Did capital go away?
The contribution of capital flight to the decline in β during and after
World War I
Mehrene Larudee, Visiting Scholar
Economics Department
University of Massachusetts
Amherst, MA 01003
March 2015
JEL codes: D31, F39, N24, N34
Keywords: Piketty, inequality, capital flight, wealth, Swiss banks, Keynes
Abstract
Piketty’s β, the wealth/income ratio of a country, dropped sharply during and/or
after World War I for France, Germany, and the UK. This paper explores what part of this
drop is likely due to capital flight rather than physical destruction or loss in asset values.
Piketty and Zucman (2014, Data Appendix) acknowledge that capital flight is one causal
factor, at least flight of foreign assets from Germany in 1918-19 or so. For capital flight to
Switzerland from France from 1913 to 1929, this paper roughly estimates, based on plausible
assumptions, that the top 1 percent of wealth-holders transferred over 8% of their wealth to
Switzerland, about three-fourth as securities and one-fourth as financial deposits. This was
over 22 percent of their financial deposits and an unknown percent of their securities. This
alone, however, accounts for less than one-fifth of the total 13-plus percentage point decline
in the top 1%’s share of wealth in France from 1913 to 1929. But if the wealthy borrowed
back their own flight capital in the guise of arm’s-length loans, it could account for up to
two-fifths of the decline in wealth by both removing the assets from reported wealth and
transforming them into liabilities. Loss in value of assets evidently accounts for much of the
rest. Factors that affected capital flight from Germany are similar, but not identical.
Acknowledgments
This paper was prepared for the History and Development Workshop, Economics
Department, University of Massachusetts, Amherst, April 1, 2015. A special thanks for the
supportive intellectual environment of the Economics Department, which has welcomed me
as a Visiting Scholar, and contributed to development of these ideas. For useful
conversations, thanks especially to Jim Crotty, Gerald Epstein, Carol Heim, Bob Pollin, and
members of the Piketty study group, notably Anders Fremstad and David Rosnick, whose
timely and trenchant critiques I have particularly appreciated. I am fully responsible for any
remaining errors.
I. Introduction
A central claim in Piketty’s work is that increasing inequality over more than a
century was powerfully, but temporarily, interrupted by the events of 1910-1950, and that
now “capital is back” (Piketty and Zucman 2014). The evidence is a roughly U-shaped dip
over from World War I through World War II in the generally upward-sloping plot of β, the
ratio of a country’s wealth to its income, in large European countries such as France and
Germany. A similar dip appears in the share of the top 1 percent of wealth-holders in total
wealth in these and other countries.
This paper asks how much of the U-shaped dip in β is due not to a decline in the
value of capital or to destruction of physical capital, but rather to capital flight, reducing
reported capital assets and reported income from those assets during and after World War I.
In numerous brief passages, Piketty’s Capital (2014) attributes the dip in β almost entirely to
chaos and destruction, listing causes that include inflation, taxation, currency depreciation,
wartime physical damage, rent control, capital controls, and more. But despite no mention of
capital flight in the World War I era in published work by Piketty and co-authors, Piketty
and Zucman (2014) acknowledge in their online data appendix that during and after World
War I capital flight affected the value of reported wealth. In discussing the β series they
constructed for Germany, they write:
One caveat is that estimates of foreign assets [a component of total wealth - ML] for
the inter-war are probably on the low-end for the same reasons as they are today:
they miss the foreign securities held offshore by individuals (Zucman, 2013). It was
already well acknowledged by contemporaries that a sizable amount of foreign
securities in private hands had left Germany since the end of World War I (see, e.g.,
Keynes, 1920, chapter 5, III:1). Available Swiss data show a large increase in foreign
fortunes managed by Swiss banks in the 1920s, and in all likelihood a sizable fraction
of those belonged to German households. (Piketty and Zucman 2014, Data
Appendix, 92)
Piketty (2014, 467) further acknowledges that capital hidden in tax havens is substantial
today, and should be added to total reported wealth for each country; however, his graphs of
β do not include it for any period, possibly because allocating it to specific countries is
difficult, and a full time series for it is unavailable. Evidently Piketty and Zucman’s research
on these questions is continuing.
In fact, the bulk of the 1910-1950 decline in β occurred during the 1910-1929 period,
according to Piketty’s graphs. Any needed upward adjustment would at least partly “fill in”
the U shape, making it shallower, and reducing the size of the apparent reversal of the long
run inequalizing trend. However, this paper estimates that it is likely too small to erase the
U-shaped period of reversal entirely.
The most dramatic graphs in Piketty (2014) do show a sharp decline, but give a
misleading impression of when it occurred. Piketty (2014, Figure 4.5) shows “decennial
averages” for β falling by more than half from 1910 to 1920, giving the appearance that this
decline occurred during or very shortly after World War I. However, Piketty (2010, Data
Appendix, Part 1, p. 2 and Table 2A) explains that the “decennial average” shown for France
for 1910 was actually an average of the 1910-1913 prewar β estimates, while the decennial
2 average for France for 1920 was computed, like most of the other “decennial averages”, as
the average over 1920-1929.
Moreover, for France before 1970, relatively good data on the wealth/income ratio
were only available for isolated years. The years for which wealth/income data actually had
been calculated by other researchers, and were reviewed and in some cases adjusted by
Piketty and his team, were only 1913, 1925, and 1954 (Piketty 2011, Data Appendix).
Similarly, for Germany the actual data on β were for 1910-1913 and 1927. For all other
years, the data for β were imputed by a process to be described shortly. (The UK is not
discussed in this paper, as it focuses on countries neighboring Switzerland; and data for Italy
were not available.) Figure 1 shows the years of actual data for France and Germany with
white markers, and the years of imputed data for both countries with black markers. What
we really know, then, is that between about 1913 and some time in the 1920s, which varies
by country, there was a sharp decline in the ratio of reported private wealth to reported
income.
[Figure 1 about here]
The question is: How much of the decline can be attributed to destruction of capital,
either physically or in value, and how much to capital flight to tax havens like Switzerland?
In other words, in the title of this paper, the phrase “Did capital go away?” is a double entendre.
It asks: On the one hand, did capital “go away” in the sense that its physical existence was
destroyed by war, or its value was destroyed by inflation, rent control, repudiation of debt, or
other causes? Or, on the other hand, did capital “go away” in the sense that it fled to tax
havens, mainly Switzerland, and (along with its income, presumably) ceased to be reported in
the data – so true wealth declined less than was reported?
But first: would removing the value of an asset and of its associated income from the
data – either because it was destroyed or because it ceased to be reported – always reduce β?
The effect of destruction should be no different from the effect of removal from reporting
(such as with capital flight). The effect would indeed be to reduce β, unless the lost capital
generated an extremely high percent return exceeding the inverse of current β. Labor income
is a large share of total income, and therefore the overall ratio of wealth to income from
both wealth and labor – usually no higher than about 7:1 – is lower than the ratio of the
capital alone to the income from that capital, which ranges from 10:1 to 20:1 for returns on
invested capital of between 10 and 5 percent (see Piketty 2014, 448, Table 12.2). Given a β
of 7:1, only the loss of a type of capital that consistently generated a return of 1/7 = 14.2%
or more (and its associated income) would raise β above 7:1. Hence removing from the
reported data both capital assets and their income will normally reduce reported β.
Section II reviews the data on trends in β and in wealth inequality that need to be
explained. Section III reviews reasons to believe that capital flight occurred, including
evidence of motives, means, and opportunity, as well as contemporary and retrospective
accounts. Section IV constructs a plausible estimate for capital flight to Switzerland during
1910-1929 based on the limited data available, and shows that it is consistent with changes in
the distribution of wealth and income. Section V critically discusses Piketty’s analysis of the
contribution of other causes to the sharp decline in β during and after World War I. Section
VI concludes.
