Economic History Association
History as Reflected in Capital Markets: The Case of World War II
Author(s): Bruno S. Frey and Marcel Kucher
Source: The Journal of Economic History, Vol. 60, No. 2 (Jun., 2000), pp. 468-496
Published by: Cambridge University Press on behalf of the Economic History Association
Stable URL: http://www.jstor.org/stable/2566380 .
Accessed: 18/04/2013 16:42
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
Cambridge University Press and Economic History Association are collaborating with JSTOR to digitize,
preserve and extend access to The Journal of Economic History.
http://www.jstor.org
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
History as Reflected in Capital Markets:
The Case of WorldWarII
BRUNO S. FREYAND MARCELKUCHER
Historical events are reflected in asset prices. We analyze movements in the price of
bonds issued by five European governments and traded on the Swiss bourse between
1928 and 1948, with special attentionto the war years. Some war events that are generally considered crucial are clearly reflected in government bond prices. This holds, in
particular,for the official outbreakof the war and changes in national sovereignty. But
other events to which historians attachgreat importanceare notreflected in bond prices,
most prominently Germany's capitulation in 1945.
This studylooks at changesin the value of financialassetsas reflections
of historicalevents.More specifically,the historicalevents considered
herebracketWorldWarII, beginningwith Hitler'sappointlnentas chancellor on 30 January1933 andendingwiththe two MarshallPlanConferences
in September1947. We analyze changes in the price of sovereign bonds
denominatedin Swiss francs and tradedon the Swiss bourse duringthis
period. While all the belligerentsinterferedheavily in-or even closedtheirfinancialexchanges,the Swiss government,for reasonsof neutrality,
refrainedfrom doing so (except for the two monthsfollowing the German
attackagainstthe Westin May and June 1940, when the Swiss boursedid
close). Five issuers dominatedthe Swiss government-bondmarket:Germany,the mainaggressorin WorldWarII;Austria,a countryintegratedinto
the ThirdReich well before the outbreakof the war; France,Germany's
traditionalenemy in the West;and Belgium and Switzerland,two neutral
countries,the first of which was drawninto the war, while the latterwas
spareddirectinvolvement.Therewas only very limitedtradingin the bonds
of othergovernments.
In this study we addresstwo questions that approachthe relationship
between historicalevents andcapitalmarketsfrom opposite angles. First,
to whatextentcanchangesin government-bondpricesbe relatedto historical events?Do all breaksin the price series correspondto what have been
establishedas crucial events in WorldWarII, or are there breakswhich
The Journal of Economic History, Vol. 60, No. 2 (June 2000). c The Economic History
Association.All rightsreserved.ISSN 0022-0507.
BrunoS. Frey is ProfessorandMarcelKucheris ResearchAssociateat the Instituteof Empirical
EconomicResearch,UniversityofZurich,Bltumlisalpstrasse
10,CH-8006Zurich,Switzerland.Email:
bsfrey@iew.unizh.ch.
We are gratefulto GaryS. Becker,MatthiasBenz, KnutBorchardt,RobertChirinko,WernerDe
LarsFeld,LorenzGoette,TimothyGuinnane,Jakobde Haan,RetoJegen,
Bondt,ReinerEichenberger,
GebhardKirchgaessner,Felix Oberholzer-Gee,JanOsterloh,Juergde Spindler,PeterStolz, Isabelle
Vautravers,Jande Vries, andtwo anonymousrefereesfor theirhelpfulcomments.
468
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
469
cannot(or cannoteasily) be relatedto these events?Second,to whatextent
are historical events reflected as changes in the values of government
bonds?Do capitalvalues rise or fall, andhow largeand significantarethe
changes? Are bonds issued by the various governmentsaffected in the
same or in differentways?
We arguethatthe answersto these questionsshed new light on the war.
Specifically,some eventsthataregenerallythoughtto be crucialareclearly
reflectedin thepricesof the bondsunderstudy.Thisholdstrue,in particular,
for the "official"outbreakof the war in September1939 (which depressed
the prices not only of Austrian,Belgian, and French,but also of German
governmentbonds),andfor losses andgainsof nationalsovereignty.When
Austria lost its independenceand became part of Grossdeutschland,for
example, the value of its sovereignbonds fell by 46 percent;when it regained its nationhoodat the PotsdamConference,their value rose by 12
percent. Similarly,when Belgium and Francewere overrunby German
forces in the Blitzkriegof May 1940,theirgovernmentbondsfell by no less
than35 percentand3 1 percentrespectively.Onthe otherhand,some events
to which historiansattachgreatattentionarenot reflectedin bondprices at
all: The most prominentexampleis Germany'scapitulationin 1945, which
did not affect Germangovernment-bondsprices.
HISTORICAL EVENTS, INTERPRETATION, AND CAPITAL MARKETS
Many historicalevents are generallyundisputedand their datingposes
few problems.In ourcontext,an examplewouldbe Hitler'sappointmentas
chancellor,which took place on 30 January1933 (andnot, say, in 1930 or
1936). Similarly,the Wehrmacht'sunconditionalsurrendertook place in
Reims on 7 May 1945, andwas repeatedin Berlinon 9 May (andnot, say,
in Marchor April). But even these events arenotjust facts;they are acts of
historicalsimplification.Thus, in the case of the Germancapitulation,it
could be arguedthattherewere still some Wehrmachtunits fighting after
these dates, so that a more appropriatedate for the capitulationwould be
later.Interpretationis thus a crucialelement in historicalresearch,and in
interpretingthe past greatcaremustbe takennot to distortit. In particular,
when past decisions are evaluatedthe knowledgeexisting at thatparticular
time must be takeninto account.This is particularlyimportantwhen decisions turn out to be wrong. For instance, it is not easy to understand, expost
facto, why Hitler ordered the invasion of the Soviet Union, because according to what we know in hindsight, defeat was almost inevitable. Scholars
make an effort to overcomethese dangersby carefulstudy of official and
private documents, such as diaries, which are likely to be representative of
the situationas perceivedat a particularmomentin time. But it is also a
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
470
Frey and Kucher
well-knownfact thatextantdocumentsare alreadythe resultof a selection
process(mainlythose documentsconsidered"important"
beingpreserved),
andmanyofthem arerewrittenafterwards.An exampleis Hitler's Tischgesprache im Fuhrerhauptquartier,
the transcriberof which, it is now known,
inadvertentlyinsertedargumentsand commentsmanyyears afterthe war.
It is possible thatthese laterinsertionsdo not fully reflectHitler's original
statements in 1940.1
Accordingto efficient-market
theory,capitalmarketsofferthreeparticular
advantagesover otherdatasources.2First,providedthey are correctlyrecorded(which is probablesince boursesarepublicor quasi-public),securities prices reflectthe situationobtainingat the given point in time. The future is not known, nor can it be incorporatedinto these dataretroactively.
What can be registeredare the decision makers' subjectiveexpectations
aboutthe future,which is a wholly differentmatter.Ourdatasetcapturesthe
moodexistingamonginvestorsata givenpointin time,forinstanceexpectations regardingthe likelihoodof Germanywinningthe warandhonoringits
foreigndebts.Second,investorsarelikely to evaluatecarefullythe prevailing situation,as well as any likely futuredevelopments,becauseerrorsdirectly affect theirpocketbooks.Even a Nazi sympathizerhadto weigh the
probabilityof default on, or repudiationof, Germangovernmentbonds
shouldGermanylose the war.Failureto do so incurreda greatriskof capital
loss. Thistoo distinguishescapitalmarketsfromotherdatasources,particularlysurveysandquestionnaires.A finaladvantageis thatfinancialmarkets
usually exhibit a high predictivepower,due to so-calledmarginaltraders.
This type of tradercarefullyassesses the relevantinformationandacts on a
relativelyunbiasedbasis. In the extremecase, one suchtradercan drivethe
marketpriceto the underlyingequilibrium.3
On the other hand, one must also bear in mind the limitationsof this
method.Most importantly,tradersof governmentbondsareonly interested
in the likelyfinancialconsequencesof politicalevents.Theyseekto evaluate
how a given eventaffectsthe probabilitythatdebtwill be servicedandultimately redeemed.Capitalmarkets,moreover,and especially the government-bondmarket,do not reflectthe generalstateof economicandpolitical
expectationsat a particularpoint in time. Rather,they capturethe expectations of a special groupof people, not only floor tradersbut also the much
largergroupof underlyinginvestors.It wouldbe of considerableinterestto
know exactly who the ultimatebuyersand sellers on the governmentbond
marketwere duringthe period in question.But no recordsexist as to the
1Urner,Schweiz.
