotism. The remainder of the paper is organized as follows. Section I discusses why investors' patrio is n can affiect their e-.itv.y holdi s. Section II introdAlces the data iuseed in the study. In Section III our main results are presented showing patriotism- and instrumented 3

patriotism significant in epnng the home bias. Section iV contains a variety of robustness tests, including tests on the correlates, instrumented patriotism, and investor protect ion e:xplo r atios. Thbe last sectior c ncludes. L. -Why Patriotism?^ Appeals to patriotism appear to influence purchasing decisions in many countries, )bu to our Ik:owledge no study has doculimented the impacit of patr.otisPr}, on portfolio selection. The explanations of transaction barriers, risk, and information asymmetry have not fully ac coun -)tecd for the magn:itaude the home hias. Re(cent]ly, Grinblatt arLd Keloharju (2001), Huberman (2001), and Bhattacharya and Groznik (2001) find that familiarity might also play a role in explaining the home bias. Famniliarity bias is a sp-ecial case cf ambiguity aLve rsin, i:a wihich inavestors dislikie assets wlhen they do not know the distribution of returns.3 Familiarity bias is ambiguity aversion brought on by proximity to certain firms, independently of having information about the firmrs' prospe ctsz. Huber:[mna: (2001) finds that investors prefer to invest i:,local telephorlne companies. Grinblatt and Keloharju (2001) find that Finnish investors prefer stocks of domestic firms, especially those firms which communicate in Finnish and are operated 'by Finnish executr.ives. Aithough the authlors hypothesize that local biased illvesting is due to familiarity bias the results of their study are also consistent with patriotic nrvesting. Our approach is novel because we provide evidence that a bethavior in addition to f:am iliari ty is at work i:a homtne bia.sed choi e-:s.: The famiiar ity hypo the ses WOu id predi cl that knowledge of any stock's expected return distribution, regardless of nationality, would make it attractive to investors. Our hypothesis predicts that loyalty also causes investors to invest at homne. Several closely related puzzles of homne-focused choices can help us understand how patriotism af-fects the dlecicsicn to invest. Te first is in the sports betting mnarket, ard In lihis 1961 experiment. Eiilsberg firids that people prefer to bet on draws fron. an urn whenl they learn that the proportion of winning balls is one half than when they do not know this proportion (and thus the best estimate of it is one-half. 4A

the second is in employees' 401(k) asset allocation. Consider an amateur sports bettor. If she is risk averse and derives non-pecuniary utility from a home victory, expected utlilitv threory predicts that she will hede by bettirng on the opposin.gte-a, tLo deive a. payoff in the case of a loss. Contrary to this prediction, individuals tend to bet on their home~ team, even ih unfavorable odds (Cray and Cray, 1997). Strumypf (2003) withan faoab eleodd (ra and observes t egal okaks tiat at ialho. I oeonakrs tilt tehe oeds agast ho tn bets in l ai aemt to balance their books. A second related puzzle is the pattern of 401(k) portfolio allocations. People invest dis pr, oport ion:ate a-mounts of their ldicr:etionary 40](k) rireltrement assets, in their own company stock. For example, B:enartzi (2001) inds that Coca Cola employees allocate 76% / of their discretionary contributions to Coca.-Cola shtares.4 In Benartzi's survey, only 16.4% of respondents believed that their company stock was more risky than a diversified portfolio. The observed 401 (;) allocations strongly contradict nmean-varialmce thelory i that they fail t;o diversity away a firn rs u:nsyste:iati c risi.5 Even worse, the employer stock is the asset most correlated with the employees's human capital, and thlus (ignoring the benefits of signalling to shareholders) a rational agent should short emnployer stock, not hold large q uantities of it. The home team betting and 401 (k) phenomena can be explained by familiarity alone: proximity to thhoe ho teamn and to the firm allows the investor to understand the distri butio of oexpected retlrnsl thus lessening arbigi ty aversioni. I fact, Boyle et a] (2003) do just that. rThs far: there has been no way to distinguish between the familiarity theory a-nld the alternate theory of investor loyalty towards the home entity. The only other paper to our knowledge which tackles loyal behavior affecting investment decisions. Cohen (2003) shows that emuployees invest less in their firm if it is a conglomerate and ssuggests that loyalty is a driving force behirnd enmployee investmnent in thei:r own firm. in conglomerates, each dollar of investment is spread amnong all divisions in the conglomerate, and thus dilutes the investment in the homie fir.m. In the case of patriotic 4LHe also finds that past returns influence, but do not fully explair, these decisions: employees extrapiaLte past ret:urn i their deciskins to invet in company stocks. The retirement fund of CocaCola as a. whole allocated 90% of its value to the firm stock. 5See. e.g. Huberaln and Sengroller (2002). 5

investing, we are able to begin to disentangle loyalty from famniliarity. But familiarity alone cannot explain why more patriotic nations invest less abroad. Why is this so? A country's res'idelts ~ may bDe Ilmore familia.r witht their own rmarkets0; however fao i.i ari ty with a market should be independent with the strength of patriotism. In this paper, we show that loyalty is at least partly responsible for home biased investm.ent behavior. P:atriotic feelings. or loy)alty towa'rds holme country, c'a ai ffct investmeirn]t choices in both steps of an investment decision-making process: evaluating the performance of a securit (a stock or a tearn) and making a portfolio allocation decision. The effect of loyalty in the allocation de~cision is a:dressed first. Investors may correctly estm,,ae the mean-variance characteristics of a home stock, team or firm but are driven by loyalty to over- in]vost in, i. A ex:plaine inl Akerlof (1 983) lo(al loyaliy rma brin i; t agii.ble I neiits, sucih as jobs, improved infrastructure,7 and tax receipts from local corporations, some of which may be returned to the investor. The following anecdote provides an example. A central bank i: n ar e.neging inarket broulht ii:1 n(ew diLrector otf t;il pension friund delpartient. After s tudyihng the portl-fo.ik), the new ditect or approached the head of the central bank with an international diversification strategy. The central bank head quickly rejected the plan, explaining that the fund investment was not to go abroad because the country needed capital for growth. Allocation decisions may be influenced by loyalty even if there is no monetary externality associated with such investment. Investing at home may generate utility in the forli of the approval ofofilers an]( the ofeei f tr ib tiu; to societ-iy, 13an this lnmay reduce regret if the investment has low or negative returns. Bu Isinre ss week,capturtes the Sep Lte.mber 1 th loyalty in cvestnment sentiment i n t~he following quote: Patriotism is especially evident when it comes to the financial services sector. Suddenly, buying stocks in a down market is a duty. 60r ii; the case: ofi: Groznik and Bhattacharya (2002), i.rnmigrLation may imply lthat residents nLay be more iamniiar with particular other countiri ' mrakets. 'Huberman (1997) shows that people invest a large portion of their portfolios in their local phone coumpani. "Diane Bradley, Business Wek ek November 8, 2001