3 II. The data to be explained
The decline in wealth/income (β) during and after World War I
Piketty’s β fell by about two-thirds during 1910-1950 in France and Germany. In
fact, as shown in Figure 1, in both countries β fell by more than half during and after World
War I. The trough in β for France and Germany was about 1925. To be precise, by 1925 β
was 63% lower in France than it had been in 1913, by 1927 it was 48% lower in Germany
than it had been in 1910-1913 (Piketty and Zucman 2014, France.xls, Table FR.6f;
Germany.xls, Table DE.6f). In contrast, despite a small rise in β during the late 1920s and
1930s, the percent decline recorded for World War II was smaller than during and after
World War I. Yet by all accounts, as Piketty notes, the physical destruction in World War II
was far greater.
This raises the puzzling question: Why did β fall so far, so fast during and after
World War I? This paper explores the hypothesis that capital flight accounts for a substantial
share of this decline in the World War I pre-depression era, 1910-1929, Using financial data
for various countries for 1929 and earlier in League of Nations (1931). If the amount of
capital that went into hiding was substantial, quite possibly much of it remained in hiding
through 1950, helping to account for the smaller decline during the World War II era.
France and Germany
Figure 1 plots annual data on β for France and Germany from the online data tables
from Piketty and Zucman (2014). However, as their data appendix explains in detail, before
1970 it was only in a very few isolated years that the wealth/income ratio was directly
estimated – by other researchers, with some later adjustments by Piketty’s group. For France,
these years were 1896, 1913, 1925, and 1954. Piketty’s group then imputed annual data for
the intervening years, as explained below. For Germany, Piketty and Zucman have fairly
good estimates for β for around 1913 and 1927, as well as for 1950, and similarly impute β
for intervening years. Figure 1 plots the combined actual and imputed annual data for β for
both countries, with white markers for years when β was actually measured, and black
markers for years of imputed data. Looking at the white markers in Figure 1, the message is
that for France, β was high in 1913, and quite low in 1925, fluctuating after that. Better
understanding the imputation process will help clarify what we do and do not know about
exactly when the fall occurred; to this we now turn.
[Figure 1 about here]
For France, Piketty reconstructed the β series between 1913 and 1925 (as between
other years of known β such as 1925 and 1954) by linking the known β1913 with the known
β1925 – the results of which are displayed in online Table A17 for Piketty (2011).1 Known
annual data series on growth of national income, saving rates, and wartime destruction were
entered into their respective columns.2 Into two columns of unknown, or mostly unknown,
1
See the online Data Appendix for Piketty (2011) at piketty.pse.ens.fr, especially section A.5
(pages 45-59), and Table A17.
2
Wartime destruction is “known” in the sense that Piketty obtains the series for it by using
the estimate by Divisia, Dupin, and Roy (1956) of total physical destruction to France in
4 quantities (the β column, years 1914-1924 for which the cells were empty, and the column
for rate of increase of total wealth), formulas were entered linking these unknown to other
known and unknown variables. The last stage was to enter into the “rate of capital gains”
column, in all the cells from 1914 to 1925, a uniform guess (same value in each cell in this
range) for this variable, defined as the excess of asset price inflation over CPI inflation – and
then to observe whether it delivered the known value of β1925. If the β1925 so generated was
too low, a new, higher uniform rate of capital gains was entered for 1914-1924; if too high,
the guess was adjusted downward. By trial and error, the average rate of capital gains over
the period was found that reproduced the known β1925.
This produced the imputed annual data series shown in Figure 1. However, it is not
the true data series, precisely because, as Piketty states in the data appendix, asset price
inflation is quite volatile. Hence a uniform value of qt, the “rate of capital gains”, is likely to
be wrong, possibly far wrong, in any given year, and the resulting imputed annual value of βt
is only a rough approximation. In the end, we mainly know what we already knew: that for
France, β fell sharply from 1913 to 1925 and, all else equal, it would have fallen less in
periods when saving rates were higher and income growth lower. The results are more
informative than a simple linear interpolation; but we end up not quite knowing when during
this period the decline mainly took place; and the imputed data do not really support
regressing β on lagged variables to examine causation and deliver valid results.
Decline in the top 1%’s wealth, and decline in flow of inheritances
Ultimately we are interested not just in β, but in the relationship between trends in
inequality of wealth and income inequality and the trend in β. Any explanation for the
decline in β must also be consistent with trends in inequality. For example, the share of the
top 1 percent in reported total wealth and income fell during and after World War I in
France and other countries, as shown in Figures 2 and 3. (Other data on inequality show
similar trends: a sharp decline in the reported flow of inheritances, and in the share of
inherited wealth in total wealth (Piketty 2014, Figure 11.1). An important feature of these
changes is that, as shown in Figure 2, the share of the top 10% in wealth (the curves above
the 80% level) fell much less than the share of the top 1% (the curves in the 45%-70%
range). Figure 3 shows a similar pattern in the evolution of income shares in France: in this
period, only the income share of the top 0.1% of income recipients fell appreciably from
1910 to 1930; the next 0.9% lost less, and at the end of the period the next 9% had the same
income share as in 1910. A plausible hypothesis for the 1910-1929 period is that only the top
1 percent were able to engage in capital flight, because the large but fairly fixed transaction
costs involved created significant scale economies, so that only the wealthiest would reap a
positive net benefit from paying the costs of capital flight. This motivates the assumption in
section IV that it was the top 1 percent who engaged in capital flight.
[Figure 2 about here]
[Figure 3 about here]
World War I, with some adjustments; assigning one-fourth of it to each year from 1915
through 1918; and then separating out the loss in foreign assets from wartime physical
destruction.
5 Other hypotheses are possible, however. It might be thought, for example, that the
top 1% suffered larger losses in wealth than the next 9% because the top 1% held more
government bonds, which lost value because of inflation. This is a plausible hypothesis, but
not one that is supported by Piketty’s data on wealth composition for France for 1912: he
wealth composition data show that the share of wealth held in the form of public bonds was
almost identical for the top 1 percent (19 percent of whose wealth was public bonds) and the
next 9 percent (18 percent of whose wealth was public bonds), suggesting that inflation that
eroded the value of government bonds affected the two groups equally (Piketty 2014, Table
10.1, 371). Of course it is possible that during the war the top 1% were the main buyers of
bonds, and so suffered losses much greater than those of the next 9%; however, to date I
have been unable to find evidence of this.
III. Reasons to think that capital flight occurred
There are two kinds of reasons to think capital flight took place. One is direct
comments by observers at the time or later. As Piketty notes, in The Economic Consequences of
the Peace (1920, Chapter V, section III.1.iv), Keynes explicitly describes capital flight from
Germany, explaining that wealthy Germans urgently sought to place their assets out of reach,
because the Allies intended to confiscate their wealth for war reparations:
It is certain that since the Armistice there has been a great flight abroad of the
foreign securities still remaining in private hands. This is exceedingly difficult to
prevent. German foreign investments are as a rule in the form of bearer securities
and are not registered. They are easily smuggled abroad across Germany's extensive
land frontiers, and for some months before the conclusion of peace it was certain
that their owners would not be allowed to retain them if the Allied Governments
could discover any method of getting hold of them. These factors combined to
stimulate human ingenuity, and the efforts both of the Allied and of the German
Governments to interfere effectively with the outflow are believed to have been
largely futile.