See for exampleFama,"EfficientCapitalMarkets."
I On the marginaltraderand the Hayek hypothesis see Smith, "Markets";or Forsytheet al.,
2
"Anatomy."
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
471
identity of these persons;their characteristicscan be determinedonly in
generalterms(thatwill be exploredlater).
Ouranalysisof break-pointsdoes not identifyhistoricalfacts, but rather
the acquisitionand assessmentof informationrelevant for bondholders.
Investorsfollowedthenews not forits own sake,butin orderto predictwhat
wouldhappento thebondstheyownedorconsideredacquiring.Eventswere
evaluatedwith respectto theirimpacton theprobabilityof governmentdebt
service. Some events of the war years were deemed importantand thus
influencedbondprices,while othersdidnot affectthe perceivedprobability
of debt service, andthereforewere not reflectedin bondprices.
In Switzerlandduringthe war,informationwas very quicklyandreliably
disseminatedboth in the press and on radio.4This raises the question of
whetherfinancialmarketsmighthaveregisteredhistoricalfacts in advance.
In such a case, a breakwould be visible beforethe event, or completely
absent,dependingon the speed of adjustment.Eitherway, no breakwould
be visible at the date of the event itself. Thereexists suggestive evidence,
however,thatfinancialmarketstendto overreactto the arrivalof news.5The
overreactionhypothesisimpliesthateven if many investorshave predicted
an event well in advance,and financialmarketshave adjustedaccordingly,
a breakin the price series can still be identified.
THE GOVERNMENTBOND MARKET
DuringWorldWarII, as so oftenbefore,belligerentgovernmentsdirectly
or indirectlyintervenedin marketsundertheircontrol,includingfinancial
markets.In Germanyin particular,manyforeign-exchangerestrictionswere
either introducedor tightenedsoon after the Nazi takeoverin 1933.6 The
only marketwhere bonds of the governmentsunder considerationwere
freely tradedwas the Swiss bourse. For reasons of neutrality,the Swiss
governmentcontrolledneithertransactionsnor prices, and there were no
restrictionson foreigninvestors.Tradingwas haltedonly duringMay/June
1940, when it was unclear whether Germanforces would outflank the
MaginotLine to the north(throughBelgium andthe Netherlands)or to the
south (throughSwitzerland).
Manygovernmentsissuedbondsin Switzerlandduringtheinterwaryears.
Here we are only consideringobligationsof national governments.The
foreign governmentsthatborrowedmost in the Swiss capitalmarketwere
FranceandGermany,followed by BelgiumandAustria.The value at emis4 See for exampleSalis, Chronikor Moos, Grosse Weltgeschehen.
I See for exampleDe BondtandThaler,"Doesthe StockMarket?"
6 Manyof the Germancapitalcontrolshadbeenintroducedduringthe bankingcrisesin September
1931andwere only tightenedby theNazi government.TheNazis did,however,addsome new restrictions, such as those on transfersof interestpayments.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
472
Frey and Kucher
sion of Germany's sovereign debt was roughly SFr 3 billion, France's
SFr 3.6 billion, Belgium and Austria's SFr 1 billion and SFr 590 million,
respectively(all 1999 values).7Ouranalysisconsidersa weightedindex of
the valuesof all governmentbondsissuedin Switzerlandafter1922 foreach
of these four countries.It is importantto notethatall of the aforementioned
bonds were issued and tradedin Swissfrancs.Bondholderswere therefore
protectedagainst debased repayments.Exchange-ratefluctuationscould
theoreticallyalterthe probabilityof debt serviceby changingits real cost.
But since exchangeratesof most currencieswere fixed againstthe Swiss
franc duringWorldWarII (the sole exceptionbeing the U.S. dollar),the
lattereffect was probablyof slight importance.
Due to the largenumberof Swiss governmentbondstradedon this market, we restrictourselvesto thetwelve largestissues.A value indexhasbeen
constructedby comparing,the averagerateof returnon these bonds to the
average monthly rate of returnon the twelve largest Swiss government
bonds over the periodfrom 1906 to 1925.'
No informationis availableon who tradedat the Swiss stock exchange
duringWorldWarII. But as we have mentionedbefore, even if we knew
their identity,it would remainunclearwhose money they were investing.
Given the high degreeof opennessof the Swiss financialmarket,it seems
likely thatinvestorsfrom all over Europeused this "safehaven."Thereis,
however,some limitedinformationavailableconcerningthevolumeoftrading in governmentbondson the Swiss bourse.The Swiss NationalBankdid
not keep recordson turnoverin stocksor bonds;butturnoverwas taxedand
the returnshave been used to estimatethe extent of trading.Accordingto
HubertSchwab,tradein foreigngovernmentbonds fell from aboutSFrlggg
18 billion in 1937 to about 3.5 billion in 1943, reboundingthereafterto
about 7 billion in 1946.9Germanand Frenchgovernmentbonds each accountedfor roughly30 percentof the annualturnover,whereasthe respective sharesof BelgiumandAustriastoodat 7 and6 percent.In 1937 trading
in Swiss governmentbondsamountedto abouthalfthatof all foreignbonds.
During the war, investing in governmentbonds of the belligerentsgrew
7All amounts indicated in this paragraph are in 1999 Swiss francs. For the conversion of war-era
prices into 1999 equivalents we only took inflation into account. Since the Swiss CPI is nowadays about
6.9 times higher than during World War II, values at emission were multiplied by 6.9 in order to get
1999 Swiss francs. So, for example, the actual value for the 31 German government bonds at time of
emission was only roughly SFr 460 million. However, some researchers (such as Jost, Politikund Wirtschaft) point out that not only inflation, but also the development over time of national income, should
be taken into account when converting prices. This, of course, would yield considerably higher values
in 1999 Swiss francs.
8 For December 1939, for example, the average returnofthe twelve bonds was 4.25 percent. Comparing this to the 4.42 percent average return for the period from 1906 to 1925 yields an index of 104.00
((4.42/4.25). 100).
9Schwab,SchweizerischeEffektenmarkt.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
473
increasinglyrisky,so investorsshiftedmoneyintoSwiss governmentbonds.
The best estimatesavailableindicatean annualturnoverof approximately
9 billion in 1937, 4.5 billion in 1940, and 13 billion in 1946.
SFr1999
WorldWarII "officially"startedwith the Germaninvasionof Polandin
September1939, andendedin the Westwith the unconditionalsurrenderof
the Germanforcesin May 1945.In manyrespects,however,the war started
earlier,perhapswith the occupationof the Rhinelandby Germanyin March
1936, or the invasion of Czechoslovakiain March 1939. It could even be
arguedthatthewarwas a directconsequenceofthe Nazi takeoverin January
1933. To answerthe questionwhetherit makessense to look at this period
as a formof warprecedingthe officialdeclarations,we includemonthlydata
extendingfrom December 1933 (December1928 for Swiss bonds) to December 1948.10All datahave been collected from monthlyreportsof the
Swiss NationalBank."1
ECONOMETRIC METHODS
Our basic method is to search for structuralbreaks in these series of
government-bondprices. In contrastto an event study,the startingpoint is
not a list of dates, with the data then revealing which of them "matter."
Rather,the methodused here allows the datato speakfor itself, withouta
priori specificationof the dates.
The basic idea behindthe procedureused is to estimaterandomwalks
within small time windows and then to test for differencesin mean bond
prices between these windows. This will provide informationon threats
commonto all the countriesconsidered. Wile this methodmightbe useful
to answera varietyof questionsregardingevents thataffectedall bonds, it
also meansthat nothingcan be said aboutcountry-specificthreats.Therefore, in a second step, we test for country-specificthreatsby estimating
conditionalrandomwalks andthentestingfor differencesin the meanbond
pricesfor each coumtry.Theterm"conditional"mustbe stressedhere, since
the secondprocedurecorrectsfor factorsinfluencingall governmentbonds
10Unfortunately, higher-frequency data are not available. While weekly or daily observations are
econometrically unnecessary, monthly data might mislead: Suppose that Event A raised bond prices
early in the month, while Event B lowered them later on. The data will report this as a wash and we will
miss two potentially important events. Or the data might say that something happened in, say, November; but if several things happened in November, then one might fmd oneself at a loss to say which
particular event it was that moved the prices. However, while we cannot exclude the possibility that we
missed some dates in our study due to the usage of monthly data, we have never encountered two
important events in the month before a break-point. So while daily data might enable us to identify
events with greater precision, we do not think that they would lead us to new insights.