in essence, loyalty enters the utility funiction, whether or not the home entity benefits from the investment. Home betting or own lirm 401(k) investing umay generate a pride tlility in supportinrg the horne entity. The second way in which loyalty can affect investment decisions is by biasing investors' estimates of the risk-reward characteristics of a home investment proposition. According to Kahnemlan a~nd Lovallo's (1 993) 'inside view'' bias. loyal insiders may iden — tify strongly with an organization and may find it difficult to hold an independent view on the expected returns to coumpanly stock. Investors may have similar trouble evaluatin, their homire market or nhom te t ob ec ly. Selectively Se ive thinking; caused by an inside view could elicit either overestimation of the mean return (optimisnm) or underestimation of the associated risk (overconfidence). Optmiism can feed on itself; invesitor cr an be convinced that their home market is superior and selectively ignore contradictory evidence10 SimilarlIv. Heath and Tversky (1991) find evidence that overconfidence increa.ses with fanmiliasrity.` We have outlined the i.mporta.nce of loyalty atnd the ways in which it can affect investment. In the next section we will test whether this effect exists. II. Data The Appendix provides sources for all the variables used in the study. Patriotism data come front the W\orld Values Survey (\V'iVS) conductod during two periods, 1990 -1992 a.nd:1995 —:1997.1 Adults ove:r tle age o: 18 were i:nterviewed face —to-face in 53 countries. We focus on individuals' responses to the question: "How proud are you t'o be /.substiutnte rtional'tyI/?". Oulr measure of patriotism is the mneall country score. Table Kilhka and Weber (2000) provide survey evidene Lhat Germlan and American students overestLnimate their owvn country's future performance, and Stronmg and Xu (2003) find this bias among fund mnanagers in countries around the world. Tesar and Werner 199I5) estimate that the lack of diversification due to the home bias could be explained by a 620 to 800 basis point over-esitiatio of dom.estic returns. ]For more on belief perseverance and confirmatory bias, see Flabir (1998).: Ov.erconfidence is expioredi ibr examlpl]e in Odea'n, (1999) and Barber and Odean, (2000). T'ihe "illusion of cont.rol" (Langer 1975), in which people treat chance evewnts as controllable, may cause investors to perceive lower risk due to their infiuence on the outcome. lnThe survey is conducted and held by the Inter-University Consortium for Political and Social Research at the University of Michigan.

1 lists the patriotism scores for the two surveys. From the first to the second survey, the mean score increased marginally from 3.28 to 3.3:4.13 The 190-1992 r survey suggests that New \Vorld coultrlieLs are more patriotic than Old NWorld co untries, but thIe arg-er 1995-1997 survey results are more diverse withi in come and location groupings. Outr measutIre of foreign equity holdings is the perce-rlntage of foreign equity in a cou'ntry's aggregate equity portfolio. We calculate a weight of foreign holdings relative to total holdings (FEH) as follows:l'4 B l-Foreign equity hoidn1,gs Market capitalizatiron Fo eigr eqlity holdiings Foreign equazty liabilities Foreign equity holdings and foreign equity liabilities are from the IFS database. Market capitalization data are from the World Bank's World Development Indicators (WDI) database. While thle sLIrvev includes 53 coultries, the availability of forei-gn holdings restricts our sample to 40 observations, encompassing 33 countries. Seven countries have data for both 1991 sand 1996.:5 In order to supplement this saample, we collect sinil ar datai on:foreign i:nvelst.rernt for the nine Ul nitted States Celnsusi regionsi froml the 1 997 Survey of Consumer Fiinances. There are tour standard ex. planatio'ns for the e uity home bias. The first is that the presence of transactions costs limits investing abroad. NVe measure transactions costs with an indicator of Capital Account Restrictions from the IUF ExLchange Arrangements and Exchange Restrict-lons AnnuHal re;ports. An alter ative measlure of transactions costs is tariff and non-tariff barriers to trade from the Economist Intelligence Unit (EIt) Database. The index ra nges fromn 1 (few barriers) to 5 (marny barriers). The second common explanation for the home bias is the lack of information about foreign equities. Our mneasure of information is airline departures per capita from the WD1 diatasbase. We also consider other informrnat ion variables such as mobile phones and newspapers per capita, from WDI as well as distance and gravity measures from Great 3Ftur the countries participating in both surveys, the average score increased form 3.29 to 3.39. 14Our results renmain if we simply normalize foreign equity holdings by market capitalization. 6A:Reclkman selection test fails to reject that country selection is ins ir depeent of C DP and mnarket capitalization with respective p-values of 0.79 and 0.42.

Circle. The third explanation for the home bias is a lack of familiarity with foreign investmenrlt opportunities. Following the principles in Croznik and BhattalCharya (2003), our' familiarily measure]- is the percentalge of tthe popula.ition that is fo'reign: born, t aken from the WDI database. The forth standard home bias explarlation is the lack of an effective diversification beLne:fit. T lie risk-reward it:radeoff variable is constr. lted as thie country Sharps)e ra. ti nminus the rest of the world Sharpe ratio. We use five years of prior monthly returns for market indices from Datastream. We also calculate simple correlations for each of these indices with the rest of tthe world. We test whether economic, political, and social correlates of patriotism could be driving our fitdi.n,-s. Economnic variai)les which are correlated with patriot;ism score are GDP per capita from XWDi and tax revenue collections per GDP from the University of Michigan.s VWorld Tax Database. Political correlates are two rmeasures of political awareness firorm the WVS. The political nmeasure Discuss Politics follows fromn the question " When youH get together UNth youlr fzelds, woulld you say that you discuss political matters frequentily (t), ccaszoraly ('), or never (1)?". Trust Covernment is derived from "Could io tel i, me how m'inchL conrfidentce iyou have zin the governimenlt in ln isert capital city':: a great deal (), q'uite a lot (),?not very nmuch (2). or ntne (!)?". Social correlates of patriotism include percentage of rural population, othinolinguistic fractionalization, and ch:arity. Ru1ral populati:on perct i tehe pae:[rcen tagie o:f popullat;ion that li ves in rura] areas; from the WDI database. Ethnolinguistic fractionalization is a nmeasure of iin;uistic and ethnic group dispersion ranr;ging frorm 0 (not; very diiverse) to 100 (very diverse) from Easterly and Kraay (1 999). Charity is from the W*VS question which asks whether the respondent is an active rmemrrbe.r (?), inactive mremb.r (2), or not a mrember (I1 of a recogn7ized char-itabie orjganization.

II. Empirical Results T he cii:tra] fi:in! of the paper is de pited in FOilre 1i plots of paltriotism agaJinslt ioreign equity holdings for the two survey periods. IThese figures suggest a decreassing linear relation between patriotisn and foreign equity holdings. As reported in column 1 of 'Iable 11 as simple regression includin;g a d ummy.'or the 1995 —1997 nsur.vey period shows that patriotism is significant in explaining foreign investmnent with an K-square of 0.2'16 A one-unit increase in the patriotism score is associated with a 11 percentage point decrease in the propotrtkin of tor'eign holdings in ar averag!-e couint rys portto]li of equities. iTo furiher inves; igate the role ot patri-oti;.si on foreign equity holdin.gs we study the severn countries that are present for both surveys in our sample. Our hypothesis is that when patriotism rises foreign holdings decreases more (or increaseses less) than in instanic::es wbhen patriot tism falls. Sin:e data. constraisnts inuhibit more fo rmal t ests, we plot changes in foreign equity holdings against changes in patriotism. Figure 2 shows that while there are only seven countries in the sample that have data from both surveys, forei;grn ivestrent increases less vs wen resideitst,; becomre mnor]e pattriotic.[7 A. Additional U.S. Investor Evidence We have presented initial cross sectional and time series results showing the negative association between patriotism and foreign holdings. Would patriotism explain portfolio variationa within countries? To control for cross coLintry variation, re obtain patriotismL data from the 1995-97 WVS for nine regions in the United States and compare them to foreign equity holdings reported in the 1997 Survey of Consumer Finances. We use househo]d incom e aso; a proxy for net worth anid normXsalize foreign holdings by this figuire. Ioreign holdings are plotted against patriotism in Figure 3. T'he more patriotic regions:t'"[he sigrnficance of patriotisn holds irrespective of the inclusion of potential outliers. No variahle has a D])FBeta la.rer than 0.50. Cook's distances and iFFits measures are vwthin, reasonable limits. I7'[The figure als o sugges that both increasirlg gioba]isrn and in: creasirg pai.iotisnm charaicerizes the period between the 1990-1992 s urrve and the 1995- 1997 survey. GCermna.ny is the exceptional in its large decrease in patriotisr, possibly because of the economic strains fror unification during the period. 10