In fact, since the various German states such as Prussia had already had wealth taxes since
the 1890s, there is reason to think that German wealth-holders were already somewhat more
experienced than those of other countries in concealing or transferring their wealth to reduce
taxation. Piketty and Zucman (2014, Data Appendix) remark on the apparent degree of tax
evasion in the wealth data for Germany in 1913, when a new “defense tax” (Wehrbeitrag) was
imposed; they note, too, that German wealth-holders reported a surprisingly high level of
liabilities compared to wealth-holders in other countries with no wealth tax, and seem to
imply that the liabilities were likely associated with tax evasion.
The second kind of reason is indirect, based on evidence of motive, means, and
opportunity. Motives for the rich to move their funds to safe havens included changes such
as rising inflation, currency depreciation, and new and increased taxation that fell heavily on
the rich. For countries sharing a border with Switzerland, including Germany, France, and
Italy, the means for the wealthy were simply the chance to travel across the border, perhaps
paying for insurance or physical security to protect funds and financial securities in transit.
We reasonably assume that the top 1 percent had the means to afford safe transport, since
the return from doing so was presumably sufficient to cover the cost. Finally, the
opportunity was available in Switzerland as in the 1920s, and to some extent even earlier, it
6 rapidly became a tax haven: a country with a stable currency, stable prices, and lower
effective taxation than elsewhere.
Motives for capital flight
As noted above, for Germans a prime motive for capital flight at war’s end was the
near-certainty that much wealth in Germany would be taken to pay war reparations. Piketty
and Zucman (2014, data appendix) note that considerable wealth “left Germany”; Keynes
suggests something slightly different: that in Germany foreign securities, which were largely
bearer securities whose ownership was unregistered, were relatively easy to smuggle out of
Germany – but an unknown amount of wealth was also effectively hidden inside Germany.
Germany also experienced a rapid and growing rate of inflation during and after
World War I, reaching hyperinflation, rapid currency depreciation, and then complete
currency collapse by 1923, after which the mark was replaced by a new currency
(Eichengreen 1992; Farquet 2012). Other European countries also faced considerable
depreciation. Figure 4 shows the fluctuation of several currencies against the Swiss franc,
including the dramatic depreciation of the German mark up through 1923. A similar plot of
depreciation of the German mark against the dollar can be generated from League of
Nations (1931) data.
[Figure 4 about here]
Wealthy households in Germany and France also had other reasons to engage in
capital flight. These included inflation, and especially in France, new income and inheritance
taxes which fell especially heavily on high incomes. Since the large majority were in danger of
sacrificing their lives or at least their health in the war, measures were implemented to relieve
them of financial burdens which in any case they were likely in most cases unable to pay.
Farmers and urban tenants received exemptions from rent payments if they had been
mobilized to fight in the war. In part because of exemptions, but perhaps also due to tax
avoidance or deficiencies in tax collection, tax payments in France fell to three-fifths of their
previous normal level by the end of 1914 (Eichengreen 1992). Though statutory taxes were
increased in France during the war, collection was spotty: tax collectors in France were told
that households in which at least one member was in the military should not be prosecuted
(Eichengreen 1992, 75-76). Figure 5 shows the rise in statutory tax rates on top incomes in
France, Germany, and the UK, and Figure 6 shows the rise in statutory tax rates on top
inheritances in France and Germany, with data from Piketty (2014).
[Figure 5 about here]
[Figure 6 about here]
Means for capital flight
Not all wealth-holders had the means to move their capital abroad or effectively hide
it, but we presume the top 1 percent of wealth-holders did. This assumption is consistent
with Figure 3, and based on transaction costs involved – presumably payment for financial
advice and services, the time and effort required to learn about the costs and comparative
risks involved, and the cost of traveling to Switzerland to carry out the transactions, as well
as the cost, in security and/or perhaps insurance, for moving financial assets to Switzerland.
Another factor that made the 1 percent more likely to engage in capital flight was that a
7 somewhat larger share of their fortunes in France were moveable assets, although in 1912
the difference was not large: 65 percent compared to 58 percent or less for those with less
wealth (Piketty 2014, Table 10.1).
Opportunity for capital flight
Switzerland was viewed as an excellent safe haven for capital by the 1920s, and even
to some extent before. An ideal tax haven needs a stable currency, bank secrecy, low taxes,
and political institutions and governance that create confidence that these will remain in
place. Switzerland had all of these by the early 1920s, after it brought moderate wartime
inflation under control and stabilized the Swiss franc, backed by gold. Strongly confirming
the role of Switzerland as a tax haven, a detailed 2007 history of the Swiss National Bank
says,
During the First World War, Switzerland had developed rapidly as a neutral
intermediary, and then in the 1920 [sic] appeared as an island of stability in a wild sea
of currency inflation and depreciation. In addition, a lack of transparency in
corporate ownership and control meant that Swiss financial institutions offered
attractions as part of a chain of holdings. In consequence, German and other central
European corporations and individuals saw Swiss banks and their affiliates as ways
of lending money back to themselves, with some tax advantages. Switzerland thus
had a role as a secure tax haven. (Abegg, Baltensperger, et al. 2007, 71)
The same source discusses capital movements in and out of Switzerland over the 1920s, and
then notes:
In 1930, a private Swiss banker told a representative of the Bank of England that he
would be afraid if he knew how much foreign money (presumably French and
German) was flowing into Switzerland. (Ibid., 72, citing an archived Bank of England
memorandum)
With regard to bank secrecy, Palan, Murphy, and Chavagneux (2010, 108) cite
Fehrenbach (1966) as asserting that bank secrecy was standard in Switzerland by 1912, even
though violation of bank secrecy was not criminalized by legislation there until 1934. Palan
et al. also note, without citation, that there is “talk of the emergence of offshore trusts in
Switzerland in the 1920s, used primarily by wealthy Italians to protect their assets” (115). In
the present paper Italy is little discussed because of lack of data for this period. However, in
section IV we will assume Italian residents engaged in capital flight to the same degree as did
French and Germans.
Naylor (1987, 33) cites evidence from an internal Chase Manhattan Bank memo that
by the mid-1960s Switzerland was already by far the leading tax haven in the world. He
notes, too, that at least as early as 1932, American criminals were depositing funds in banks
in Switzerland and using loan-back schemes to reduce tax liability (21-22).
Another of Switzerland’s advantages as a tax haven was lower tax collection in
practice than other countries. Figure 7 shows that tax revenues collected as a share of GDP
in Switzerland were lower than those in France, Germany, and the UK in this period. As
Farquet (2012) explains in detail, Switzerland’s fairly progressive official tax structure was
not a good guide to the actual tax burden on foreign assets held in Switzerland. This was in
part because taxes were collected by cantons, which not only were in tax competition with
8 one another, but typically relied on self-reporting of income, wealth, and inheritance, and
often exempted or partially exempted from taxation those who proved that they were
domiciled outside Switzerland. The increase from 1920 to 1922 shown in the graph probably
represents an increase in Swiss holdings of flight capital rather than a change in the tax rate.
With respect to Germany, for example, hyperinflation probably explains the sharp decline of
deposits in German commercial banks, reported in League of Nations (1931, 138-139), and
it is likely that much of these deposits fled to safety in Switzerland.
[Figure 7 about here]
Another Swiss advantage was the stability of its currency relative to other currencies.,
already illustrated in Figure 4. That is, in the diagram the Swiss franc is the reference
horizontal line, and fluctuations in the value of other currencies measured against the Swiss
franc are shown. In fact, the value of the Swiss franc did vary a bit during World War I,
appreciating against the US dollar about 6 percent from 1913 to 1918; yet variation of other
currencies against the dollar was typically greater. After the war, in 1919 and 1920, the Swiss
franc depreciated from $0.2043 in 1918 to $0.1540 in 1920, but by 1921 it had returned to
$0.1953, close to its 1913 level of $0.1930. After a brief 5% depreciation over the next two
years, it returned to its 1921 (and 1913) level and remained there for the rest of the 1920s,
backed by an ample supply of gold (League of Nations 1931).
