The Swiss National Bank

The Bank of England, the Banque de France, the Reserve Bank of India: all started as private corporations and were nationalized over the course of the twentieth century. The Swiss National Bank, founded in 1907, never was. Its shares still trade on the SIX Swiss Exchange.

The Swiss National Bank

The Swiss National Bank (SNB) is a special-statute joint-stock company. Its primary mandate is to pursue a monetary policy that serves the overall interests of the country. The SNB was founded under the law on 6 October 1905 ¹, which entered into force on 16 January 1906, and it officially commenced business on 20 June 1907.^{2 3}

The SNB has 100,000 registered shares at CHF 250 nominal value, traded on SIX. About 75% are held by cantons and cantonal banks as quasi-permanent positions; private shareholders hold ~24%. Monetary policy is set by the Governing Board, whose members are appointed by the Federal Council; shareholders elect five of eleven Bank Council members (the supervisory body, responsible for oversight of management and approval of the annual accounts) but have no influence over monetary policy decisions. The dividend is capped at 6% of nominal (CHF 15 per share); any remaining profit goes one-third to the Confederation and two-thirds to the cantons. As of late 2025, total assets were CHF 854 billion (88% foreign-currency reserves, 9% gold), with CHF 143 billion in equity, or CHF 1.43 million per share. The SNB has paid its CHF 15 dividend every year since 1907 except in 2013, 2022, and 2023, when large foreign-currency losses exhausted the distribution reserve.

The share

Shares of the SNB trade on the SIX Swiss Exchange as SNBN. Typically not more than 50 shares change hands per day, and the average bid-ask spread is ~2.7%. With ~25,000 privately held shares and such thin volume, an illiquidity discount would in principle apply; in practice the share has historically traded at a premium to the fair value, suggesting any illiquidity discount is more than offset by other premia. As of late 2025, the price stands above CHF 3’000, which prompts the question, how much are these shares really worth?

The price

The most direct starting point is the price itself.

Discounted cashflows

A share of the SNB can be modeled as a perpetual bond with a 6% coupon, treating the CHF 15 annual payment as the perpetual cash flow.⁴ The three missed dividends (2013, 2022, and 2023) appear as episodes in the historical z-spread record and are discussed below. More broadly, the z-spread has risen from deeply negative toward zero in recent years, signalling that the market is pricing in higher risk of another interruption. The SNB share is modeled as a perpetuity, discounted at the long end of the Swiss sovereign curve (100-year rate, extrapolated flat). The yield curve is extrapolated using a two-factor Nelson-Siegel model. The entire Swiss sovereign yield curve was negative during 2020-2021, so a moving average is applied to smooth out these values.

Liquidation event

Were the SNB liquidated, shareholders would receive the book value of equity pro rata.

The nationalization bet

Today, central banks with any private shareholding are the exception. As of late 2025, the list includes Japan, Switzerland, Belgium, Greece, South Africa, Liechtenstein, San Marino, Italy, the United States, and Turkey. For most of the twentieth century, the list was much longer, and then governments nationalized them.

• The Bank of England was nationalized in 1946 ⁵. Its ~17’000 shareholders were issued 3% perpetual government bonds (callable after 1966). The face value of each share was GBP 100 and it historically had a dividend of ~12%, so each shareholder received GBP 400 of the perpetual bonds per share.
• The Bank of France was nationalized in 1946 ⁶. The compensation to shareholders was set to be negotiable bonds, the face value “equal to the liquidation value of the share […] [which] shall not exceed the average quoted price during the period September 1, 1944 […] and the date on which the liquidation value shall have been fixed”. The bond was to be amortized in less than 50 years and the interest rate should not exceed 3% (so as to not exceed the previous dividend rate).
• The Reserve Bank of India was nationalized in 1949 ⁷. The compensation was set at the average monthly market price over March 1947 to February 1948 (approximately INR 118.6 per share), paid in 3% government bonds of equivalent value. The dividend had been capped at 3.5%, so the payout was close to market price.
• The Central Bank of the Argentine Republic was nationalized in 1946 ⁸. The compensation to shareholders (private banks) was established to be equal to their contribution (interpreted as the nominal value of the shares) and to be paid with 2.5% government bonds (without specified maturity) or in cash. The face value of the shares was approximately half of their book value.
• The Bank of Canada was nationalized in 1938 ⁹. The face value of the shares was CAD 50, and the book value was slightly higher. The compensation to shareholders was CAD 59.20 per share.
• The Cuban National Bank ¹⁰, the State Bank of the Russian Empire ¹¹, the various State Banks of China ¹² were never privately owned and thus never nationalized.
• Privately held shares of the Bank for International Settlements (sometimes referred to as the “central bank of central banks”) were mandatorily repurchased in 2001 ¹³. The total number of shares was 529’165 (a small fraction of which were in private hands). The payment was CHF 16’000 per share (~USD 9’950), representing a 95%-155% premium over market price. The face value of the shares was 2’500 gold francs (25% paid up) (~USD 4’850, GF 1 = USD 1.94149 at the time). The book value of each share was GF 6’548 or USD 12’712 (GF 3’465 million total equity). The dividend per share was approximately GF 95 or USD 183.