" Data for the Swiss series can be found in the Swiss National Bank's Monatsberichte, tables 13
(1928 to 1930), 12 (1931 to 1938 and 1942 to 1944), 9 (1939 to 1941 and 1947 to 1949), and 10 (1945
to 1946); indices for foreign government bonds are taken from tables 14 (1934 to 1938 and 1941 to
1946), 18 (1939), 17 (1940) and 12 (1947 to 1949).
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
474
Frey and Kucher
in a similarway. Wewill thereforenot find breaksin, say,the Germangovernmentbond index as a resultof changinginflationor real interestrates.
This holds also for mean differences:the second procedureonly tests for
significant structuralmean breaksin bond prices conditionalto the price
movementsof all governmentbondstradedin Switzerland.Forexample,an
eventthatled to a 10-percentincreasein Swiss government-bond
pricesand
to a 5-percentincreaseof the prices of all governmentbonds will increase
the conditionalmean of the Swiss bond prices by 5 percentonly. It also
meansthatan eventthathas the sameeffect on the bondpricesof all countries will not be recognized.12
In orderto find all possible datesfor structuralbreaks,we apply a fourstep procedurebasedon the workof AnindyaBanerjee,RobinLumsdaine,
James Stock, and PierrePerron.3 Similarmethodshave previously been
applied effectively by Russel Sobel, KristenWillard,TimothyGuinnane,
andHarveyRosen in orderto identifybreaksin exchange-rateseries."4For
a full accountof the technicaldetails,see the Appendix.
FROM STRUCTURALBREAKS TO HISTORICALFACTS
CommonThreats,1933-1948
As can be seen in Figure 1, thereis a strongdownturnin the index of all
governmentbondstradedin Switzerlandfromlate 1933 up to the outbreak
of WorldWarII. During the war, the index remainedrelatively stable at
around40 percentof par.One interestingfeatureis the peak in mid-1944,
just aboutwhen Allied forces invadedNormandy.
Theprewarfall in bondpricesmentionedpreviouslyis particularlyinterIn
esting since all five countrieswere maintainingtheirinterestpayments."5
the
Swiss
decided
on
a
devaluation
of
addition,
government
currency
approximately30 percent,as a consequenceof which the bankcould be expectedto loosen the monetarypolicy.16Sinceboththese factorsshouldhave
raisedbondpricesratherthanloweredthem,it seems likely thatthe steady
12
it will of courseshow up in the first(unconditional)estimationprocedure.
13 Banerjee,Lumsdaine,and Stock,"Recursiveand SequentialTests";andPerron,"GreatCrash."
14Sobel, "ExchangeRate";andWillard,Guinnane,andRosen,"TurningPoints."
Mostofthe countriesstoppedinterestpaymentsaftertheGermaninvasion.Forthe countriesunder
consideration,thiswas the case forBelgiumas well as fortheremainingpartsof France,which ceased
interestpayments
inthe summerof1940andNovember1942,respectively.Germanycontinuedinterest
paymentssomewhatlonger,untilJune1943.It is interestingto notethatnoneof the countriesformally
repudiatedits outstandingdebt, and all of them resumedinterestpaymentsafterthe war (for more
details,see the sectiontitled"PostwarFateof Bond Prices").
16On26 September1936thegovermmentrepealedthe
law requiringtheSwissNationalBanktoback
the francwith gold. Nonethelessthe bankwas still restrictedin its monetarypolicy by a government
requirementthatit maintainan exchangeratioof roughly205 mg. gold per franc.This decisionwas
takenabovetheBank'sobjections.See CrettolandHalbeisen,Wdhrungspolitische
Hintergriinde,
p. 14.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
475
80 -
70 -
60
50-
V
402030
20-
10
0
12/33 12/34 12/35 12/36 12/37 12/38 12/39 12/40 12/41 12/42 12/43 12/44 12/45 12/46 12/47 12/48
FIGURE 1
INDEX OF ALL GOVERNMENT BONDS TRADED IN SWITZERLAND, 1933-1948
Source: Swiss National Bank, Monatsberichte, 1933-1948.
decline had somethingto do with the ongoing war preparations.But it remainsunclear,primafacie, whicheventstradersdeemedimportant.Inorder
tojudge amongthe alternatives,the methodjust describedmaybe fruitfully
applied. Since in this section we are interestedin events that affected all
governmentbonds,we shall seek to isolatebreak-pointsin the index of all
governmentbonds by estimatingan unconditionalrandomwalk.
The four-stepprocedureidentifiesnine statisticallysignificantbreaksin
the time seriesof all governmentbonds;thesearedisplayedin Table1. Each
eventhas the "expected"sign. Withthe exceptionof the Swiss devaluation,
and to some extent the MarshallPlan Conferences,all of them are related
with eithercausing,prolonging,or shorteningthe war.Thetwo events indicatingan earlierendto the war-the Normandyinvasionand,of course,the
Germancapitulationitself-had a positive impacton bond prices. On the
otherhand,all eventsindicatingthattherewouldbe war,thatthe warwould
lastlonger,orthatit wouldinvolvemorecountriesthanpreviouslybelieved,
had a negativeimpacton the index. Most prominently,the actualoutbreak
of war reducedbond values by morethanone-quarter.
Most of the events identifiedwill be describedin detailin the following
sections.In this sectionwe will commentonly on two eventsthatcould not
be identifiedin anyofthe country-specificestimations.Thefirstis thedeval-
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
476
Frey and Kucher
TABLE1
STRUCTURALBREAK-POINTSAND CORRESPONDINGHISTORICALEVENTS:
ALL GOVERNMENTBONDS TRADEDIN SWITZERLAND
Date
(yyyy.mm)
Changein OverallBond Indexa
(percentage)
1935.03
1936.10
1939.02
1939.09
1941.12
1942.11
1944.06
1945.04
1947.07
a
-5.6***
+2.3***
-7.7***
-26.4***
- 1.7**
- 1.9**
+5.4***
+4.3***
-5.2***
MajorEvents
GeneralDraftin Germany
Devaluationof the Swiss Franc
GermanInvasionof Czechoslovakia
GermanInvasionof Poland
JapaneseAttackon PearlHarbor
RussianCounteroffensiveat Stalingrad
Allied Invasionof Normandy
GermanCapitulation(s)
MarshallPlan Conferences
Percentagechangein absolutemean.
** - Significant at the 5 percent level.
= Significantat the 1 percentlevel.
Sources: See the text.
uation of the Swiss francon 27 September1936, which had the expected
positive impacton bondprices.This "event"cannotbe identifiedin any of
the country-specificestimations,probablybecauseit hadapproximatelythe
sameeffecton all bonds,andthereforedisappearsintheconditionalrandomwalk analyses.
The second such event is the MarshallPlan Conference of July 1947.
The idea that a conference deciding on a program to rebuild Europe
should have had a negative impact on government bonds seems quite
paradoxical,but it was also seen to markthe beginning of the Cold War.
The first conference of the ThreePowers, in Parisfrom 27 Juneto 2 July,
was planned to conceptualize the proposals made by U.S. Secretaryof
State Marshall in June 1947. It soon turnedout that the differences between the United StatesandBritainon the one side, andthe Soviet Union
on the other,could not be resolved, andthe conference endedwithout the
intended results. This failure had importantconsequences: the common
Europeanprogram,which Marshallhadin mindwhenmakinghis proposals, hadturnedinto a WesternEuropeanprogramwith several anti-Soviet
elements.17 The confrontation culminated as first Poland and later all
countriesunderSoviet occupationwithdrewtheirpromisesto participate
in the follow-up conference.18The negative breakin the index of all government bond prices suggests that even in 1947 the importanceof this
withdrawal,which nowadays is widely regardedas the beginning of the
Cold War,was understoodby the capital markets.
17 Hardach,Marshall-Plan,pp. 48ff.
18ParrishandNarinsky,New Evidence.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
477
Germany, 1933-1948
Figure 2 shows the monthly price index of the 31 German government
bond issues traded on the Swiss bourse. Visual analysis reveals a secular
decline. The steep drop between 1933 and 1935 indicates that bondholders
feared that the Nazis would seek to renegotiate their foreign debts, or simply default. This fear was strongly nurtured by official pronouncements
advocating extreme autarchy; heavy-handed interventions in the capital
markets further depressed bond values.19A moratorium on Versailles reparations payments was declared in mid-1933, at the same time that the government redeemed many foreign bonds in an effort to become as autarchic
as possible.
The partial recovery in 1937 and 1938 may be attributed to the (shortterm) success of the expansionary fiscal policy that accompanied rearmament: national income picked up, and unemployment fell sharply. The ratio
of foreign to total government debt fell from 18.7 to 5.4 percent.20The Nazis
thereby regained some financial respectability with foreign investors. While
the German government amassed a huge internal debt to finance armaments
and other government expenditures, the probability of foreign-debt service
was considered to have improved.