In addition, a conservative government and weak federalism helped preserve the
status quo. In terms of secrecy, low taxation, price stability, and a stable currency, then, by
the early 1920s Switzerland offered a golden opportunity as a safe haven for capital.
IV. Estimating capital flight
We begin by estimating what share of on-balance-sheet financial deposits held in
Swiss banks, credit unions, and other financial institutions in 1929 was flight capital, using
League of Nations (1931) data on bank balance sheets for that year. Combining this with
Piketty’s wealth distribution and wealth composition data for France, and a few plausible
assumptions, allows us to check the plausibility of estimates by calculating the approximate
implied share of their wealth that the top 1 percent moved to Switzerland. Using other
information, an estimate can also be made for flight capital in the form of securities, which
evidently were typically held off-balance-sheet in Swiss banks, usually as the holdings of trust
companies controlled by the banks.
Estimating flight capital to Switzerland in the form of on-balance-sheet financial deposits
How much flight capital was there in Switzerland in the form of on-balance-sheet
financial deposits in 1929? Table 1 shows, by country, aggregate financial deposits in
commercial banks, credit unions, and other financial institutions divided by population. The
13 European and Scandinavian countries with the highest per capita aggregate deposits,
including the UK and Ireland, are shown; the US is also shown for comparison. Figure 8
shows just the per capita aggregate deposits for these countries, excluding the US. The per
capita aggregate deposit amount of $178 for all 13 countries together is also shown, obtained
by multiplying each country’s per capita figure by its population, summing them, and
dividing by the total population of all 13 countries.
9 [Table 1 about here]
[Figure 8 about here]
We assume that the excess of Switzerland’s per capita financial deposits over the $178
average per capita deposit ($714-$178 = $536) represents flight capital; this is about threefourths of the total. Multiplying by the Swiss population of 4.022 million (Maddison 2010),
total flight capital in deposits in Swiss banks in 1929 is estimated at $2.26 billion in current
dollars ($31 billion in 2014 dollars).
France, Germany, and Italy are three major countries that border Switzerland (along
with Liechtenstein and Austria); and Switzerland has French-speaking, German-speaking,
and Italian-speaking cantons. For these reasons, but also because we have already noted
specific accounts of capital flight from each of these three countries to Switzerland, we
assume that all the flight capital estimated in the first step came from these three countries
alone, and that flight capital constituted the same fraction of each country’s domestic
financial deposits in 1929.3 Using 1929 population data (Maddison 2010) and 1929 per capita
domestic deposits in France ($90), Germany ($122), and Italy ($76) we find that an amount
equal to 14.7% of the aggregate domestic deposits of these three countries was flight capital
in the form of deposits in Swiss financial institutions.4 Details are in Table 2.
[Table 2 about here]
Roughly consistent with this approximation, Farquet (2012) cites the finding of Guex
(1993, 150) that holdings of Swiss banks were 2 billion Swiss francs (CHF) before World
War I, according to contemporary accounts, but by 1920 had grown to 10-20 billion CHF.5
Farquet (2012, 4) further notes that “Although…the cumulative balance sheets of the major
Swiss banks represented only 26% of those of their French counterparts in 1913, they
subsequently amounted to 73% in 1929.” Figure 9, based on League of Nations (1931) data,
shows this increase. Such a change could result both from an increase in foreign-owned
deposits in Swiss banks, and movement of French deposits out of French banks.6 Simple
algebra shows that a transfer of 10% of domestic French, German, and Italian deposits to
Swiss banks, with no other change or growth (and if German and Italian domestic deposits
totaled three times French deposits in 1913 as they did in 1929), would bring about the result
3
There is actually no implication in this assumption based on 1929 data that the deposits
arrived in Swiss financial institutions after the beginning of World War I; possibly some
arrived before, although the bulk of it probably arrived afterward.
4
League of Nations (1931, 9) explains some of the difference in per capita aggregate
deposits among countries by the different degree to which financial intermediation was used
to mobilize savings for investment; in some countries households were more likely to invest
directly into firms, while in others they were more likely to deposit funds in banks, which
lent funds to firms for investment.
5
Indeed, judging by the fact that big banks did almost no mortgage lending (League of
Nations 1931), it may be that the big banks’ customers were almost exclusively foreign
nationals, and that Swiss nationals typically used cantonal banks, local banks, credit unions,
or the post office for banking services; this is, however, only a guess.
6
However, it is only fair to acknowledge that they could also result from other behavioral
changes, such as shifts within Switzerland in use of major banks vs. use of local or cantonal
banks.
10 described by Farquet.7 This differs from the 14.7% estimate, but is in the same plausible
range; and the categories do not match up well enough to expect them to coincide.
[Figure 9 about here]
Now consider France and go back to the 14.7% estimate. If total deposits that
French residents owned in both France and Switzerland combined were 114.7% of the
deposits in France alone, this would imply that 14.7%/114.7% = 12.8% of the total deposits
held by French residents were in on-balance-sheet financial deposits in Switzerland, that is,
flight capital. However, deposits were only a small share of total wealth (Piketty 2014, Table
10.1); we need to estimate flight of securities to Switzerland as well.
We assume that all the capital flight was carried out by the top 1% of wealth-holders,
due on the one hand to economies of scale in transactions costs such as hiring financial
advisors, traveling to Switzerland, and paying any fees required, and on the other hand to the
fact that the top 1% typically hold a somewhat larger share of their wealth in moveable
forms such as financial assets, and did so in 1912, according to Piketty (2014, Table 10.1, p.
371). For France, we can use this assumption to derive the share of all domestic deposits
that were held by the top 1%, and then to calculate the share of those deposits in the total
wealth held in France.
Table 3 shows data that allow us to calculate what share of the deposits held by the
top 1% in France they had moved to Switzerland by 1929. For the top 1%, the next 9%, and
the next 40% (assuming the bottom 50% had no net wealth), we have data on the share of
each in total private wealth in 1910. In addition, we have data on the composition of each
quantile’s wealth: Table 3, column 2 shows the share of “other financial assets” (“cash,
deposits, etc.”) in their wealth holdings (Piketty 2014, Table 10.1). We assume are the same
fraction f of “other financial assets” is deposits for each wealth group, and f drops out in the
following calculation, so that we equivalently assume that all of “other financial assets” are
deposits. For each group we multiply these two quantities together to each group’s deposits
in France as a share of total wealth in France (column 3). Finally, summing these three
categories to find that deposits were 9.6% of total wealth, we divide each wealth group’s
deposits as a share of total wealth by total deposits as a share of total wealth (4.84%/9.61%,
and so forth) to get each wealth group’s share of total deposits held in France. For example,
the top 1% owned 4.84%/9.61% = 50.4% of all domestic financial deposits. Since this
group owned 60.5% of all wealth, and its financial deposits were a smaller share of its wealth
than for the less wealthy, this seems plausible.
[Table 3 about here]
If the wealth composition shares for 1912 were still valid in 1929, then the amount
held in Swiss financial deposits, which was 14.7% of all French domestic deposits, was
14.7%/51.0% = 29% of the top 1%’s holdings of French domestic deposits, or 29%/129%
= 22.5% of all the financial deposits held by the top 1% of French resident wealth-holders
(deposits in France, plus those in Switzerland); see Table 4. Again, this is well within a
plausible range. In fact, the top 1 percent’s holdings had declined relative to total wealth in
France, so quite likely somewhat more than 22.5% of their financial deposits were held in
Switzerland in 1929.