of the United States Wlest South Central, Mountain and NWest North Central invest the least in foreign equities. Evidence again supports the hypothesis that more patriotic in vesto0ts ar;e more el homne bia:ised in their inrvest]rents. B. Standard Explanations for the Horne Bias The four standard expla.n ations r proposed:or the eCquitvy ho:ie i i::e i tra:nsactio:ns barriers, information, familiarity and risk measurement.1' We briefly discuss these standard ex:xplalnat;ions and inc]hlue nmea;sures for themr in thle estiinatjioni m:odel. B.I:. Transactions Barriers Tra'nsaction barriers should:anegatively affect fo:reigri i nvest: n ent, so:mnewhat like a tax paid on the absolute value of the holdings of foreign stocks (Stulz, 1981). Barriers may be restrictions on capital outflows or frictions in repatriation of capital gains and dividends. Th e irh e bi as literature genera.lly c includes t hat t;anmsaction barriers are significant in impeding flows but are not economically large in explaining the bias.19 Our pr]incipal measure of transaction barriers is the TMF s cap ital account liberalization indicator variable.2' Columns 3 of Table iI reports that transactions barriers are negative and statistically signilicant n explaininrg foreign holdings, indicating that tran.sactio.n barietrs isnmpede outflo of s int o fns o oreign equiti,;.21 When we n:ach]de patriotism in column 4 both variables remain significant; the results suggest that patriotism explains a. portion of the home bias separate from that explained by transactions 'aD riers. I8A fifth argument from recent vwork by Rowland and Tesar (2000) is that individuals diversify by investing in rmultinationals. 09See Cooper and Kaplanis (1994), Tesar and Werner (1995), Errunza,:Hogar and IRlng (2000), (C'assman and Riddick (2001) and Fourth, Ahbearne, GCriever and WIarnock (2001). 2'One could measure barriers in terms of ]iberaiization ot capital markets (Bekaert and:larvey, 2000: Chari Land Heiinry; 2003). Hoowe.ver liberalization tuidies tend ito considrer only the opening of a market, to fiowi from abroad, not ouitfows of inves Linent. Additionaily, Bekaert and Hairtey's (2000) c:Aollt:.:Iio of financial liberalization dates all pre-date our first survey period 1990-1992. We seek a measure of differential liberalization, not liberalization per se. 21 A measure of tariff and non-tariff barriers from the El. U database yields similar but weaker results. 11

B.2. information Brennan and Cao (1997) develop a model of international equity portfolio investment fliws a.nd show thatl. differenltial ilformallion amrcing investors caan affect their investm.ent decisions. A number of empirical studies confirm this. There is conflicting evidence, however, on whether information affects foreign holdings. iTsinlg Korean data from 1997 -1998, Choe, Kho, and Stulz (2000) find that doti-estic individ(ual, but not institutional, investors have a small information advantage over foreign investors. In U.S. data, Coval and Moskowitz (2001) show that mutual funds earn abnormal returns when they invest in n.ea.riy fir::m. F'rocit, O Cnneli and Seasholes (2000) shiow tlhat daily iVnterltnationa. flo icws can forecast future equity returns, suggesting that foreign investors may have superior information. in a first test, we investigate whether home biased investors have superior information about their homne market. This is unlikely because all countries in our study are extrie:mely lhoine biased, and all inmarkets cannot be above average. lndeeed, we find that the more home biased countries do not have subsequently higher five-year monthly index returns. it is possible that investors only have an information advantage over individual stocks. If this is true, then investors who have more information about foreign stocks are more likely to buy th-em. Airtp]anre depa.itures per ca.pitaa is a, nat;turail m ea.sure of inforrnoation about investment opportunities in other countries. Airline departures captures commerce being transacted, as business travellers aere usually a large fraction of airline lie xtl e. Add iti on:ally. in: as mrniucl as air plane depuJartares ]are intervational, departlures captures potential occasions for learning about foreign investment opportunities.22 Table II, columns 5 anld 6 shows that sum of market capitalizations is significantly and positively relate-d to foreri;n equity h ciltingsi. Tbe aer the e ad eomi - 2"Mobile phones and newspapers per capita yield he smiar results to the results using airplane departures. In fact, correlation is larger than 0.65 amonPg newspapers, mobile phones and airplane departures. G(ravity and distance measures such as the average distance between a country and all other countries, the number of i ctiguos countries, an the physical size of a country are not significant in explai[ing foreigo ho ldings. Usipg the sum of market, capitalizationsE of:oatigiiuo)s c:ountries yieliV sirilar resu]ti as airplane departures. 12

nificance of nearby countries; the more willing residents are to hold foreign equities. When we add patriotism to the model, both variables remain significant, suggesting that bot1h inforlation and patriotism are import,'ant in explai.ni ng the home bias. The results from cohlmn 9, however: indicate that inclusion of the transactions barriers along with the information variable and patriotism negate the significance of the information variabl)e.23 Our data arve not sutfftiently rich to identlify tihe role of information sepasrate Irom that of transactions barriers. B.3. Famriliarity The home bias could simply reflect the fact that people dislike ambiguous situatioins and prefer to invest in the fanlliar even without superior information.24 Heath and Tversky (099.1) find thlat people underiweight choices of wviicih they ha:ve diffiiculty understanding the distribution of outcomes. Bhattacharya and Groznik (2002) find that U.S. investment in a country increases with the income of the U.S. nimigrant popular tion from t.hat counttry and conclude th:atl i:mn]r ir.: ants prefer to invest their countries of origin. interestingly, the authors find that the level of investment in foreign countries is unrelated to language or physical distance, which should affect information acquisition. B.hattac,.harya and Groz:nik's results, like GrLinbiatt a:nd Keloja:rIus (2001) fin:dings with Finnish investors, are also consistent with the possibility that immigrants invest close to home out of patriotism. Our measure of familiarity is the foreign percentage of foreign-born residents of a country. Table II, columnns 6 and 7 show that it is very significant in explaining foreign holding-s, w'hich c onfirmis BhattacJhar}ya anid G:roznik's results. The pat ariotism variable, however, remains significant in regressions incllding foreign population. This suggests that above and beyond familiarity, people invest at home out of patriotism. '3Even withoilt the incluisiorn of SharpeWorid, our risk measure, Airplarne:[)partures loses its sigiificance in the presence of iBarieers. 24 Literature on lamiliLarity i; srummarized by Barberis an.d Thaler (2002). 13

B.4. Risk An investor may choose to invest abroad to mitigate financial risk, and the benelits to doing so' may vary by country. rThe vextent of thtese diversificatiol benefits is diffi clt to estniate, however; because correlations between countries tend to increase in bear manrkets.2 Our risk variable is the dilference between a country's Sharpe ratio and the rest of wo rld Sharpe rat, constr ucted using 5 prior years of Vmo:nthly [returns.2' SharpeW\orld captures the difference in Sharpe ratio between investing at home and investing in the world portfolio. A larger SharlpeWrorld would be indicative of less notivation to invest atbroad. Resultts fror] this mnodel are presente-d in c olutnns 7 and S of TC ble lI. I Our risk variable is not significant in explaining foreign holdings. B.5. Econor ic Significance of Pat;riotism Is patriotism economically significrant? Column 9 of Table II brings together our measures; of tralnsac;tijons barriers, intforrmaioon, famiiiarity and risk. CJearly lthe variaibles are collinear, and the number of observations is small, but we present this model to highlight the persistent significance of patriotism in explaining the home bias. The re;ression coefficients of p atriot ism on foreign hold ings in Table Ii ranges fiomn -0.080 to -0.132. lence, for example, a ten percent decrease in the observed 1996 T.S. patriotism score of 3.73 is associated with a rise in foreign holdin-gs firom 11% to 14-16'% of total equity holdings, (a 2-45% increase). Give the U.S. total h ing in 1996 of $9 trillio:n this implies U.S. investors would have allocated $260 to $ 44 billion more in equities to foreign positions. For the melan country in our sample, similar calculations yield an iaverage. rise i'n toreign equity holdingl's fromn 9% to 1. 6-13.3% of total equity (a 30-60% increase) for a ten percent decrease in patriotism. The estimated effect of patriotism is economically significant. S.ee Ani T and BekLart (2.00) 2e\yV a.iso construct a mo. iiied vwfrEio)n capturing past returnsi to ouitreicy poiticlns, but this variable oroiy increases the noise in Laeas uremient. 27Interestinly, using simple correlations of the country index with a rest of world index yields a significant and positive coefficient. This counters portfolio theory but supports the idea that markets nay be segmrented. inclusion of the transaction barrier variable negates this significance. 14