But Hitler's aggressive foreign policy and increasing isolation led to
another drastic fall in German bond prices from mid- 1938 to the outbreak of
war in 1939. Bond traders feared that the impending war would reduce
Germany's willingness and ability to service its foreign debt. There was
again a rise -in the value of German government bonds after the successful
Blitzkrieg in the spring of 1940, but it did not last long: from the second half
of 1941 there was a permanent fall in German bond values, suggesting that
investors expected early on that the Nazis would lose the war, that the debt
would no longer be serviced, and that the capital invested would be lost.
Econometric analysis identifies six break-points for Germany. Table 2
gives a survey of the resulting break-points and the corresponding percentage changes in the conditional mean price index. Germangovernment bonds
experienced a large and statistically significant surge beginning in the summer and autumn of 1936. In July/August ofthat year, the conditional average
index rose by more than 7 percent relative to the conditional mean (that is,
to the index of all other government bonds tradedon the Swiss market). This
might be attributed to the Olympic Games in Berlin, which took place in
August and which made the Nazi regime look peaceable to many observ'1 German
economicpolicy in the prewarandwarperiodis discussedin Boelcke, Kosten;K6llner,
Militar und Finanzen;Milward, War,Economyand Society; Federau,Zweite Weltkrieg;Fischer,
andErbe,NationalsozialistischeWirtschaftspolitik.
Wirtschaftspolitik;
20Erbe,NationalsozialistischeWirtschaftspolitik,
p. 51.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
478
Frey and Kucher
70
60 -
50
~a400.
3 30 -
20 -
10
0
12/33 12/34 12/35 12/36 12/37 12/38 12/39 12/40 12/41 12/42 12/43 12/44 12/45 12/46 12/47 12/48
FIGURE2
INDEX OF 31 GERMAN GOVERNMENT BOND ISSUES TRADED IN SWITZERLAND,
1933-1948
Source: Swiss National Bank, Monatsberichte, 1933-1948.
ers.21 ThemarketremainedbullishthroughJanuary1937, at whichpointthe
boom was particularlymarked.
In March 1939 Germanyinvadedthose partsof Czechoslovakianot already ceded at the Munich Conferencein September1938. Accordingto
manyhistorians,this heraldedthe beginningof WorldWarII.22 The governof history.The value of Germentbondmarketssupportthis interpretation
mangovermment
bondsfell by no less than17percentcomparedto the average marketvalues, indicatingthattradershad lost even more confidencein
the Germangovernment'scapacityto service its debts. The invasion of
Czechoslovakiawas the first time Hitlerannexednon-"German"
territory,
which was taken as an indicationthat he would not stop there, and that a
majorwarwas likely.However,someuncertaintyremained;some actorson
capitalmarketsobviouslythoughtthatthis conquesthad satisfied Hitler's
demands.Accordingly,the value of Germangovernmentbonds dropped
only half as much as they would at the "official"outbreak.
21Forexample,the Frenchdelegationusedthe fascistsaluteuponenteringthe stadiumatthe Olympic openingceremony.
22
See for exampleWeinberg,Worldat Arms.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
479
WorldWarII
TABLE 2
STRUCTURAL BREAK-POINTS AND CORRESPONDING HISTORICAL EVENTS:
GERMANY
Date
(yyyy.mm)
1936.07
1939.03
1939.09
1941.12
1942.11
1945.02
a
Change in German Bond Indexa
(percentages)
Major Events
+8**
- 17**
-39***
-5**
-7***
-34***
Olympic Games in Berlin (30 July-16 August)
Invasion of Czechoslovakia (15-16 March)
Invasion of Poland (1 September)
Japanese Attack on Pearl Harbor (7 December)
Russian Counteroffensive at Stalingrad (November)
Yalta Conference (4 -11 February)
Percentage change in the conditional mean (that is, the parameter Yrfrom equation A2).
** = Significant at the 5 percent level.
= Significant at the 1 percent level.
Sources: See the text.
Worldwar became a reality after 1 September 1939, when Germantroops
invaded Poland. Since the end of 1938 capital markets had been interpreting
Nazi actions in a strongly negative way. The actual start of the war sent the
value of German government bonds plummeting 39 percent. Obviously,
traderswere extremely pessimistic about the prospects of a German victory.
As already noted, the Swiss bourse was closed in May/June 1940, so the
immediate effect of the German Blitzkrieg victories are not reflected in our
data. But Figure 2 clearly shows that the average price of German government bonds rebounded to prewar levels. It is worth noting, however, that it
did not surpass them. This may be interpretedto indicate that after the Blitzkrieg peace was considered a likely prospect, with "normal"prewar conditions expected to resume.
The fourth structuralbreakis identified in November/December 194 1, but
the decline of average bond prices is rather small (around 5 percent). It
reflects another major war event, namely the Japanese attack on Pearl Harbor (7 December) and the consequent war declarations of the United States
(and the United Kingdom) on Japan, and of Germany (and Italy) on the
United States (8 and 11 December, respectively).
Yet another signiflcant drop in German bond values (about 7 percent)
occurred in November 1942. In that month the Soviet army launched a massive counteroffensive against the German invaders. More than 300,000
German troops were encircled at Stalingrad. Traders on the Swiss bourse
evidently considered the counteroffensive as even more detrimental to Germany's ability to service its debt than was the capitulation by Field Marshal
Friedrich Paulus three months later (2 February 1943).
The last break-point indicated by the German data occurred towards the
end of the war, in February 1945. At the Yalta Conference the Allied great
powers decided that only a complete capitulation of all German forces on all
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
480
Frey and Kucher
120-
100
80
60
40
20
0
12/33 12/34 12/35 12/36 12/37 12/38 12'/39 12'/40 12'/41 12/42 12/43 12/44 12/45 12/46 12'/47 12/48
FiGuRE3
INDEX OF 9 AUSTRIAN GOVERNMENTBOND)ISSUES TRADEDIN SWITZERLAND,
1933-1948
Source:Swiss NationalBank,Monatsberichte,1933-1948.
fronts would be accepted,and that Germanywould be divided into three
militaryoccupationzones.This was interpretedto be a decisive blow to the
Nazis-more so even thanthe formalcapitulationof the Germanmilitaryin
May 1945-and resultedin a fall of Germanbondpricesby 34 percent.
Austria,1933-1948
Figure3 tracksthe monthlypriceindex of the nine Austriangovernment
bond issues traded in Switzerland.In contrastto Germany,it shows a
markedincreasebetween 1933 and 1937. Therewas a huge dropwith the
Anschiussof March1938, and it remainedquitedepressedthereafter.This
drasticfall may be due not only to politicalfactors,but also to the fact that
with annexation,they became subjectto the severe Germancontrols on
capitalandforeignexchange.It is worthnoting,however,thatthe Austrian
index remainedmuch below its Germancounterpartuntil mid-1944, even
thoughGermanyhad formallyassumedall Austrianforeigndebts.
Econometricanalysisof the Austriangovernmentbond index identifies
threestatisticallysignificantbreak-points(see Table3). The indexfell by no
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
481
TABLE 3
STRUCTURAL BREAK POINTS AND CORRESPONDING HISTORICAL EVENTS:
AUSTRIA
Date
Change in Austrian Bond Indexa
(yyyy.mm)
1938.03
1939.09
1945.08
a
(percentages)
Major Events
-46***
-46***
+12**
German Annexation of Austria
German Invasion of Poland
Potsdam Conference
Percentage change in the conditional mean (that is, the parameter Y, from equation A2).
** = Significant at the 5 percent level.
= Significant at the 1 percent level.
Sources: See the text.
less than 46 percent upon the Anschluss in March 1938. A significant drop
is visible as of the beginning of the year, when the Nazi government prepared that event. It is noteworthy that traders on the Swiss bourse did not
consider the seemingly enthusiastic Austrian supportfor the Anschluss to be
relevant to their interests. The same holds for its near-unanimousratification
in a plebiscite undertaken on 10 April of the same year.23
As with German debt, the outbreak of war deeply depressed Austrian
government bond values (by 46 percent in September 1939). Oddly, the
German capitulation of May 1945 is not reflected in these data. One reason
might be that Austria's political future, and thereby the servicing of its debt,
was taken to be uncertain;traderswere unable to predict clearly how defeat
would affect that part of the Reich which had, after all, been annexed by the
Germans. This uncertainty was mitigated in July/August of the same year
when the Potsdam Conference settled on Austrian independence, a decision
reflected in a 12-percent increase in average bond prices.