7
If F is initial French deposits, and K denotes French capital flight in deposits to Swiss
banks, then .26F + 4K = .73(F – K) or K/F = .47/4.73 ~ 10 percent.
11 [Table 4 about here]
By how much would such capital flight in deposits alone have reduced total reported
wealth in France? French deposits were only a 9.61% share of total French wealth, so the
claim here is that the amount of wealth that the top 1% held in Switzerland in 1929 was
equal to 14.7% of the wealth held in France in 1929, or (14.7%)(9.5%) = 1.4% of total
French wealth held in France. There is reason to think, however, that a much larger value of
securities was also moved to Switzerland, so we also estimate the size of that form of flight
capital.
Estimating flight capital flight capital to Switzerland in the form of securities
Keynes (1920) strongly asserts in the passage quoted above that there was likely
considerable flight of securities from Germany, at least, to Switzerland just after World War
I. Securities in Germany were bearer bonds, and therefore easier to transfer to Swiss banks
than other securities. They apparently were typically held off-balance-sheet in trusts
controlled primarily by large banks. Fortunately, we have some fragmentary information
about the size of such off-balance-sheet holdings. Farquet (2012, 4), relying on research by
others, writes:
A committee of experts working on Swiss history during World War II who,
exceptionally, had access to the in-house archives of the banks, showed that holdings
of securities, generally in off-balance-sheet bank custody accounts, represented, in
the two biggest Swiss banks in 1931, more than three times their balance-sheet.
Hence, the figure of 1,359.6M Swiss francs (CHF) for 1937 that was given during the
first survey undertaken by the Swiss National Bank (SNB) on the assets of foreign
clients was amply underestimated.
We estimated that about three-quarters of all 1929 on-balance-sheet Swiss financial
deposits were flight capital, so if off-balance-sheet securities were worth more than three
times the balance sheet, the amount of flight capital off-balance-sheet as securities could
have been more than 3/(3/4) = 4 times the value on balance sheet, at least for the two
largest Swiss banks to which Farquet refers, presumably Credit Suisse and Union Bank of
Switzerland (UBS). However, this refers to 1931, a year when there was a banking crisis in
Germany (Kenwood and Lougheed 1999, Chs. 12-13), when one would expect an upsurge in
flight capital to Switzerland, at least from Germany. In fact, an increase, both in deposits and
in the holdings by trust companies controlled by the big banks, began around 1926 and was
in full swing by 1929 (League of Nations 1931, 258-278, esp. 266). If the banking crisis
increased flight of securities more rapidly than flight of deposits (and we do not know this
was the case), perhaps a conservative figure of 3 times the value of deposits is more realistic
for 1929.
Let us further assume that the amount of securities in Switzerland from each of
Germany, France, and Italy was proportionate to these countries’ domestic deposits in
Switzerland, and that all the securities came from the top 1% of wealth-holders. For the top
1% in France in 1929, our estimates above imply that capital flight as deposits made up
2.33% of their total domestic wealth (see Table 4). If the value of securities that they moved
to Switzerland was 3 times this value, then both types of assets together totaled 4 times this
amount. So for the top 1%, that means they moved a value equal to 9.33% of their wealth
12 held in France, or in other words, 8.53% of their total wealth (in France and Switzerland
combined) to Switzerland. (And this was 5.64% of all private wealth held in France.) This
might seem low, but of course not all wealth was easily moved, unless it was sold and the
proceeds of the sale deposited in Swiss banks. Piketty (2014, Chapter10TablesFigures, Table
S10.4) shows that for the top 1% of Parisian households, foreign equity was 9% of wealth
holdings in 1912. If this was representative of French households, this analysis implies that
on average they moved 7 percentage points (over three-fourths) of that 9% to Switzerland.
Moreover, Hautcoeur and Sicsic (1999, FNs 12 and 13) note several independent sources of
estimates of tax evasion on foreign securities in various years, including a statement in the
Chambre des députés in 1922 that implied that tax evasion was about two-thirds of tax
liability, and possibly a bit more. An estimate for 1930 found that about half of these taxes
were evaded. More than half of securities in France were bearer securities, and a tax was
imposed on the issuer of bearer securities, but in addition rentiers (those who lived on the
proceeds) were required to report the income and pay an additional tax on it (Hautcoeur and
Sicsic 1999, 38). However, these authors write, “these rates were probably never paid since it
was very easy to avoid paying IGR [impót general sur le revenu] on bearer securities, because
coupons were paid to the bearer without any requirements enabling his future IGR
declaration to be checked” (Ibid., 40). The flight of foreign securities is consistent with the
reported decline in net foreign assets from 123% of French national income in about 1913
to 6% of national income in 1925, although the repudiation of Russian bonds also
presumably played a role, affecting the Germans and the British, not just the French (Piketty
2014, Chapter3TablesFigures.xls, TS3.2; Keynes 1920); see section V.
What share of falling reported wealth inequality does capital flight explain?
The top 1 percent’s share of wealth in France before World War I (in 1910) was 60.5
percent, and for 1930 is given as 47.4 percent in Piketty (2014, Chapter10TablesFigures,
Table S10.1), a fall of 13.1 percentage points. The cumulative effect of the estimated flight of
both deposits and securities on the top 1 percent’s share of all wealth held in France would
only reduce its reported share from 60.5% to (60.5% - 5.65%)/(100% - 5.65%) = 58.1%, a
reduction of 2.4 percentage points in its share of wealth in France. Thus even the entirety of
French capital flight estimated in these categories would appear to account for less than onefifth of the 13.1 percentage point fall in the top 1%’s share of total wealth from before
World War I to 1929.
For the top 1%, however, the over 8% of their wealth that this estimate concludes
they moved to Switzerland was substantial. In addition, however, there were other ways and
other places to hide wealth; and there were other potential benefits to capital flight besides
directly avoiding taxation on the capital moved abroad.
However, one other potential benefit was reducing tax liability by using loan-back
schemes, as Abegg, Baltensperger, et al. (2007) mention in the passage quoted above. For
example, a French depositor’s flight capital in Switzerland could be lent back to that person
in the guise of an arm’s-length loan, creating interest payments that could reduce the tax
liability of the French depositor; the liability would also be deducted from the depositor’s
reported net worth. Piketty and Zucman (2014, Data appendix, 83-84) remark on the very
high liabilities that appear in the German data, and note that, according to official wealth
data gathered for the German “defense tax” (Wehrbeitrag) imposed in 1913 to help finance
the anticipated war, reported liabilities constituted over 20 percent of the gross value of
13 assets, and therefore over 25 percent of reported net worth. This was so large that the
researcher upon whose work Piketty and Zucman drew adjusted the value of assets upward,
assuming that much or all of this was tax evasion. Indeed, loan-back schemes are mentioned
frequently in the capital flight literature.
As an upper bound, if all French flight capital were lent back and the “loans” so
generated were reported as liabilities and deducted from net worth held in France, then they
evidently could nearly double the contribution of capital flight to the decline in the share of
the top 1 percent in total wealth held in France; the reduction in tax payments would reduce
the loss in net worth due to deduction of the liabilities. That is, the consequent top 1
percent’s lower share of reported total wealth due to capital flight would be (60.5% 2⋅5.65%)/(100% - 2⋅5.65%) = 55.5 percent, and capital flight to Switzerland would therefore
account for 5.0% of the 13.1% decline, or nearly two-fifths of it. (However, if they were
successful in reducing tax liability, that could lessen the reduction in net worth.)