Table TII also shows the estimated change in foreign holdings for a one standard deviat ion chan o te each of th variables of the full model of column 8, Table II. A oniest Lalard deviation positive chan.e independently in pariotsm -or transaction barriers results in a change in the foreign holdings percent by -3 of total equity holdings. The final colut.mn. of Table III presents the chan-le i R-Square that occurs vhen adding each variable) t-) the otherw:ise full miodel. Our results show t.hat c ompaare to a. regression of barriers, information and risk on foreign holdings, the regression of patriotism, barriers, information familiarity and risk explains 7% more of the foreign holdings vanriaion. Three concIusions:arise fror this sect lon. F irst, patriotis]I is sig:nifi Lan:lt l:nI xplainring the home bias, controlling for measures capturing the standard explanations. Second. the results pr esen t strt)ng evidence for the role of t;ransaction b: arriers, some evidence for the role of information and familiarity and no evidence risk in explaining the home bias. Third, patriotism has an economically significant efTect on foreign equity holdings. C. Endogeneity and Instrumental Variables VW explore whether the significance of patriotism is due to endogeneity. Could it 'De;hat a c untry'se residents -gr w diffe- rentially rpa:trioic bec ause of varyiang econ ovic prosperity created by markets or economics? Using Li.BOR as the risk-free rate; we coalstruct the one, lthree and five —year lagged excess market ret urs for each co.. Onty in our sample. in an unreported regression, we find that lagged returns do not explain the proportion of foreign equity holdings, either by themselves or as an olfset to patriotism. Thlese returns- also do not exp)la in the patriotismi: score. 'o control for the potentiai that an omitted economic variable is simultaneously driving foreign holdings and patriotism, we instrument patriotism with a \V/S question asking respondents how i:npo rtant religion is. Rel:gio:a a is an appe )aling instsrltilent ior patriotism for a ummber of reasons. Atta.chment to the nation may be stronger for societies in which religion plays a strong role.2` Religion is not an economic var2 Counit.ry loyalty is encou raged by some religio:ns. For example, Lhe New Testamientq qiuotes Jesus as saying, "Render to Caesar the things that are Caesar's" (Mark 12:17). In Judaism the obligation of 15

able. its importance to respondents should not be correlated with any omitted variable that is endogel ous to patriotism anld foreign holdings. The correlation between religion andd p-atriot;ism is 0.57. As a first test of t;he app:rop:riateness of our instrumnent, the correlation between religion and the residuals of the full estimation model of foreign holdings patriotismr barriers, information. amid risk on foreign holdingsiss -0.01 and not sig:nificatly di.fferent firon zero. in the first stage of a. 2SLS procedurie, our instru;ment significantly covaries with patriotism with an R.-Square of 0.44. The coefficient of instrumlented patriotism on FEH is -0.108 with a t-statistics of 2.61. A Ilausman test of the equivalence of the instrunen ted a:nd the OLS pa:triotism coefticients strongaly fails to reject the equivalence. IV. Robustness Tests Three key arguments arise that might threaten the robustness of our results. First, a covariate of patriotism might be causing its significance in explaining the home bias. The second 1ro,11 bstlness conc:ern: is that equity investmlent is typically tundeltaken only by the upper income earners. Finally, the nature of investor protection of a country might be correlated with both country pride and investment choices (Dahlquist, Pinkowitz Stulz and Williianrnson, 2003). We test these possibilities and find that they do not:negate the effect of patriotism on the home bias. A. Correlates of Patriotism Perhaps factors associated with a country's patriotism score drive the patriotismfo0reigon equity Vholdings relationship displayed in Figu re f i ti is seltio, we coinsider whether economic, political or social correlates of patriotism drive the home bias result. We find that patriotism is correlated negfativelv with country incore, education and [oyalty towards once's Land (e.g., Jeremiah 29:7) is written by the prophets. One of the core teachings of K' ng Fu I Tzu (ConfuciusAwas Clhung or ]oyalty to the State. 16

the burden of the government. Patriotism is positively correlated with faith in the government axnd social identities. if coiuttry —wi]de wealth and educatio]n are reflective of soc ities being riore p.l eased with the performance of their governmenits wealth and education should be positively related to patriotism. Conversely if more wealth, educated people know more about the rest of, the world, wealth aend ed.uc ation soiluld be negat ]iveiy related to p)atrio tisa]m. Partial correlations among the correlates variables and patriotism are reported in Table IV. Controlling for the survey year patriotism is strongly negatively corielated with GDP per capitla a3'1, witih the percentage of te latlion with secondlary el ca.tion. Regression results of foreign holdings on patriotism and either GDP or education are reported in ta-le V. Both GDP and edlucation can explain part of fo, ign. ho]dings beyond the effect of patriotism, but neither negates the significance of patriotism. The burden of government is measured economically as tax revenue collected norlmalized by GDP ' ard pohlitically as the WVS resonse score to thie q(uestionrl: 'How oflten: do you discuss politics at home?'. La.rger tax and discussion burdens may cause resentment. especially if tax burdens are not accompanlied with reciprocally large provision of services. OC the othelr hald, high t;axes a,:d m re discussions could imply a l.rger role for government in residents' daily life, and thus may be positively correlated with patriotismn. 'Tble IV shows that higher tax burdens are observed with less patriotism. ln Table V, howevevir, neitiher tax burtden nlor discuss poilitics is signrifi:cnt in expaining the home bias in the presence of patriotism. W~e proxy tolr the other 1 olii(cal dimensio:n of psat: iotisrm, faith il the goverlrn eint, with the WVS responses to "How mucin do you trust your national government.?'. Trust of government is positively related to patriotism, but the partial correlation is not signlifica tt. Th ius, we do not p rese nt t hei i1si grli fic:r Lt res lts of for t eifgo 11O I 1 ol S.]- l t rust of government. Finally, we consider social ideAntity correlates with patriotism. Individuals who have important tribal, ethnic; linguistic, township, or other local organization affiiagtions may have stronger identification of themselves as members of a group. A resident's ability 17

to identify with a group enlhances the identification with citizenship, thus strengthening patriotisml. TTnder the social identity hvpothesis, ethnolinguistic diversity as well as the percentage of popullatjn that is runal, and chanital t endenCieLs sh uJd a.l] be positively correlated with patriotism.29 The social identity correlates are significantly correlated with patriotism, supporting the social gro.p hypothesis.3 \Te are interested in knowing xvwether any of thfe social i:dentity c:rreilat es is di:riving the- sit!gnifi6cant. expl]anatc ry piwer of patriotism on foreign holdings. Columns 5-7 of Table V present the regression results of foreign holdings on patriotism, and social group correlates. Consistent with all of our nfidings, no ne of the so::cial identiy correlatesi ataes the sign:ifica:ncse of patr.iotism in explaining the home bias. Only the charity is significant by itself or with patriotism in explaining forei.gn holdings. XNe conclude from the correlate analysis that patriotism seems to capture a nunmber of fundamental chalracteristics of a country's residents. The residents of more patriotic colntries are govermenl..al mi imanistic, poorer, less ed.ucated, and affila.ted wlitbh social identity. None of these characteristics are however, the driving factor in why patriotism explains the equity home bias. B. Upper Income Earners A poss'"ible coer:icn with the rmieasure:mente of p1at;riotisrmt res.il s fromn identlif yin[g whic invests in equities. During the bull market in the United States, only 24% of wage earners under 830,000i held stock, while 84i%: of Americans with income over $75,000 malintained equity portfolios (L anger 2001). Clearly, the divergence gro-ws even larger for poorer countries in which disposabie income for the poorer half of the population is limited. Since the more affluent account for the majority of stock holdings, it may be that our cpatriotis,1 score does not capture the sentimenti; of upper income res pol dents. We re-run our analysis restricting the W7VS responses to reflect only the upper half of income earners in each couitrv. Our results ale not materially different excet tlhat "iOuir social identity hypothesis builids on thoughts of social capital from Coleman. (1990) and Guiso, Sapiensa and Zingales (2000). i1t. is adimittedly ulrprising hat et i:tiingusil ic diierity is. posiively corr.elated with patriotism. 18