France, 1933-1948
The raw data for French government bonds (Figure 4) show a fairly stable
value up to mid- 193 8, followed by huge drops coinciding with the "official"
outbreak ofthe war, the invasion by German forces, and the French capitulation (22 June 1940). After trading resumed at the Swiss bourse, French
bonds experienced a secular increase in value up to the end of 1945. It is
interesting to note that the value of French government bonds remained
above 20 percent of par, even though France suspended interest payments
in November 1942 and did not resume debt service until after the period
considered. The fact that French government-bond prices did not drop to
zero implies a surprising degree of confidence that France would reemerge
as an independent nation and resume debt service.
23 Approximately 4,453,000 ofthe 4,484,000 electors voted "yes," only 11,924 voted "no," and 5,776
spoiled their ballots. See Henschy, Freedom.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
482
Frey and Kucher
120
100
,
80-
60
60
40
20
0
12/33
12/34
12/35
12/36
12/37
12/38
12/39
12/40
12/41
12/42
12/43
12/44
12/45
12/46
12/47
12/48
FIGURE4
INDEX OF 12 FRENCH GOVERNMENT BOND ISSUES TRADED IN SWITZERLAND,
1933-1948
Source: Swiss National Bank, Monatsberichte, 1933-1948.
Oureconometricprocedureidentifiesfive statisticallysignificantbreakpoints in the Frenchseries (Table4). Frenchgovernment-bondvalues suffered a blow when the Germansoccupied the demilitarizedRhinelandin
May 1936.Investorsmayatthispointhavelost someconfidencein France's
willingnessandabilityto resistNazi aggression.The "official"outbreakof
war reducedprices still further,as did the defeat of May/June1940. The
Allied invasionof Normandyin June1944was greetedas a sign of military
and politicalrecovery,andraisedFrenchgovernmentbondvalues.
TABLE4
STRUCTURAL BREAK POINTS AND CORRESPONDING HISTORICAL EVENTS: FRANCE
Date
(yyyy.mm)
1936.05
1939.09
1940.05
1944.06
1946.01
Change in French Bond Indexa
(percentages)
-4**
-25***
[-31]b
+16***
-14***
Major Events
German Occupation of the Rheinland
German Invasion of Poland
German Invasion of the Low Countries and France
Allied Invasion of Normandy
None Identified
Percentage change in the conditional mean (that is, the parameter y, from equation A2).
in the bond values between the day when tradingwas stopped and when it was resumed. For
methodological reasons it is not possible to identify such "breaks"by the econometric techniques used.
** = Significant at the 5 percent level. *** = Significant at the 1 percent level.
Sources: See the text.
a
bDifference
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
483
140-
120 -
100
ct 80-
60-
40 -
20-
12/33 12/34 12/35 12/36 12/37 12/38 12/39 12/40 12/41 12/42 12/43 12/44 12/45 12/46 12/47 12/48
FIGuRE5
INDEX OF 11 BELGIAN GOVERNMENT BOND ISSUES TRADED IN SWITZERLAND,
1933-1948
Source: Swiss National Bank, Monatsberichte, 1933-1948.
Belgium, 1933-1948
The value of Belgian government bonds traded in Switzerland exhibited
marked variations (Figure 5). A strong increase from 1934 to 1937 was
followed by an even stronger fall, to about 30 percent of par in 1940. Bond
values then recovered over the remainder of the war, right up to 1947.
TABLE5
STRUCTURAL BREAK POINTS AND CORRESPONDING HISTORICAL EVENTS:
BELGIUM
Date
(Yyyy.mm)
1937.09
1939.08
1940.05
1943.02
1944.06
1945.04
Change in Belgian Bond Indexa
(percentages)
-
-3**
10***
35]b
+10**
+6**
+7**
Major Events
None Identified
German Invasion of Poland
German Invasion of Low Countries and France
German Capitulation at Stalingrad
Allied Invasion of Nornandy
German Capitulation(s)
Percentage change in the conditional mean (that is, the parameter y, from equation A2).
Difference in the bond values between the day when tradingwas stopped and when it was resumed. For
methodological reasons it is not possible to identify such "breaks"by the econometric techniques used.
** = Significant at the 5 percent level. * - Significant at the 1 percent level.
Sources: See the text.
a
b
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
484
Frey and Kucher
Our econometricanalysisidentifiessix break-pointsin this price series
(Table5). The "official"startof the war,and to a much greaterextent the
Germaninvasionof May 1940, sent pricesplummeting.Allied victories at
Stalingrad(February1943), on the beachesof Normandy(June 1944), and
at the very end of the war (April 1945) predictablypushedup the values of
Belgian governmentbonds.
Switzerland,1928-1948
The overallvalue of Swiss governmentbonds shows a long-termrise of
about30 percentoverthe twentyyearsfrom 1928to 1948 (Figure6). Values
tendedto fall in the 1930s. The strongincreasein value in 1936 can be attributedto a devaluationof the Swiss currencyin September.However,this
economiceventdoes not correspondto a statisticallysignificantbreakin the
data,most likely becauseit also affectedthe values of all otherbondstraded
in Switzerland.Pricesexperienceda markeddropin the threeyears before
the "'official"outbreakof the war, until the Blitzkriegof May 1940. After
tradingwas resumedlaterthatyear, Swiss governmentbonds increasedin
value untilthey regainedthe prewarlevel of 1936/37.
160
140
120
100
80
60
40
20
0
12/28
12/30
12/32
12/34
12/36
12/38
12/40
12/42
12/44
12/46
FIGURE 6
INDEX OF 12 MAJORSWISS GOVERNMENTBOND ISSUES, 1928-1948
Source:Swiss NationalBank,Monatsberichte,1928-1948.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
12/48
WorldWarII
485
TABLE6
STRUCTURAL BREAK POINTS ANTDCORRESPONDING HISTORICAL EVENTS:
SWITZERLAND
Date
(yyyy.mm)
1933.04
1935.03
1936.09
1939.10
1940.06
1941.06
a
Change in Swiss Bond Indexa
(percentages)
-4***
-6***
+7***
+3**
+4***
+4***
Major Events
Establishment of the Nazi Dictatorship (24 March)
General Draft in Germany
Olympic Games in Berlin
German Invasion of Poland
German Invasion of the Low Countries and France
German Invasion of the Soviet Union
Percentage change in the conditional mean (that is, the parameter y, from equation A2).
** = Significant at the 5 percent level.
= Significant at the 1 percent level.
Sources: See the text.
Econometric analysis reveals six statistically significant break-points,
summarized in Table 6. The consolidation of Hitler's dictatorship in the
spring of 1933, as well as the reintroduction of the general draft in March
1935, were considered to be negative events from the point of view of investors in Swiss government bonds. The Olympic Games in August 1936 gave
the Nazi government a convenient propaganda forum; Swiss bond values
jumped in September 1936, which may be interpretedas a sign that Hitler's
goverment was thereby able to gain some goodwill with investors (as well
as with many British, French, and Italian politicians). In view of Switzerland's neutrality,the "official" outbreakof war in September 1939 increased
Swiss government bond values the following month. It is likely that funds
were shifted into Swiss government bonds, which seemed to be safer than
those of the four other countries considered here (all of whose values fell).
A similar pattern appeared in June 1940, after the German invasion of the
Benelux countries and France, and again after the invasion of the Soviet
Union in June 1941. In both cases, the attacks were directed at countries
other than Switzerland, so that this country's position as a safe haven improved. In particular, Hitler's decision to outflank the Maginot line in the
north, instead of south through Switzerland, was a lucky event.