Moulton and Lewis (1925, 27 and 32) provide useful data on the foreign investment
and debt position of France in mid-1919 and at the end of 1923. The December 31, 1923
debt position lists “War and post-war sales of domestic securities and real estate” with a
value of 21.0 billion francs, while “Paper francs held by foreigners” are 15.0 billion francs.
Together these constituted about 20 percent of the total foreign debt of 180.4 billion francs.
(The remainder is nearly all debt to the UK and US, including capitalized interest). With the
franc in 1923 worth 5.25 cents of a US dollar (League of Nations 1931), this means paper
francs held by foreigners were $0.79 billion (a little more than the $0.54 billion in implied
flight capital as deposits in Swiss banks estimated above – which is natural because the $0.79
billion refers not just to Switzerland). The sales of domestic securities and real estate to
foreigners could account for more assets held abroad, if some or most were sold to entities
whose beneficial owners were French; but French-owned foreign assets placed abroad, such
as in Swiss trust companies, were evidently in a different category. The same source for 1919
lists “Losses on other pre-war investments” of 23.0 billion francs (about half of the 45.0
billion franc value of all pre-war investments), or (at 1 franc = .092 cents in 1919), about
$2.1 billion, reminding us that some of the losses were, in fact, genuinely losses. An
interesting question is what happened to assets that began to lose value: to what extent did
their owners sell them (perhaps at a loss) and to what extent did they simply hold onto them
in the hope that they would later recover their value? And did wealth-holders ever sell assets
at a loss to “themselves”, disguised as foreign entities, to keep the assets as part of their
wealth but be able to report reduced net worth?
For simplicity several factors have been omitted from the analysis. One is that
deposits in commercial banks in the Netherlands approximately quadrupled from 1913 to
1919, although they leveled off after that; and we know that the Netherlands also had some
of the features of a tax haven, such as a quite stable currency and apparently a relatively low
tax rate (Farquet 2012), and so suspect that at least some of this was capital flight. In fact,
Farquet says, “Switzerland was positioned at the forefront of tax dumping in presenting,
maybe with the Netherlands, the most favourable conditions for imported capital” (13). For
these and other reasons, the estimate here should be regarded as a lower bound for the
amount of capital that remained in possession of the top 1%, but shifted from being
14 reported to being unreported either because it fled to tax havens or was hidden in some
other way.8
V. Piketty’s explanations for declining β
Piketty (2014) offers two kinds of information relevant to explaining the decline in β
during 1910-1929. One, in chapters 3-4, is a decomposition of β into four categories: net
foreign assets, housing, other domestic assets, and land, each of which contributes to overall
β. For France, the contribution of net foreign assets to β reportedly fell from 123% to 6%
between “1910” and “1920” (data points for wealth are for 1913 and 1929, but were
transformed as described in the online appendices to Piketty, Postel-Vinay, and Rosenthal
2006), the most dramatic decline of the four categories (Piketty 2014,
Chapter3TablesFigures.xls, TS3.2). The contribution of housing fell from 168% to 90%, of
other domestic capital assets from 249% to 134%, and of (apparently agricultural) land fell
from 142% to 60% (Ibid.). Evidently the only two of these four categories in which capital
flight might show up were net foreign assets and “other domestic assets” (a category that
presumably includes financial deposits). Overall, the ratio of private wealth to national
income fell from 692% to 293%, a fall of well over half. (This does not quite match the sum
of the four above, which refer to national wealth rather than private wealth; but the
difference is small.)
For Germany, data are not available for 1920, only for 1909-1911 and then for 1927
(see Piketty 2014, Chapter4TablesFigures.xls, TS4.1); they are displayed as 1910 and 1930 in
the figures. The overall decline in β was somewhat smaller for this period – from 608% to
275% – than it was for France for a shorter period (Piketty and Zucman 2014, Germany.xls,
TableDE6f), but this may be because after a deep trough there was a rise again by 1927; at
least the data on bank assets and liabilities in League of Nations (1931) show a drop to
around zero in 1923, followed by a recovery by 1928 (of deposits expressed in US dollar
equivalents) to above their 1918 level. As for the four categories of assets, the contribution
of net foreign assets to β fell from 39% to -11%, while that of housing fell from 125% to
64%, that of land fell from 138% to 48%, and that of other domestic capital assets fell from
337% to 234% (Ibid.). It is perhaps worth noting that, given rapid inflation, what actually
happened is that private wealth only about doubled in value while national income was
multiplied by 5. It is unclear what, if anything, compelled assets to be “marked to market” by
their owners, and thus rapid overall price inflation could account for some of the reported
decline in β (corresponding to the negative “rate of capital gains”), and it is unclear whether
this reflected the true value of the assets.
It is worth mentioning one caveat regarding the data: Some foreign bonds, such as
those of Austria, whose value collapsed after the war due to skepticism that Austria would
ever resume coupon payments, did in fact start paying around 1926, due to agreements
reached at the urging of Allied governments and organizations of bondholders (Council of
the Corporation of Foreign Bondholders, UK 1929). Thus some foreign assets valued at or
8
Keynes (1920, Ch. 5, III.1.c.iv) estimates that at least $0.5 billion of securities in Germany
either fled or were hidden within Germany by the time of his writing. The estimate here
implies a much larger amount: $1.2 billion flight of financial deposits from Germany to
Switzerland and $3.6 billion of securities.
15 close to zero around 1920 recovered value later on. However, Piketty’s data on these include
no entries between 1920 and 1950, so this is not reflected in the decomposition figures.
Piketty’s analysis of influences on β, 1910-1929
In Piketty (2014), the various discussions of the causes of falling β 1910-1950 (at
least pages 41, 236, 237, 275-6, 369, and 396), as well as Piketty and Zucman (2014, 1260),
offer a list of close to a dozen causes that affect β. Among these, we ignore here those that
seem to refer mainly the post-1929 period: bankruptcy, nationalization (a reference to postWorld-War-II French and British nationalizations, evidently), and substantial wartime
destruction in Germany during World War II.
Of the remaining factors, the main ones seem to be these four:
(1) CPI inflation that exceeded asset price inflation, which evidently so
deeply eroded the value of government bonds fixed in nominal terms that Piketty
refers to “the inflation-induced euthanasia of the rentier class that lived on
government debt” (2014, 275);
(2) Rent control, which was apparently one cause of a sharp decline in the
reported value of housing;
(3) The repudiation after the Russian revolution of several billion dollars’
worth of Russian bonds by the revolutionary government, reducing their value to
zero (Keynes 1920); about $1 billion were held in the UK and another $1 billion or
so in Germany, at least; and
(4) A fall in saving rates, evidently including personal savings, retained
earnings, and government savings – although it is possible that this item also reflects
some capital flight, which could exaggerate the fall in savings.
It is not completely clear how the quantitative decomposition of the evolution of β
described at the beginning of this section is related to all of the above factors, although
possibly further intensive study of the data appendices and tables would lead to greater
insight on this point.