the magnitude of the patriotism coefficient is consistently larger. The significance levels are very similar to the results presented. VnW infer that we may be undereestimating the eco:nomic: si;nificance of the pat;riotism effect. C. Investor Protection As a, fiLinal set of roblrstiness check s, we c c o:lnsi er the corpo )rate goverl[nance envi:ron:luien[t of the investing country. Motivation for this section stems from Dahlquist, Pinkowitz. Stuiz and William]son (2003) (DP'SW). DP'SW consider th:e environm-rLent of the eceiving; country of U.S. outward investment. In particular, they focus on the amount of closely held shares in a country and the nrvestor protection ratings. The amount of closely eeld shsares captures th ole rC of thite large shareholders in the country in. controlling the market portfolio. Their hypothesis is that the closely held percentage measures provides an indication of the amount the market that is available for outside investment. DPFSW find thiat indeed e t he closely hel ci measure is co:nsistently positively sigunifiica.t, in explaniing the U.S. outward investment to that country. Additionally expropriation risk explains a portion of tihe variation h in vestment positions. VWe focus OIn the cross-section of ilnve-stors: courntries rather than the cross-section of host countries for investment. In DPSW, the governance affects the amount of inflows of investment. In our study, t he overall o uttflows m]ay be affected bty the governance at home. iWe construct an average measure of closely held shares for each country by using the average closely held percentage for a country for all firms in \Toridscope. Then we take the difference between the closely held mneassure tor a country and tle rest of the world average. 'able NViI presents the result that the closely held percentage neither increases nor decreases the foreign 'holdings. Closely held may have two offsetting effects. On the one haend, the relative per-ceitage of closely held shares romay rflect poorer,gover — nance; thus encouraging more foreign investment. Conversely, the closely held measure is a reflection of the investor group for a country. Thus, a higher closely held percentage may re-fect more inwarld- foculsed i:nvestm.ent if thie lar;est sharelhold ers are extr acting private benefits of control. LaPorta, Lopez-de-Silanes, Sh]eifer and Vishny (1998) (LLSV) 19

show that the concentration of ownership of shares is negatively correlated to investor protection. ie follow D'SW A and i nc]ludet first investor pro tect.io antd then exropriation protection from EIL[:. We also include a measure of corruption of public officials from the University of Goettingen and Transparency International. Table VII shows that a country witL bette investor and expropriat ioni potect; ions aniL less co:rru ptito inlvest rmore abroad. This is inconsistent, withi the concept that investors flee their own poor governance countries and is consistent with the hvpothesis that a country s overall investmileni posil.io is ai ffected b'y p:rivate benef6its of control. XVe conc lude th1at governa may indeed e Jnfluencing the portfolio allocations of investors across countries, but this iln.erestincg fi ndi!-j do es not ero ie any) of tlthe sig]nifica.nce of platiotisrm. 'o further explore the role of investor protection oni portfolio decisions, we explore the legal origii of investing countries. LLSV (1998) show that co.mmon law countries have stronger prot ect ion tor investor rights over civil law countries,. 'Thuls, if counitries with weaker protection hold less foreign holdings, possibly due to the taking of private benefits, civil law should be neg vel relaiv r ed to foreign holdings. Table VII shows that an idi cator,enoting civil lei!al tr ad:ittio is Ilegative but.no.t significant in explialni:i!g ioreign holdings.' One inference is that legal origin may be a noisier measure of the taking of private benefits than investor protector indices.o2 V. Conclusion In the ilrst (to our knowledge) study of the effect of country loyalty on investment, we find that inve:stors in more p atriotic cou ntr les o hol sinmaller fo reign ecquiity position:s. Supporting this finding, investors in more patriotic T.S. regions hold less foreign equity, and changes in patriotic behavior asre negativelv related to chalnges in foreign equity ni n an urnrepor. ted regression, we ihnd no signiicant fixed effec, anmong commno:n iaw. French civil law, and, other civil law. "'2We also consider the influence on investment decisions of a country beingv bank-based. WTe find no significant effect. 20

holdings. This result is robust to controls for transactions barriers, risk information, and familiarity, which our study is able to distinguish from loyalty. Economic political atid social co:,rrelates of patriotismr ddod not ne!gate the s-i;rnificance of patriotis:nm. WVe find strong support for the transaction barriers explanation for the home bias in addition to patriotism. Inclusion of patriotism accounts for 16.5% in variation of foreign holdings, and a 10% decrease in patriotism is associatetd wvith a rise in foreign equity holdings of 2 —4% of total country holdings, a 29-48%( increase. Our results are robust to tests of endogeneity. Neither instrumenting patriotism with the inrportat.ce of religion nor inrcluding past (or future) count ry inrdex l retu Lrs n11,egates the statistical or economic significance of our patriotism result. \We also test for the o butstcess of the patriotismn variable st e arst moeasures of transacltionas barriers, risk; and information. \We find evidence that domestic investors in poor governance countries choose to invest at home, possibly to take advantage of private benefits of conLtrol. No.ne of the robtstness tests of eludogeneity, alternative 1( -easur-es and corporate governance negates the effect of patriotism on the home bias. Two implications call be drawn from our study. The first is that Ipatriotic' behaviorL appears to have a ]large part in explain ing the phenomenon that mal.ifses its elf in the home team bias, the 401(k) investment allocations, and the equity home bias. The second implication of this study is that policies aimed at reducing the home bias may bie less successful in countries where: investors prefe r not to invest a;broad bec3lus they are patriotic. Patriotism results in a winner's curse in the sense th at the person valuing a. stock most hi; ghiy will utltitmate'ly be the hi;i.heat bidder in any auction. The citizens of a country will likely be the highest bidders for their ownv country's assets, thus possibly driving up the price (and capitalization) of their own market. In a more general equilibriurl setting, people invest disproportionately at home, but a portion of this over-investment is offset by the home biased choices of other countries. This may push prices upward in more patriotic countries, and downward in less patriotic countries. As in BarbLeris and 'T aler (2002), however, price distortion does not necessarily point to a "free lunch. iolding domestic equities may enter directly into the utility 21

function. Thus; if one were to hold the mean-variance efficient portfolio in lieu of the home bias portfolio, one's utility may be lower. Prices may be umnaffected if there exists a groutp cf:deep pocki:eted arbitrageurs whto are not; affected b)y p atriotismrn anId tumdersta;n;fd the market capitaiization biases caused by countries' home biases. They can take short (long) positions in a country if domestic home bias dominates (is subsumned by) the bias of foretign invesi.tors choos ]ing not to invest in that couintry. The effect of patriotism may also explain the home bias in consumer products. Lewis (1999) finds that the equity home bias and the home bias in consumption are linked. Concurrently, Bennett atnd Young's (1999) t;heoretk:al nmodel suggests thal optimal port;folios should be biased towards equities in commodiies that attract a large share in its c(>:sumptrion expenditure.33 An interesting future streamrt of research will expJlore whether the effect of patriotism on the equity home bias comes via consumption choices, and whetthlr patiortitsm has t+he potential to affect prices. n3This is controversial: Uppal's (1993) theoretical model suggests that it is unlikely. 22