HISTORICALFACTSAND GOVERNMENTBOND VALUES
This section analyzes first, whether historical "facts"related to World War
II, and generally considered to be importantby historians, show up as statistically significant break-points in the government bonds ofthe five countries
considered, and if so when; second, in what direction, and by how much, the
bond values changed. And third, as pointed out in the introduction, it is also
importantto consider what historical "facts"are not reflected as break-points
in government bond values.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
486
Frey and Kucher
The literaturewe have consulted takes the following seven events to
have been turningpoints in WorldWarII:the Germaninvasion of Poland
on 1 September1939, markingthe "official"outbreakof war;the German
invasion of Benelux andFrancein May 1940; the Germaninvasion of the
Soviet Union in June1941;the warentryofthe UnitedStatesfollowing the
Japaneseattackon PearlHarborin December 1941;the Germandefeat at
Stalingradin February1943; the Allied invasion of Normandy in June
1944; and the Germancapitulationof May 1945, markingthe "official"
end of the war.24
One interestingquestionwe are in a position to answer is whetherthe
datesmarkingthe "official"beginningandend of WorldWarII correspond
to the evaluationsof investors.As virtuallyall historiansagreethatthe war
was initiatedby theNazis, we includethe followingseven importanthistorical eventsoccurringin Germanyin theperiodbeforethe"official"outbreak:
Hitler'sappointmentas chancellorin January1933,andthe EnablingAct of
March 1933 which gave Hitler essentially unlimitedpower; the Rohm
Putschof June/July1934, wherebyHitlervitiatedthe SA andreestablished
the Wehrmachtas the sole militaryforce;the reintroductionof the general
militarydraftin March1935;the invasionof the demilitarizedRhinelandin
March1936;the OlympicGamesin Berlinin JulyandAugust 1936;theAnschluss of Austriain March 1938; andthe invasionof Czechoslovakiain
March1939,wherebyHitlerbrokehis formalpromise,given atMunich,that
the Sudetenlandwas his last territorialambition.
In the samevein, the followingtwo historicaleventsareincludedin order
to test whetherthe "official"end of the waralso markedthe end according
to capital-marketdata:the YaltaConference,where the principleof Germany's unconditionalsurrender,andits divisionintothreesectorsof occupation,was decided.This conferencetook place in February1945, thatis,
andthePotsdam
beforecapitulation,butit referredto postwararrangements;
Conferenceof August 1945,where(amongotherissues)Austria'srebirthas
a nationwas decided.
Table 7 lists the 16 historicalevents just mentionedand indicates the
statisticallysignificantchangesin the government-bondvalues of the Axis
powers(GermanyandAustria),the neutralcountries(SwitzerlandandBelgium), and one Allied nation (France).The table speaks for itself and it
suffices thereforeto concentrateon the most importantaspects. Only one
event,the "official"outbreakof the war,produceda statisticallysignificant
break-pointin all five countriesin our sample.It greatlyreducedthe value
of the governmentbonds of all belligerents:investorsthus did not "picka
winner"butconsideredthe warto be a generalthreatto theirassets.Neutral
24In view of the capital-market
dataavailable,we restrictourattentionto the Europeantheaterof
WorldWarII. A comprehensiveoverviewcan be foundin Weinberg,Worldat Arms.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
487
WorldWarII
b
a
n.d.
War
Date
The=
The
Prewar
Postwar
Sources:
"breaks"no
1936.03
1939.03
1935.03
1938.03
1945.05
1939.09
1944.06
1941.06
1941.12
1945.08
1945.02
(yyyy.mm)
1940.05a
Seewith Swiss
data
1933.01-03
1934.06-07
1936.07-08
thethedifference
in Stock
1942.11-1943.02
text. the
available
bond
Exchange
econometric
was
Yalta
values
U.S.
Battle
Rohm
Allied
General
Invasion
Invasion
Invasion
Olympic
Potsdam
of
Outbreak
Surrender
Anschluss
closed
Occupation
of
of of of
techniques
of
of
Putsch
10
Establishment
of Draft
the
between
Gamesin
theInvasion
Declaration
War
of
Conference
in
used.
in
theMay
Stalingrad
of
Benelux
Austria
Conference
Event
Nazi
Soviet
day
War and
Rheinland
BerlinGermany
through
Czechoslovakia
Wehrmacht
Union
when
Normandy
8
FROM
France
July
Dictatorship
trading
1940
"WAR-EVENTS"
TO
was
and
there
stopped
andwas
no
when
it trade
-37
-7 -5
+8 -39
-17
+8
n.d.
(percentage
Germany TABLE
7
BREAK-POINTS
IN
changes)
waswith
-46
+12
-46
n.d.
French
resumed.
or
For
Austria
GOVERNMENT
Belgian
BOND
Bonds
+4 +4 +3
+7
-6
-4
SERIES
Switzerland
methodological
between
reasons
it May
not1940
and
possible
to
February
identify
1941
such
+7 +6 +10
-35-10
n.d.
Belgium
+16
-3lb-25
-4
n.d.
France
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
488
Frey and Kucher
Belgium was also negativelyaffected,probablybecauseinvestorsthought
thatit would likely be drawninto a militaryconflictbetweenGermanyand
France,a predictionthatprovedcorrect.Thefallingpriceof bondsissuedby
belligerentgovernmentsstandsin contrastto a modestincreasein value of
Swiss governmentbonds,becausethatcountrywas considereda (relatively)
safe haven.Econometricanalysisthusleadsto the same evaluationas does
traditionalhistoricalresearch.It is worthnoting,however,thatmajorevents
beforethatdateare also clearlyreflectedon the bondmarket,most importantlythe Germaninvasionsof the Rhinelandandof Czechoslovakia.
Anotherset of historical"facts"clearlyreflectedin the bondmarketare
major changes in nationalsovereignty.When a governmentdisappeared
(Austriain 1938, Czechoslovakiain 1939, and Germanyin 1945) or was
oecupied(BelgiumandFrancein 1940), the priceof its debtexperienceda
very markeddrop.Conversely,when a governmentreemerged(Austriain
1945), its bondvalues rose sharply.
Severalwartimeevents consideredimportantby historiansare reflected
as significantbreak-pointsin bondvalues.In additionto the majorGennan
invasions (Rhineland,Austria,Czechoslovakia,Benelux, France,and the
SovietUnion),thisalsoholdstruefortwo engagementsgenerallyconsidered
decisive in Geimany's ultimatedefeat:Stalingradand Normandy.On the
other hand,viewed fromthe bond market,the end of WorldWarII is defined less by the Germansurrender(which affectedBelgian bond values
only) than by the Yaltaand PotsdamConferences,where the fate of the
defeatedcountrieswas decided.
THE POSTWAR FATE OF BOND PRICES
The readermay wonderwhathappenedto bondpricesafterthe war.Was
the marketcorrectin its assessment-suggested by the very low value of
bondpricesat the end of the war-that most Europeancountrieswould not
servicetheirdebtsfor a considerabletime to come?
Table8 depictsthe value of bonds issued by four of the five European.
governmentsunderstudy.(Switzerlandis excludedbecauseit neverstopped
interestpayments,andvaluesfortheSwissgovernmentbondsremainedabove
par almostthroughoutthe war.)The most obvious featureis that the four
countries'bondpricesdevelopedvery differently.WhileBelgiumresiuned
interestpaymentsalmost immediatelyafter tlhewar (such that its bonds
reachedparas earlyas 1946),Germany,for example,didnot resumegeneral
debtservice until 1954. Commonto all countriesunderconsiderationis the
factthattheirpostwargovernmentsacknowledgedall sovereigndebtanddid
eventually resume service. None of the countries, however, offered investors
compensation for the interestpayments foregone during the war.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
489
TABLE 8
VALUES OF FOREIGN-GOVERNMEN'T BONDS TRADED IN SWITZERLANI) AFTER
WORLD WAR II
(percentageof par)
Date
(yyyy.mm)
Gennany
1945.12
1947.12
1949.12
1951.12
1953.12
1955.12
21.18
16.72
39.50
48.15
89.89
99.84
Austria
France
Belgium
39.99
30.40
49.03
64.09
102.95
108.54
59.67
40.41
44.65
73.72
78.70
83.89
90.45
97.97
101.52
99.37
104.77
103.85
Source#:Swiss NationalBank,Monatsberichte,1948-1956.
There is a large literatureon when and why governmentsrepudiate
debts.25There are several models of the conditions underwhich regimes
decideto repudiatedebtsincurredby eitherearlierregimesor in the service
of aims for which they do not thinktheirpeople shouldpay (such as fighting againsttheNazis). In the light of these models, it seems quiteclearthat
those countrieswhich relied most heavily on new foreign creditstried to
resume payments as soon as possible. In our sample, this was the case
especially for Belgium and(to a lesser degree)Austria.As a consequence,
Belgian governmentbonds had already reached par in 1946, Austrian
bonds in 1952.