In any case, the main additional causes that Piketty mentions are
(5) An increase in taxation (although the size of the net effect, given some tax
avoidance, is unclear), which could reduce private wealth;
(6) Wartime physical destruction, although this was not a large factor for any
country in World War I, the highest estimate being for France; and
VI. Conclusion
There is little doubt that all of the following factors played some role in the decline
in β during and after World War I: loss in reported real value of assets due to a large rise in
inflation, repudiation of Russian debt after the 1917 Revolution, a temporary or permanent
collapse in values of other foreign assets, and capital flight. Changes in saving behavior may
have played some role as well, although it is unclear how much, or whether some capital
flight would show up as a decline in saving rates. There are two factors that might have
16 mitigated the impact on the top 1% of wealth-holders, however. One was the certainty (since
it is mentioned in the Swiss National Bank history) that some, perhaps many, used tax haven
facilities for loan-back schemes to reduce tax liability by not only reducing their reported
wealth once by removing it and putting it into the guise of third-party ownership, but
reducing their reported net worth by up to an equal amount through lending it back to
themselves, at interest. The second is that, as Moulton and Lewis report, a substantial value
of domestic securities and real estate were sold during the war to foreign residents, some of
which (again) may have been created in tax havens as apparent third-party entities, but with
beneficial owners actually being French. Without knowing what were the reporting practices,
we do not know how this would have showed up in the data. France’s net foreign debt is
reported in Moulton and Lewis (1925) as 22.3 billion francs in mid-1919, but 128.8 billion
francs in 1923, an enormous (but smaller) increase from $2.0 to $6.8 billion in US dollars.
This paper arrives at a rough estimate for capital flight as a cause of declining β and
of declining reported wealth and income inequality, arguing that, in France at least, less than
6 percent of total French private wealth (but nearly 10 percent of the wealth of the top 1
percent) was in Swiss banks and trust companies by 1929. This could explain perhaps 40
percentage points of the over 400 percentage point decline in β in France – or perhaps
nearly 80 percentage points, if all the flight capital were lent back to its owners and reported
as liabilities. The estimate is consistent with the fact that it was the top 1% whose share of
total wealth plunged.
This research has arrived an estimate of flight capital in Switzerland in 1929 from
France, Germany, and Italy, and has shown for France that under reasonable assumptions,
the implied share of their financial deposits that the top 1 percent of wealth holders held in
Swiss banks was a plausible 23 percent. It has further estimated that this group probably held
nearly 9 percent of their total wealth in Switzerland, of which around three-fourths was
securities and one-fourth was financial deposits. If this wealth was lent back to its owners in
the guise of arm’s-length loans and used to reduce tax liability, the impact on reported net
worth could be almost double – or perhaps a little less, due to the reduction in taxes. In
addition, these estimates are probably low, in light of other locations that reportedly received
flight capital, such as the Netherlands, and in light of other ways to hide wealth within one’s
country of residence. The analysis here has relied heavily on data from France, while data
from Germany are relatively sparse, and from Italy seem to be unavailable. The point is not
that this is the correct figure; there are far too many uncertainties to be at all certain about
the size of capital flight. The point, rather, is that estimating capital flight by triangulating
with the various pieces of information available is a useful exercise that confirms the
plausibility of capital flight as a partial explanation for the reported decline in wealth/income
and the reported decline in wealth and income inequality in France and Germany during
1910-1929. In other words, the space of parameters that fit all the known data about
possible capital flight is non-empty.
A number of these possibilities, and the overall story, could be further checked
(among other methods) by extensive searching in contemporary periodicals, including
financial journals, and by looking carefully at trends in reported liabilities during the period.
That remains for future research.
17 References
Abegg, Werner, Ernst Baltensperger and others. 2007. The Swiss National Bank, 1907-2007.
Zurich, Switzerland: Neue Zürcher Zeitung Publishing.
Colson, C. 1903, 1918, 1927. Cours d’économie politique. Several editions. Paris: Gauthier-Villars.
Council of the Corporation of Foreign Bondholders, UK. 1919, 1920, 1929. Annual Report.
Digital archive, Stanford University Library.
Divisia, F., J. Dupin, and R. Roy. 1956. A la recherche du franc perdu. (volume 3 : Fortune
de la France), Paris: Serp, 1956.
Eichengreen, Barry. 1992. Golden Fetters: The Gold Standard and the Great Depression 1919-1929.
Oxford: Oxford University Press.
Farquet, Christophe. 2012. “The Rise of the Swiss Tax Haven in the Interwar Period: An
International Comparison.” European Historical Economics Society (EHES)
Working Papers in Economic History, No. 27. Oktober.
Fehrenbach, R. R. 1966. The Gnomes of Zurich. London: Leslie Frewin.
Guex, S. 1993. La politique monétaire et financiére de la Confédération suisse. 1900-1920. Lausanne:
Payot. Cited in Farquet (2012).
Hautcoeur, Pierre-Cyrille, and Pierre Sicsic. 1999. “Threat of Capital Levy, Expected
Devaluation and Interest Rates in France during the Interwar Period.” European
Review of Economic History 3: 25-56.
Kenwood, A. G., and A. L. Lougheed. 1999. The Growth of the International Economy 1820-2000:
An introductory text. London: Routledge.
Keynes, John Maynard. 1920. The Economic Consequences of the Peace.
http://www.gutenberg.org/files/15776/15776-h/15776-h.htm
League of Nations, Economic Intelligence Service. 1931. Memorandum on Commercial Banks
1913-1929. Geneva.
Maddison, Angus. 2010. Historical Statistics of the World Economy, AD1-2030, horizontal
file_02-2010.xls
Moulton, Harold G., and Cleona Lewis. 1925. The French Debt Problem. New York:
Macmillion Company.
Naylor, Robin Thomas. 1987. Hot Money and the Politics of Debt. New York: The Linden
Press/Simon and Schuster.
Palan, Ronen, Richard Murphy, and Christian Chavagneux. 2010. Tax Havens: How
Globalization Really Works. Ithaca, NY: Cornell University Press.
Piketty, Thomas. 2014. Capital in the 21st Century. Cambridge, MA: The Belknap Press of
Harvard University Press.
Piketty, Thomas. 2011. “On the long-run evolution of inheritance: France 1820-2050.”
Quarterly Journal of Economics 126(3): 1071-1131.
18 Piketty, Thomas, Gilles Postel-Vinay, and Jean-Laurent Rosenthal. 2006. “Wealth
Concentration in a Developing Country: Paris and France, 1807-1994.” American
Economic Review 96(1, March): 236-256.
Piketty, Thomas, and Gabriel Zucman. 2014. “Capital is Back: Wealth-Income Ratios in
Rich Countries 1700-2010.” Quarterly Journal of Economics 129: 1255-1310.
Zucman, Gabriel. 2013. “The Missing Wealth of Nations: Are the US and Europe Net
Debtors or Net Creditors?” Quarterly Journal of Economics 128(3): 1321-1364.
19 Table 1. Approximate amount in dollars per head of population of commercial banks’ total
assets and deposits, of deposits in other credit institutions, and of aggregate deposits in 1929.
Switzerland
US
Denmark
Norway
Sweden
UK
Scotland
England & Wales
Ireland (Free State & N. Ireland)
Netherlands
Czechoslovakia
Belgium
Germany
Austria
Finland
France
Italy
Commercial bank
deposits
580
351
152
101
153
250
238
205
114
53
102
62
57
56
60
41
Other credit institutions
deposits
134
87
161
212
139
53
Aggregate
deposits
714
438
313
313
292
292
7
61
81
30
60
38
38
30
35
212
175
134
132
122
95
94
90
76
Source: League of Nations (1931), p. 8, Table 1.
Notes:
Canada's commercial bank deposits and aggregate deposits exclude deposits in branches abroad.
Commercial bank figures for the UK include private banks, but exclude British overseas banks and English
branches of foreign banks.
UK figures in the two right hand columns include discount houses, trustee, and P.O Savings Banks, including
those in N. Ireland.
Australia commercial bank figures exclude offices abroad, and the two right-hand columns include deposits
with the Commonwealth Bank.
New Zealand commercial bank figures exclude offices abroad.
The first two columns for the Netherlands are based on the assumption that the leading banks account for
60% of commercial banking.