0.45................... 0.4 0.35 0.3 o 0.25 t z W 0.2 0.1 015 Belgium Netherlands Austn a + UK Denmark 4 Gemmany France" 0.1 0.05 0 o Italy Canada *USA Spain *Finand Argentna S. Afnca 2.5 2.7 2.9 3.1 3.3 WVS Patriotism Score 35 3.7 39 4.1 Figure la. Negative Relation Between Foreign Equity Holdings and Patriotism, 1990-1992 XVorld Values Survey. The line is the fitted regression line. 23

nA 0.4 * Switzerland 0.35 0.3 31 4 Germany 0.25 W 0.2 - 9Sweden a.15 Estonia NewZealand 0.1 Austraia.USA 0905 -Latvia a C* Slove 0.05 4 Armena F n and Ltuaa Ru i Ma dova Bulgaria entia Inda Pol ad SAri e 2.5 2.7 2.9 3.1 3.3 3.5 3.7 3.9 41 WVS Patriotism Score Figure lb Negative Relation Between Foreign Equity Holdings and Patriotism, 1995-1997 lVorld Vailes Survey. The linle is the fitted regression line. 24

0.16 0.14 * Germany 0.12 S 0.1 0.08 008 \ Sweden 0.06 ~1 0USA 0.04 Finland 0.02 Argentna ^Africa -0 -0.2 0.15 0.1 0.05 0 0.05 0.1 0.15 0.2 0.25 03 0. 5 Change in Patriotism Score Figure 2. Negative Relation Between the Change in Foreign Equity Holdings and the Change in Patriotism. The changes are measured from the 1990-1992 survey to the 1 995-1 997 survey. Z~

"'it - ~ c -.'- t t- 5 Ct~ ci 'i c "'- -tc;tI (t~ i 1<) r! - 5> c - A -~r-r - - t cz 2 K k A> CD -CD t~ao - -i C - CO -v Foreign Equity Holdings????? 3 c0 I Ul m *~. *m o ON ON 57 m CA w a0 a 0 = 0 '0 0 I a M0 0 I

9 m 6) I If,.2. 5 If U o - h " m Z 4 6 > za, 0 6, 2t. u s 0 'C C=L PH o 0 0 6) 2C oC 0) - o,, < 3 0 0 > 3P = X. o i - r 2 0 0) I 06 C - I s C0 0 t "a u S - s - ~0 1C *SI. t, 0 0 0), C 0), VI +z E 0 +, I I cn........"..... cn i i r CO CO - N r A 0C- 0' o r On On n On n On n On n On On On Cm C I S i M 0 I, 0 o r EE 0) CL 0) II on I 0' 0) I C C 0 0\ C 0.c +^ Cc'.Inon S Im I II I I I IIII7IR I I I - c Iri. iri-MIr I I I I I I ^I I I I I I I I I.I." " " " ir I In on I 0 U 0 < ~ s g 9 a 1 C E C < C l4l 0) s 3 zz m 0) CL 0) 0 C ' E 9 'z u < a0 'C 0) g C. I 'C 7a C 0) 'C I 0 U 0 s.4 i.2 0) CL 0) z 0 N o e! 0 1s N P H, -Z-= m 9 0 C I 8 I u o 3 _j C 5 6) 'C 6) o o 'ir 0 6, C oN 6) oN oN '3 M.a> C C s) ' t~CN 'C av I I C toN oN C1 C u 0N

Table II Estimations Showing Patriotism's Significance in Explaining the Home Bias OLS estimates for regressions of foreign equity holdings on patriotism score and measures of standard explanations for the home bias. Foreign holdings is a percent of total equity holdings and is calculated from IFS data. Patriotism is from the World Values Survey (WVS) and is an increasing index from 1 to 4. Barriers is an IMF indicator variable of Capital Account Restrictions. AirDepart/Pop is the number of airplane departures per GDP from the WDI database. ForeignPop is the precentage of a population that is foreign born. SharpeWorld is the difference in the Sharpe ratio at home versus the rest of world Sharpe ratio. Significance at the 10%, 5% and 1% levels are indicated by *, ** and *** respectively. T-statistics are in parentheses. Cr Dependent Variable: Foreign Equity Holdings _1 2 3 4 5 6 7 8 9 10 Patriotism -0.112 * -0.080 * -0.132 * * -0.081 * -0.108 * -0.094 * (2.40) (1.90) (3.02) (1.73) (2.52) (1.97) Barriers -0.109 * * -0.097 ** -0.062 * (3.87) (3.48) (1.74) AirDepart/Pop 0.004 * 0.004 * 0.002 2.09 2.63 1.07 ForeignPop 0.006 * * 0.005 * 0.002 (2.71) (2.10) 0.83 SharpeWorld -0.036 -0.052 -0.039 (0.77) (1.18) (0.75) SurveyDummy Y Y Y Y Y Y Y Y Y Y Obs 40 40 40 38 38 40 40 37 37 37 R2 0.21 0.35 0.41 0.19 0.36 0.23 0.29 0.14 0.29 0.47

Table III Patriotism is Economically Significant Several measures of economic significance of patriotism and other explanations of the equity home bias. The full model refers to the regression of foreign equity holdings on the survey year dummy, patiotism, transaction barriers, the sum of neighboring countries' market capitalizations, and modified sharpe ratio. The first colum contains the coefficient of the variable in the full model. The second column is the standard deviation of that variable. The third column reports the change in the percentage of equity holdings that are foreign for a one standard deviation change in the row variable. The final column shows the change in RSquared that occurs when the row variable is added to the full model. Li) Change in Foreign Estimated Coefficient Equity Holdings for 1 Change in R2 when when Regressed on Standard Deviation of Standard Deviation Variable is Added to the Variable Foreign Equity Holdings the Variable Change in Variable Full Model Patriotism -0.094 0.310 -0.03 0.07 Barriers -0.062 0.479 -0.03 0.05 AirDepart/Pop 0.002 0.449 0.00 0.02 ForeignPop 0.002 8.221 0.02 0.01 SharpeWorld -0.039 0.449 -0.02 0.02

Table IV Partial Correlations, Controlling for Survey Year, of Patriotism and Economic, Political and Social Variables Partial correlations, controlling for survey year, are presented for economic, political, and social correlates of patriotism. Economic correlates are GDP/Cap, income inequality, and Taxes/GDP. GDP/Cap is from the World Development Indicators. Taxes are scaled by GDP. Political correlates are crime, trust of government and propensity to discuss politics. Crime is the number of violent crimes per 100,000 inhabitants in 1993. TrustGov and Discuss refer to the World Values Survey questions of whether the respondent trusts the government and whether he or she discusses politics. Social correlates are entholinguistic dispersion, rural population, charity and religion. Ethnolinguistic dispersion is an index from Easterly and Kraay (1999). Rural is the percentage of rural population. Charity is the response to the World Values Survey question of charitable activity. Significance in the partial correlations is denoted by *, **, and *** for the 10%, 5% and 1% levels respectively. _Patriotism GDP/Cap Education Taxes/GDP TrustGov Discuss Ethnoling Rural Patriotism 1 GDP/Cap -0.22 *1 Education -0.40 ** 0.76 1 Taxes/GDP -0.62 ** 0.31* 0.54 *** TrustGov 0.15 0.11 0.12 -0.08 1 Discuss -0.32 *** -0.01 0.22 * 0.15 -0.17 1 Ethnoling 0.35 ** -0.35 ** -0.40 *** -0.11 0.23 0.03 1 Rural 0.28 ** -0.40 *** -0.03 -0.25 0.38 ** -0.07 0.62 *** Charity 0.25 ** 0.31** 0.09 0.18 -0.12 -0.02 0.10 -0.18