As is well known, Germanylost a greatdeal of its productivecapacity
duringthe war(and,in the Sovietzone,thereafteras well). Whilea currency
reformwas undertakenin the Westernzone in 1948to cope withthe ensuing
hyperinflation,rationingcould not be totally lifted until 1950. Until the
the new Germangoverncurrencyreformin 1948, it was unclearwvhether
mentwould pay its predecessor'sforeigndebt.As a consequence,the price
of its bonds fell to as low as 15 percentof par.It was only in August 1953
thatWestGermanybeganto service Englishand Swiss foreign debts;full
service on all foreigndebtswas resumedin the thirdquarterof 1954.26
In France,governmentdebt had tripledbetween 1939 and 1945 while
industrialproductionfell by 80 percent.In orderto cope with the resulting
high inflation,the post-Vichygovernmentconducteda currencyreformin
1946, accompaniedby the introductionof heavytaxes on capital.The result
was a deeprecessionin 1947/48,whichthe governmenttriedto combatwith
heavyinterventionsfinancedthroughnew creditsfromtheUnitedStatesand
fromthe EuropeanRecoveryProgram(ERP).The Frenchgovernmentonly
resumedservicingits foreigndebtat the end of 1949. As can be seen from
25Foragood surveyon how debtswererepudiatedinthe 1930s,see EichengreenandPortes,"Debt."
26 See for exampleDie
Wirtschaftslage,
1953-1954.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
490
Frey and Kucher
Table8, bondtradersneverthelessremainedquitepessimisticaboutFrance's
long-termcapacityto service its foreigndebtrightup to the end of 1955.
CONCLUDING REMARKS
The approachsuggestedhere focuses on the capitalmarketand seeks to
identify statisticallysignificantbreak-pointsreflecting historical events.
This allows a quantitativeassessment of beliefs at a particularpoint of
time, uncontaminatedby later events that might otherwise influence the
evaluation.This approachhas been appliedto an importantrecentperiod
in Europeanhistory,from Hitler's rise to the MarshallPlan. It has been
made possible by the fact thatthe Swiss marketfor the bonds of various
governmentsinvolved in the waressentiallyfunctionedwithoutregulatory
interventionsduringthe whole period.Analyzing break-pointsin the values of governmentbonds obviously focuses on a very specific area, and
can thereforeonly captureone aspectof the wartimeexperience.The analysis suffersalso froma lack of informationon the buyersandsellers of the
governmentbonds tradedin Switzerlandduringthe war. But it does not
follow thatcapital-marketdataprovidea systematicallydistortedpicture.
Persons and institutionsactive on capitalmarketshave a directpecuniary
incentiveto evaluatethe prospectsofthe bondsofthe variousgovernments
as "objectively"as possible. They cannotafford-at least not in the long
run-to follow theirown politicalpreferenceswith respectto the countries
involved in the war,becausesuch actionwould resultin systematiclosses.
Wehave analyzedthe value of governmentbondsof five countriesmore
or less affected by the war over the period 1933 to 1948. We find thatthe
statistically significant break-pointsidentified by the econometrictechnique describedin the Appendixreflect historicalevents also considered
importantby historicalresearchbased on more conventionaltechniques.
We find, in particular,that the Germaninvasion of Poland prompteda
major downward shift in the bond values of all four countries directly
involved in the war,while Switzerlandbenefitedmodestly.Investorsconsidered it more likely that the Swiss governmentwould be betterable to
service its debt thanwould Germany,Austria,France,or Belgium.
Significant changes in the value of governmentbonds also occurred
whennationswereinvaded,as happenedto Austriain March1938, Czechoslovakiain March1939,BelgiumandFrancein May 1940,the SovietUnion
in June 1941, andFranceagainin 1944. The final capitulationof the Wehrmachtin May 1945 did not affectgovernmentbondvalues (except for Belgium's), which suggeststhatthe Germandefeatwas predictedmuchearlier
by the traders,and was thereforealreadyreflected in bond prices. More
relevantwas the decisionof the Allied powersat Yaltato acceptonly a total
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
491
capitulation on all fronts. The Potsdam decision to restore Austria's statehood was predictably associated with a significant rise in the price of that
govemment's bonds.
Analyzing breaks in capital-marketdata are of most use when considered
along with events identified and interpretedby more conventional historical
research. This article has followed two different tacks: from the break-points
determined by the econometric methods to historical events, and from the
major historical events to break-points. In both cases the correspondence is
incomplete-some break-points and their signs remain difficult to link with
historical "facts." There are several reasons why such deviations may have
occurred: we may be ignorant of important "facts" of which professional
historians are well aware; historians are themselves unaware of the "facts,"
or have chosen to ignore them; or no such "facts" exist, at least not in the
grand arena of high politics: specifically, break-points may have been driven
by purely financial news.
Conversely, there are a number of reasons why historical "facts" may
not show up as break-points: First, bond-market data may be of poor
quality, for instance if there are too few transactions. Second, governments may have intervened in the bond market as buyers or sellers, or by
imposing controls of one sort or another. An important case in point
occurs when governments want to prevent the reflection of a political (or
economic) event on the financial market. What might particularly affect
our analysis are changes in capital-market restrictions. Third, the econometric technique applied may be unable to identify break-points relating
to historical events, even though they are in the data. Fourth, a "fact" may
be important from the historian's point of view (it relates to the fate of a
nation, country, or population), but does not affect government debt
service. And fifth, the "fact" may not exist, or is not as important as historians believe. Here the quality of historical research is in question.
However, it would be misleading to assume that all historians identify the
same "facts" as important. So the issue is which historical school or
which individual historians have identified which historical "fact," as
well as the importance attributed to it.27
One of the next steps in research should be to identify which of these
reasons are relevant, and under which circumstances. The purpose of this
article has been to suggest that such an interaction, between quantitative
and qualitative historical research, and between capital market data and
data derived from other sources, is capable of providing new insight into
historical processes.
27 See for exampleKozicki,Developments.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
492
Frey and Kucher
Appendix:EconometricEstimationProcedure
Ourapproachis designedto fimdstructuralbreaksin the series of bond prices. In what
follows, we will only discuss technical details of the second methodapplied,that is, the
conditionaltests.Fortheunconditionalestimation,we adaptedtheprocedureby simplynot
correctingfor the overall index.All othersteps involvedremainthe same.
To address this task, we follow a sequential test procedurebased on Banerjee,
Lumsdaine,Stock,Sobel, Willard,Guinnane,andRosenin theiranalysesofthe Greenback
market.28In orderto find all possible structuralbreaks,a four-stepprocedureis applied.
Using data from a 36-monthwindow startingin December 1933 (December 1928 for
Switzerland),we first estimatethe regression
+e,
Inp,=A0+/8Inp,_ +32 1np,_1
(Al)
wherep, standsfor the index-valueof all governmentbondsof the countryconsideredon
datet, p, is the indexof all governmentbondstradedin Zurich(whichwe use as a measurementofthe marketperformanceas a whole),the,/'s arethe parametersto be estimated,and
e is a white-noise errorterm.A Waldtest associatedwith the hypothesisthat there is a
structuralbreakin the middle of the window is then calculated.The idea behindstep one
is to estimate a randomwalk and then check for changes in the constant, which is the
procedurefollowed in recent stock-marketstudies.It implies that bond prices follow an
exponentialBrownianmotion.29(Wealso ranregressionwith autoregressiveprocessesof
up to the sixth order,but did not find differentresults.)The inclusion of a measure of
marketperformanceas a right-handvariableallows us to estimatethe randomwalk ceteris
paribus:forexample,we allow forfactorsthatmightinfluencethevalue of all bondstraded
(such as fluctuatingreal interestratesand inflation).
In a second step, we estimatethe regressionagain,this time using a 36-monthwindow
beginningone monthlater.This step is repeateduntilthe entireperiodhas been covered.
An exampleis given in AppendixFigure1, which depictstheF-statisticsfromall the Wald
tests for Germany.By searchingfor peaks in the series of F-statistics,the first two steps
identifyseven datesfor Germany,five for Austria,seven for Switzerland,six for Belgium
and five for France,where the null hypothesis of no structuralbreaks is most strongly
questioned.The third step consists of choosing the windows aroundthese dates. As an
example,a time window aroundFebruary1945 is markedin AppendixFigure 1.
In the fourthstep,we test for statisticallysignificantstructuralbreakswithineach of the
windows isolatedin stepthree.Wedo thisby estimatinga seriesofthe following equations,
which in comparisonwith equationAl have been extendedby a dummyvariable:30
Inp,=A+AInpIj+fl2InP
l+YsDs, +?
with s = 6, ..., 42
(A2)
where Ds, equals one if date t is on or afterdates, and zero otherwise.The parameterys
measuresa changein the conditionalmean(thatis, a shift in the meanprice index ceteris
paribu,s)that occurson dates. Since all the prices are in logs, yscan be interpretedas the
percentagechangein the conditionalmean.WeestimateequationA2 repeatedly,eachtime
moving s by one month.For each resultingequation,we test whether ys is differentfrom
zero using a conventionalF-test. The date associatedwith the highestF-statistic is then
28Banerjee,Lumsdaine,andStock,"RecursiveandSequentialTests";Sobel,"ExchangeRate";and
Willard,Guinnane,andRosen, "TurningPoints."