In the first two columns for Germany the estimate for all commercial banks, including Staats- and Landesbanken,
is based on the last official interim return for 1929.
The two left-hand columns for France are based on the assumption that the leading banks account for 75% of
total commercial banking. The two right-hand columns for France are for savings banks only; banques
d’affaires are included under the first two columns.
20 Table 2. If flight capital from Germany, France, and Italy to Switzerland were 14.7% of each
country’s domestic deposits, and there were no flight capital in Switzerland from any other
country, then flight capital would be about 75% of Swiss deposits in banks and financial
institutions, and the remaining 25% would amount to $178 per capita for Swiss resident
owners of the deposits. All monetary data are in current US dollars of 1929.
Per
capita
aggregate
deposits
Switzerland
All 13 countries
Germany
France
Italy
714
178
122
90
76
Country
per
capita
possible
flight K
millions at
midyear
1929
Population
Aggregate
deposits
(millions)
Flight capital
portion of
Swiss deposits
(millions)
536
4.022
2,871.708
2,155.685
18
13
11
64.739
41.230
40.469
7,898.158
3,710.700
3,075.644
14,684.502
14.7%
14.7%
14.7%
Total
Ge, Fr, It deposits
Flight capital
(000s)
1,159,450
544,731
451,505
2,155.685
Ge, Fr, It
flight
capital
Source: Author’s calculations from League of Nations (1931), Table 1, p. 8, and population data from
Angus Maddison, Historical Statistics of the World Economy, AD1-2030, horizontal file_02-2010.xls
21 Table 3. Share of all deposits in France, by wealth group, based on “1910” data on wealth
distribution and wealth composition. Column [B] (financial assets as share of a group’s
wealth) is from Table 10.1 (Piketty 2014), data for 1912. Calculations assume deposits were
the same share of “other financial assets” for each group of wealth-holders. The bottom
50% of wealth-holders are assumed to have zero net worth, on average, and no deposits.
“Deposits” refer only to deposits held in France.
Group’s wealth
Total wealth
(1910)
[A]
Group’s deposits
Group’s wealth
(1912)
[B]
Group’s deposits
Total wealth
Group’s deposits
Total deposits
[C] = [A] x [B]
Wealthholders
Top 1%
[4.84%/9.49%,
2.52%/9.49% …]
60.5%
8%
4.84%
51.0%
Next 9%
28.0%
9%
2.52%
26.6%
Next 40%
12.5%
18%
2.25%
22.4%
All deposits
All wealth 9.61%
100.0%
Sources: Author’s calculations from Piketty (2014, Table 10.1, 371) and online
Chapter10TablesFigures.xls at piketty.pse.ens.fr
22 Table 4. Estimated capital flight of French residents in the form of deposits and securities
in Swiss financial institutions. The column heading is the numerator and the row label is the
denominator for each estimate.
1%'s deposits in SWI
1%'s total deposits in France
1%'s total wealth in France
All deposits in France
All wealth in France
1%'s total deposits (in FRA+ SWI)
1%'s total wealth (in FRA+SWI)
All wealth (in FRA+ SWI)
29.15%
2.33%
14.68%
1.41%
22.57%
2.13%
1.39%
1%'s securities in SWI
1%'s deposits + securities in SWI
7.00%
9.33%
4.23%
5.64%
6.40%
4.06%
8.53%
5.34%
Source: Author’s calculations; see Tables 2-3 and assumptions explained in the text.
SWI = Switzerland; FRA = France. The estimates are rough, and not to be viewed as valid
to two decimal places; the decimal places are included to make it easier to check and grasp
the relationships among variables. All securities figures in column 2 are three times the figure
in the same row in column 1, by assumption.
23 Figure 1. There was a precipitous decline during 1913-25 in the ratio of private wealth to
national income (βp) in France and Germany, accounting for most of the decline 1910-1950.
White markers are actual data; black markers are imputed data.
Source: Author’s plot of data in piketty.pse.ens.fr, Piketty and Zucman (2014), Table A5 in
AppendixPikettyZucman2013.xls, using explanation in Data Appendix.
24 Figure 2. The share of the top 1% and the top 10% in total wealth declined in France,
Sweden, and the UK, 1910-1930. In all three countries, the top 10% (the top three lines)
held 80% or more of all wealth, and the top 1% held 45% to 70% of all wealth (the three
lines in that range). In France, the only country for which we have separate data for the top
0.1% (the lowest line), this group held between 20% and 30% of all wealth.
Source: piketty.pse.en.fr/Capital21c, Chapter10TablesFigures.xls, TS10.1
25 Figure 3. In France, as in some other countries, the top 1%’s share of total wealth declined
from 1910 to 1930, but the share of the next 9% actually rose. Here the top 1% is divided
into two groups: the top 0.1%, whose share fell from 29.0% to 22.4%, and the next 0.9%,
whose share fell from 31.5% to 25.0%, with most of the decline taking place during the
1910-1920 period. The next 9%’s share rose from 28.0% to 32.6%.
Source: Author’s calculations from piketty.pse.en.fr/Capital21c, Chapter10TablesFigures.xls,
TS10.1.
26 Figure 4. Currency depreciation was a motive for capital flight to Switzerland; the graph
shows currency values relative to the Swiss franc, 1913-1929. The currencies of Switzerland,
the Netherlands, and the UK remained fairly stable, while the value of the German mark fell
to zero by 1923 with Weimar Republic hyperinflation, and the currencies of Belgium and
France also fell rapidly in value.
Source: Author’s calculations from Farquet (2012), 35 (Appendix 2, CHF exchange rates).
Data points for 1914 -1917 are missing for all countries; 1918 is also missing for Belgium.
27 Figure 5. The marginal tax rate on the highest incomes was increased sharply in Germany,
France, and the UK to help pay for the war. Germany’s formal tax increased latest, but
wealth taxes had existed since the late 18th century in Prussia and elsewhere. In addition, the
German government requisitioned gold from households during the war to help finance war
spending; many hid their wealth rather than turn it over to the government (Keynes 1920).
Source: Piketty (2014), piketty.pse.ens.fr/Capital21c, Chapter14TablesFigures.xls, TS14.1.
28 Figure 6. Top inheritance tax rates for Germany and France, 1910-1929.
Source: Piketty (2014), piketty.pse.ens.fr/Capital21c, Chapter14TablesFigures.xls, TS14.2.
29 Figure 7. Switzerland’s tax collection was below that of neighboring countries during the
entire period 1913-1929. Its total tax revenues as a share of GDP, for the years reported in
Farquet (2012), were lower than in France, Germany, or the UK, creating an incentive for
capital flight to Switzerland.
Source: Farquet (2012). Thin dotted lines connect years before and after missing data.
30 Figure 8. Per capita aggregate deposits in banks and other financial institutions such as credit
unions, top 13 countries of Europe, the UK, and Scandinavia, 1929, in current US dollars. If
only $178 of Swiss per capita deposits in 1929 (equal to the average of all 13 countries) was
owned by Swiss residents, and the remaining three-fourths of Swiss deposits were flight
capital exclusively from France, Germany, and Italy, then the flight capital from each of
these countries would amount to about 15% of its domestic deposits. See Table 3.
Source: League of Nations (1931), Table 1, p. 8, and author’s calculations. See also Tables 2
and 3.
31 Figure 9. The ratio of total liabilities of six big Swiss banks to those of six big French banks,
1913-1929, after conversion to current US dollars, rose substantially, a fact consistent with
capital flight to Swiss banks, including from France. Data were not reported for 1914-1917.
Source: Author’s calculations from League of Nations (1931, 112-113 and 271).
32