Table V Regressions Showing the Patriotism Result is Not Driven Solely by Correlates OLS estimates for regressions of foreign equity holdings on patriotism score and significant patriotism correlates. Foreign holdings is the percentage of a country's total equity holdings abroad, from IFS data. Patriotism is from the World Values Survey.GDP/Cap is GDP (in millions) per capita from World Development Indicators. Education is the percentage of the population with a secondary education from the WDI database. Rural is the percentage of rural population. Discuss refers to the World Values Survey question of how often an individual discusses politics. Enthnoling is the enthnoliguistic dispersion from Easterly and Kraay (1999). Charity is a World Values Survey question asking whether the individual participates in charitable activities. Significance at the 10%, 5% and 1% levels are indicated by *, ** and *** respectively. T-statistics are inparentheses. _1 2 3 4 5 6 7 Patriotism -0.086 ** -0.085 * -0.132 * -0.100 ** -0.088 ** -0.168 ** -0.121 *** (2.36) (1.81) (1.82) (2.28) (2.06) (2.51) (2.74) GDP/Cap 6.371 *** (3.43) Education 0.002 ** (2.66) Taxes/GDP 0.445 (1.56) Discuss 0.010 (0.10) Rural -0.056 (0.43) Enthnoling 0.000 (0.25) Charity 0.157 ** (2.09) Survey Dum Y Y Y Y Y Y Y Ohs 38 36 23 35 38 28 33 R _0.42 0.35 0.44 0.26 0.23 0.33 0.35 1 -

Table VI Patriotism and the Role of Corporate Governance for Foreign Holdings OLS estimates for regressions of foreign equity holdings on the patriotism score and corporate governance measures. Foreign holdings is as a percentage of a country's total equity holdings, calculated from IFS data. Patriotism is from the World Values Survey and is an increasing index from 1 to 4. Closely is the average percentage of closely held shares for all Worldscope country firms minus the average closely held percentage from the rest of the world. Investor protection and expropriation risk are from EIU; both of these measures are increasing in better protection. Expropriate is from the EIU database. Corruption is the Tranparency International Index increasing in corruption form 1 to 10. CivilLaw is from LaPorta et al (1999). Significance at the 10%, 5% and 1% levels are indicated by *, ** and *** respectively. T-statistics are in parentheses. Dependent Variable: Foreign Equity Holdings 3 4 5 6 7 Patriotism -0.200*** -0.143*** -0.156*** -0.120*** -0.140 ** (3.09) (2.77) (2.97) (2.82) (2.58) Closely 0.013 (0.08) InvestProt 0.037 ** (2.48) ExpropriateProt 0.074 ** (2.05) Corruption -0.023 *** (3.55) CivilLaw -0.044 (1.01) Survey Dum Y Y Y Y Y Obs 28 32 32 37 40 R2 0.34 0.42 0.38 0.45 0.23 o O

Appendix: Data sources Variabl Notation Definition and Source Mean StD Min Max Bank vs Market BANK Demirguc-Kunt and Levine (1999). Rankings are an average of economic ratios and a high score represents a more bank-based countr (we inverted the original scale for ease of intepretation) -039 078 203 066 045 051 000 100 120 017 103 181 Barriers Charty Coruption Discuss Politics Diversification Benefits BARRIERS IMF Capital Account Restrictions from the Exchange Arrangements and Exchange Reshtictions Annual Reports CHARITY WVS:"Now I am going to read off a list of voluntary organizations; for each one, could you tell me whether you are an active member, an inactive member or not a member of that type of organization? [.. Charitable organization." 3: Active membe, 2: Inactive member, 1: Not a member CORRUPT University of Goettingen and Transparency Intenational A composit index measuring the corruption of public officials Range from I (not corrupt) to 10 (very corrupt) DISCUSS WVS: "When you get together with your friends, would you say you discuss political matters frequently, occasionally or ever?" 3: Frequently, 2: Occasionally, 1: Never DIV BEN Datashteam Calculated as varance ofcountry index minus covariance of the country index with the world market index 312 215 092 750 197 015 160 230 009 033 -105 077 Ethnoliguistic Fractionalization ETHNOLING Easterly and Kraay (1999) Measure of linguistic and ethnic group dispersion ranging from 0 (not very diverse) to 100 (very diverse) 314 262 10 880 Exppropiation Risk Financial Risk Foreign Equity Holdings Foreign Equity Holdings -US GDP per Capita Imports Inequality Legal Origin Natural Distance Newspaper Circulation Patriotism EXPROPRIATE Economic Intelligence Unit (EIU) Country Indicators Database FINISK Harvey: http://www.dukeedu/-charvey/ Index of financial sector risk ranging from 0 (low risk) to 100 (high risk) For 1995 only FEH IFS Database Foreign equity holdings divided by total equity held by country residents, defined as counhry market capitalization plus residents' foreign equity holdings minus domestic equity holdings of foreigners FEH US Survey of Consumer Finance (1997) Foreign equity holdings for 9 U.S. Census regions divided by average region income GDP/CAP World Development Indicators IMPORTS U.N. Comtrade Database Imports divided by GDP INEQUALITY TheilIndex from the Univerity ofTexas at Austin LEGAL ORIGIN LaPorta et al (1999) Indicator for civil law countries NDISTANCE Great Circle Average distance from a counhty to 141 other countries NEWSPAPER World Bank World DevelopmentIndicators Database Circulation of daily newspapers per 1000 people PATRIOTISM How proud are you to be [insert nationality]? 4: Vey proud, 3: Quite proud, 2: Not very proud, 1: Not at all proud The raw scores are originally I for high and 4 for low patrotism, but we subtract them from 5 Other survey variables are subtracted from 5 as well when this eases intempretation 585 67 514 773 009 011 000 038 005 002 002 009 16219 223.2 355 019 1252 11657 1173 52 039 190 392 759 270 00 681 41841 5971 486 10 1403 2318 1572 01 5780 318 052 137 392 33

Religion Percentage of Rural Residents Sum of Market Capitalizations Taxes/GDP Tmst of Government Tariff and NonTariff Barners Televisions per capita Violent Crime RELIGION Please say, for each of the following, how important it is in your life [.. ]Would you say Religion is.. 4: Very important 3: Rather important 2: Not very important 1: Not importantt a all RURAL World Development Indicators Percentage of population that live in rural areas SumMktCap Sum of dollar market capitalizations of adjacent countries (x 10^12) TAXES Gross tax reciepts divided by GDP Prom World Tax Database, University of Michigan Office ofTax Policy Research TRUSTGOV WVS: "When you get together with your friends, would you say that you discuss political matters frequently (3), occasionally (2), or never (1)?" and "Could you tell me how much confidence you have in the government in [insert capital city]: a great deal (4), quite a lot (3), not very much (2), or none (1)? T_NTB EIU Database Barrier index ranging from I (few barriers) to 5 (many barriers) TV EIU Database Televisions per capita CRIME World Competitiveness Yearbook (1995) Murders, violent crmes and robberies per 100,000 people in 1993 362 050 291 465 258 125 35 539 046 064 000 306 009 008 000 040 226 025 184 283 228 039 1050 075 100 400 016 010 074 750 27 2725 3II