29An overviewcan be foundin Duffie, DynamicAssetPricing.
30 As suggestedby Perron,"GreatCrash."
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
493
WorldWarII
50
45-
4035-
3025-
1~~~~~~~~~~~~~~~~~~~~~...........
so 1...............................................................
sb
"
11-.1................ .......................
20-
15
10
5
0.
12/33 12/34 12/35 12/36 12/37 12/38 12/39 12/40 12/41 1/2
12/43 12/44 12/45 1~/46 12/47 12/48
APPENDIX FIGURE 1
F-TESTS FOR STRUCTURALBREAKSIN THE INDEX OF GOVERNMENTBOND
PRICES:GERMANY,1933-1948
Source:See the text.
designated as the date where the most importantmean shift took place within each window.
Since sequential break tests cannot identify breaks around the beginning or end of a sample,
we add six observations at the beginning and at the end of the windows examined. So, for
the first equation estimated in step four, s is set at date six ofthe new window (which equals
date one in the original window). As an example, the results of step four for the window
isolated in Appendix Figure 1 are shown in Appendix Figure 2.
Three furtherpoints warrant comment: first, applying only the last step ofthis procedure
to the data would yield inappropriate results, since the last step was developed under the
assumption that there is only one break-point in the series. If there is a second shift which
reverses the first, the algorithm described in step four might very well miss both of them.
To address the problem, we look for mean shifts in short "windows" only. Hence we need
steps one to three to determine which periods we should look at.
Second, since the bond price series contain a unit root, test statistics based on regression
residuals will have a nonstandard distribution. For step four, we therefore generate Monte
Carlo critical values for the Wald test under the null hypothesis of no structural breaks.
Critical values for the F-tests of no breaks are approximated with 5,000 Monte Carlo
simulations of the equation
lnpt = c+ lnpt-I
+
t
(A3)
with c = 0.1 and se(e,) = 0.1. The resulting 90-, 95-, and 99-percent critical values are 3.14,
4.32, and 8.00, respectively.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
494
Frey and Kucher
35-
301
25 -
20
;
15
10
5
0
1/43
7/43
1/44
7/44
1/45
7/45
1/46
7/46
APPENDiX FIGuRE 2
F-TEST FOR GERMANBONDS, TIMEWINDOWJANUARY1943 TO DECEMBER1946
Source:See the text.
Finally,we alsotestforvariations
of thebondindexof a specificcountiyrelativeto the
indexof all government
bondstradedin Zilrich.Thatis, we rewriteequationAl as
lnpt -lnp-, =Pio+l lnpti +A 1nPt,l+-ct
(A4)
Sucha specification
wouldseemto be morein linewiththeexcess-return
literature
frequentlyusedin financestudies.3"
However,we findthesamebreakpoints
as withthefirst
proceduresuggested,andthe size of the effectdoesnot changedramatically
(noneare
reversed).Sincewe believethatthecoefficientsof thespecification
in equation
presented
Al aremoreeasilyaccessible,onlythefirstspecification
is used.
Thecapitalmarketis simultaneously
influenced
byagreatnumber
offactors.Theeconometricmethodsuggestedhereallowsus onlyto controlsomeof them.Nevertheless,
the
resultsof ouranalysisareencouraging.
31 See for exampleCampbell,Lo, andMacKinlay,Econometrics.
REFERENCES
Banerjee, Anindya, Robin L. Lumsdaine, and James H. Stock. "Recursive and Sequential
Tests of the Unit Root and Trend Break Hypotheses: Theory and International Evidence." Journal of Business and Economic Statistics 10, no. 3 (1992): 271-87.
Boelcke, Will A. Die Kosten von Hitlers Krieg. Kriegsfinanzierung undfinanzielles
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
WorldWarII
495
Kriegserbein Deutschland1933-1948. Paderborn:Schoningh,1985.
Campbell,JohnY, AndrewW Lo, andCraigA. MacKinlay.TheEconometricsof Financial Markets.Princeton,NJ: PrincetonUniversityPress, 1997.
Crettol, Vincent, and Patrick Halbeisen. Die wahrungspolitischenHintergrinde der
Goldtransaktionender SchweizerischenNationalbankim ZweitenWeltkrieg.Zurich:
SchweizerischeNationalbank,1999.
De Bondt, Werner,and RichardThaler."Does the Stock MarketOverreact?"Journal of
Finance 40, no. 3 (1985): 793-805.
Duffie, Darrell.DynamicAssetPricing Theory.Princeton,NJ:PrincetonUniversityPress,
1996.
Eichengreen, Barry,and RichardPortes. "Debt and Default in the 1930s: Causes and
Consequences."EuropeanEconomicReview30, no. 3 (1986): 599-640.
Erbe,Rene.Die nationalsozialistischeWirtschaftspolitik
1933-1939 imLichteder modernen Theorie.Zurich:PolygraphischerVerlag,1958.
Fama, Eugene F. "EfficientCapitalMarkets:II."Journal of Finance 46, no. 5 (1991):
1575-617.
Federau,Fritz. Der Zweite Weltkrieg.Seine Finanzierung in Deutschland Tubingen:
Wunderlich,1962.
Fischer,Wolfram.Die Wirtschaftspolitik
des Nationalsozialismus.Luneburg:LandeszentralefiirPolitischeBildung, 1961.
Forsythe,Robert,ForrestNelson, GeorgeR. Neumann,andJackWright."Anatomyof an
ExperimentalPoliticalStockMarket."AmericanEconomicReview82, no. 5 (1992):
1142-61.
Hardach,Gerd.Der Marshall-Plan:Auslandshilfeund Wiederauifbau
in Westdeutschland
1948-1952. Mulnchen:dtv wissenschaft,1994.
Henschy,Reg. Freedomat Midnight:Austria1938-55: A Storyof the TraumaticYearsof
Occupation.Worcester:Billings and Son, 1989.
Jost, Hans-Ulrich.Politik und Wirtschaftim Krieg: die Schweiz, 1938-1948. Zurich:
Chronos,1998.
K6llner, Lutz. Militar und Finanzen. Zur Finanzgeschichteund Finanzsoziologie der
Militarauisgabenin Deutschland.Munchen:Bernardund Graefe,1982.
Kozicki, Henry, ed. Developmentsin ModernHistoriography.New York: St. Martin's
Press, 1993.
Milward,Alan S. War,Economyand Society.London:Penguin, 1977.
Moos, Herbertvon. Das grosse Weltgeschehen.Bern:Hallwag, 1940-1945.
Parrish,Scott D., andMikhailM. Narinsky.New Evidenceon the SovietRejectionof the
MarshallPlan. Washington,DC:WoodrowWilsonInternationalCenterfor Scholars,
1994.
Perron,Pierre."The GreatCrash,the Oil Price Shock and the Unit Root Hypothesis."
Econometrica57, no. 6 (1989): 1361-401.
Salis, JeanR. de. Eine Chronikdes zweiten Weltkriegs.Zurich:OrellFuessli, 1981.
Schwab,Hubert."DerSchweizerischeEffektenmarkt
1936-1946."Dissertation,University
of Zurich,1948.
Smith,VernonL. "Marketsas Economizersof Information:ExperimentalExaminationof
the 'HayekHypothesis'."EconomicInquiry20, no. 2 (1982): 165-79.
Sobel,Russel S. "ExchangeRateEvidenceon the Effectivenessof UnitedNationsPolicy."
Public Choice 95 (1998): 1-25.
Switzerland.National Bank. Monatsberichteder SchweizerischenNationalbank.Bern:
SchweizerischenNationalbank,January1929-January1949.
Urner,Klaus. Die Schweiz muss noch geschlucktwerden.Zurich:VerlagNeue Zurcher
Zeitung, 1990.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions
496
Frey and Kucher
Weinberg,GerhardL. A Worldat Arms:A Global History of WorldWarII. Cambridge,
CambridgeUniversityPress, 1994.
Willard,KristenL., TimothyW. Guinnane,and HarveyS. Rosen. "TurningPoints in the
Civil War:Views fromthe GreenbackMarket."AmericanEconomic Review 86, no.
4 (1996): 1001-18.
Wirtschaftslage,Die. Beilage zum Monatsberichtder SchweizerischenNationalbank.
Zurich:SchweizerischeNationalbank,1953, 1954.
This content downloaded from 193.54.110.35 on Thu, 18 Apr 2013 16:42:52 PM
All use subject to JSTOR Terms and Conditions