References Ahearne, AlHan, Wii]iam' Grieve[, and Francis Warnock, 2001. Infformation costs and homebias: An analysis of U.S. holdings of foreign equities. Federal 1Reserve Board, interrniatonal Finanlce Divsiont, WVorlinmg Paper 691. Wi'ashingto,. D.C. Akeriof, (eoirge, 1)83, Loyal:y filters, Almerican, Economic Review 73, 54 —63 Ang, A., and G. Bekaiert, 2002, International asset allocation with regime shifts, Review of Financial Sturies 15, 1],137-1187. Barberis, Nicholas, and Richard Thaler, 2003, A survey of behavioral finance: Forthcoming, Handbook of Economics of Fina'nce. (North llolland) Edited by George Constantinides, Milton Harris andi Rene Stuiz. Baxter, Marianne, and Urban Jermann, 1997, The international diversification puzzle is worse than you think, Am'oerican Econormic Review 87, 170 180. Becker, (Gary, ].957, '7Te Econormicts of D)iscrmniination, Second Editioni. (University of Chicago Press, Chicago, Tllinois). Bekaert, Gert, and Cambell Harvey, 2000, Foreign speculators andc emerging equity markets..]ournal of Finarnce 45, 565 -]13. Benartzi, Shlomo, 2001, Excessive extrapolation and the allocation of 401(k) accounts to company stock; Jojurnal of Frar:nce 56, 1747-1764. Bennett, James, and Lesle Yanrig, ] 999, lnternational stock market; equilibriulm with bet e rogenous tastes, The American Economic Review 89, 639 -648. Bhattacharya, IUtpai, and Peter Groznik, 2002, Melting pot or salad bowl: Some evidence from i S. inve stentsi abroad, f;n.r;vers;it of [n.diana WI/or king Paper. Boy]e, ' helim; Ramana Uppal, and Tan Wang, 2003, Amlbiguity Aversion and tihe Puzzle of Own-Company Stock in Pension Plans, IrFA lWor.:ing Paper. Brennan, Michael, and lienry Cao, 1997, Internationai portfolio investment flows, Journal of Finanrce 52, 1851-1880. Chari, Anmusha, and Peter Blair Henry, 2003, Sock r market: liberalization and tihe repricirg of systematic risk, iForthcoming, Journal of Finance. Choe, Hyuk, Bong-Clhan Kho; and Rene Stulz, Janruary 2001, Do donmestic investors have mote) valuable infornmation about indivdual] stockis than foreig[o investoNs?, National Bureau of Economic Research W.r/king Paper. CohLen, Lauren, 2003, Loyality -baed portfolio choice, U'nivesity of Chicago iW orl-irg Paper. Cooper, ]an, and Evi Kaplanris, 1]994, Homne bias in equity portfolios, irflatLion hedging andL international capital market equilibrium,, Review of Flinancial Studies 7, 45 60. Coval, Josh, anrd Tobias Mokowitz, 1999, Homre bias at home: Local et qity p[reference in domestic portfolios, Journal of Finance 54, 2045 2074. 35

Coval, Josh, a.nd Tobias Moskowitz, 2001, The geography of in vestment: Informned trading and isset; Lprices, Journai oJ 'Polilical EcoIrtomly 109, 81: 1-841. Dahlquist, Magnus, Lee Pinkow'iz, Rene Stulz, and PRoban Williarnson, 2003, Corporate goverr ance and. the home binas, jiiJrnal of Fi ancial and Qulantitative Analysis, forthcoming. DeBondt, Werner, 2000, A portrait of the individual investor, EJuropean Ecolnomic Review 42, 83:1 844. Demirgluc-KunLt, As[i. rind Ross Levine, 2001, Financial Structure and Economnic Grow th. A Cr~oss-Co0untry (Co7iparison of iBanks, Markets, and Development MIT Press, Cambridge Massachussets chap. Bank Based and Market Based financial systems: Cross country cormErrunza, Vihang, lied Hogan, ad Miao-nWei Hung, 2000, Can the gains from international diversification be achieved without trading abroa.d?, Journal of Finance 54, 2075-2107. Frederick, Jim, Ji, 2001, Nat. ioal sec urities, AM'oney 30, 31. French, Kenneth, and James Poterba, 1991, investor diversification and international equity markets, Am, erican Ecoiornmic Reviewu 81, 222 226. Froot, Kenneth, Paul O'Connell, and Mark Seasholes, 2001, The portfolio flows of international investors, Journal of Finoracial Economics 59, 151 193. Cribl att, Mark, and Nl at:ti KeLoharju, 2001, ]Distaince, language, and cutaLLre bias: The role of investor sophistication, Journal of Finarce 56, 1053 1073. Guiso, Luigi, Paola Sapienza, and Ltigi Zingales, 2003, People's opirlm? religion andi ecolnomlic attitttde.c, for thcom,i g in the Journnal oi JIon etary Economics. Harvey, Cambell, 1995, The risk exposure of emerging equity markets, World Bank Ecnormic Review pp. 19-50. Heatih, Chip, and Armois Tversky, 1991, Preference and belief: Amnbigiuity and conmpletence in choice under uncertainty, Journal of Risk and Uncertainty 4. Hubeirman, Gur; 2001, Familiarity breeds investment; Review of Finanicial Studies 14; 659 680. Huberman, Cur, and Paul Segmiller, 2002, Company Stock in 401(k) plans, C(olbinmia Business School Working Paper.:K ahnrei ial. an, Dd )an a ovallo 3, 993 Timid choices and told forIecas tL: A cognitive perspective on risk taking, Management Science 39, 17 31. Karolyi, Andrew, and Rene Stulz, 2003, Are financiil assets priced locally or globally? Forthcoming, Ha7ndbook of Econolics of Fainace. (North Holland) Edited b]y George Constantinides, Milton Harris and tRen Stulz. Kilka, MN.; and M. Veber, 2000, Home-Bias in Internrational Stock Return Expectations, Journal of Psychology and F/inanc:ial./larkets, 176 9 —193. Langer, Ellen, 1975, The illusion of control, Joarnal of Personality and Social Psychology pp. 311-328. 36

Langer, Gary, July 17 2001, Market gyration brings jitters, AB4C(ews.corn. Lewis, Karen, 1999, 'Ttying to explain home bias in equities and consumption, Journal of Econo'm.r ic Literrature' 37, 571-608. O(bstLe[id, Maurice, a[.d Kennreth Rogolf., 2000, 'I The six major puzzLes ih. inter'national maecroeconomics: Is there a common cause?, NBEJ,? Macroeconor ics Annual 15. Pastor, Lubos, 2000, Portfolio selection ar.d asset pricing models, Jouirnal of Finrance 55, 1'79 21)9. Rabin, iMatthew, 1998, Psychology and economics, journal of Economric Literoi.ure 36, 1 -46. Rafael iaPorta, Fokrenlcio Lopez-lde-Sianes, Andrei Shieifer, and Robert: Vishny, 1)998, Law and finaine, Journal of Political Economry 106, [1]3- 1] 55. Rowland, Patrick, and Linda Tesar, 2000, Multinationals and the gains from international diversification, U.iv.versity of Michigan Departmnent of Economics 'Working paper. Strong, Norman, and Xinzhong Xu, 2003, Understanding the equity home bias: evidence from survey data, Review of Econoriics and Statistics 85, 307312. Str'uitpl, 1Koletran, 2003, Illegal sports bookmakers, akrs niersijty of North Carolina. ChaFpel H il1 Working Paper. Stuiz, Rene, 1981, A model of international asset; pricing; Journal of Finanicial Econorrics 9, 383-406. Sulon, Bill, 2001, Savings bonds appeal to patriotic pennsylvania investors, The Patriot-News Noveamber 20. Tesar, Linda, and Ingrid Werner, 1995, Home bias and the globalization of securities markets, JournIal of irnternrational Mi.oney and Finrn.ce:14. 4li7-492. Uppal, IRuainan, 1993, A general equiibriulnm model Of internat:iocnal portfolio choice, Jou'ra.l of 1i.nance 48, 529 553. VVarnock, Francis. 2001.; Iromne bias and high turnover reconsiered, Federarl Reserve WT/VorTinrg Paper Nrumber 72S'. 37