Swiss law defines a hard floor for what shareholders may receive if the SNB is ever transferred to the Confederation or liquidated. The framework has changed twice. Under the original 1953 Nationalbankgesetz (Art. 68), shareholders were entitled to receive back their paid-up share capital (CHF 250 per share), interest at 5% for the duration of the liquidation, and up to one-third of the reserve fund, capped at 10% of paid-up capital, i.e. CHF 25 per share, for a maximum statutory payout of approximately CHF 275 per share. Any surplus of assets passed to the new central bank; shareholders had no claim on it. The 2004 revision (Art. 32) made this more explicit and less generous: shareholders receive only the nominal value of their shares (CHF 250) plus reasonable interest from the dissolution decision; the act explicitly states that “shareholders shall have no additional rights to the assets of the National Bank.” There is no BIS-style premium, no access to gold or foreign-exchange reserves, and no mechanism for shareholders to block a liquidation once enacted by Parliament.¹⁴

The statutory payout is modelled as a 10-year bond: CHF 275 face (pre-2004) or CHF 250 face (post-2004), with a 10% annual coupon (above the law’s stated 5% liquidation interest, but used here as a rough upper bound on what a court might award), discounted at prevailing rates. This gives a present value modestly above face. For periods before the 1953 act, international comparables are used directly. The model then combines the Swiss-law floor with the international-comparables estimate in equal weights (50/50), reflecting both the hard statutory minimum the Confederation would owe and the premium a negotiated buyout might command, as illustrated by the BIS precedent. The comparables are equally weighted across countries, since compensation regimes in practice reflect legal and political context more than economic size; historically they tracked some combination of face value, market price, and book value.

The composite model

The nationalization probability is estimated from historical data: approximately 56 central bank nationalization events have occurred over the 357-year history of central banking (1668-2025).¹⁵ Dividing by the approximate average number of central banks with private shareholders over that period (~15) gives a per-bank annual hazard rate \(h = 56/(357 \times 15) \approx 1.0\%\). At this rate, the probability of the SNB remaining private for its 118-year history is approximately 29%. In a Poisson model, the composite weight on the nationalization payout is \(h/(r+h)\), which varies with the prevailing interest rate \(r\) rather than being a fixed constant.

If nationalization arrives as a Poisson process with annual hazard rate h, the share value is the expected present value of dividends until nationalization plus the payout at nationalization:

\[\text{value}_{\text{all}} = \frac{r}{r+h} \cdot \text{value}_{\text{DCF}} + \frac{h}{r+h} \cdot \text{value}_{\text{nat}}\]

where \(h = 1.0\%\) is the per-bank annual hazard rate and \(r\) is the prevailing long-term interest rate. The weight on the nationalization component, \(h/(r+h)\), is time-varying: when rates are low it rises and the nationalization floor dominates; when rates are high it falls and the DCF value dominates. Book value enters only through \(\text{value}_{\text{nat}}\), which is derived from the statutory payout and international comparables; absent a nationalization or liquidation event, book value is not directly accessible to shareholders.

An alternative model fits a log-linear regression on the data:

\[\text{value}_{\text{fit}} = \exp(0.034 \cdot \ln(\text{value}_{\text{book}}) + 1.044 \cdot \ln(\text{value}_{\text{DCF}}))\]

Empirical fit

The market price tracks the DCF value closely but trades at a persistent premium, reflecting the market’s estimate of some positive probability of a nationalization or extraordinary distribution event.

The composite valuation, DCF coupon plus nationalization option, implies a testable prediction: the z-spread (the implied yield, CHF 15 divided by the market price, minus the prevailing risk-free rate) should track changes in the market’s implicit estimate of the probability of a government-triggered event, or of what shareholders would receive if it happened.

Statistical evidence

Fitting \(\log(\text{value_price})\) as a linear function of \(\log(\text{value_dcf})\) alone across 116 years of monthly data (N = 1388) explains 69% of price variation (centered \(R^2 = 0.69\)). Adding \(\log(\text{value_book})\) raises this to 72%: both coefficients are statistically significantly positive, with the DCF coefficient at 1.04 (close to one, meaning price tracks DCF value nearly proportionally) and the book value coefficient at 0.03, consistent with the nationalization floor adding a modest premium above the DCF value.

A natural follow-up question is whether changes in fair value predict future price changes. Regressing 2-year and 10-year price log-returns on the corresponding log-returns of each component shows that DCF value is the dominant predictor.¹⁶ Book value returns have no detectable predictive power at either horizon, which is expected: book value grows near-monotonically over the full sample, so changes in book value carry very little information about future price changes.

The regime split tells a coherent economic story. Running the same regression before and after Bretton Woods (split at 1947) shows that both components enter positively in both periods, but their relative weights shift. Before 1947, book value was the dominant term (\(\beta_{\text{book}} = 0.58\), \(\beta_{\text{DCF}} = 0.46\)), consistent with a period when the SNB’s balance sheet was largely gold and real assets and the dividend yield was low and stable. From 1947 onward, DCF value becomes the dominant driver (\(\beta_{\text{DCF}} = 1.02\)), while book value remains positive but smaller (\(\beta_{\text{book}} = 0.05\)).

Historical evidence

• 1919-1921: real-asset premium suppressed the z-spread. Swiss Confederation bond yields surged from ~5.75% in 1919 to ~7.25% in 1920, as post-war fiscal expansion, wartime debt issuance, and the SNB’s pursuit of gold-parity restoration kept rates elevated.¹⁷¹⁸ The SNB share price barely moved (CHF 465-490 in 1919-1920, falling to CHF 395-410 by 1921), leaving the implied yield far below the risk-free rate. The SNB’s gold reserves had quadrupled during WWI (CHF 170M in 1913 to CHF 664M in 1920), making the share a credible claim on real assets at a time when paper currencies were distrusted.¹⁹

• 1937-1938: nationalization option discounted. The z-spread was higher than usual in late 1937 through early 1938; the share was modestly cheap relative to its DCF value. Swiss bond yields had begun rising from their mid-1930s floor after the September 1936 franc devaluation ended the Gold Bloc’s deflationary trap. But the deeper reason the share lagged is that pre-WWII political uncertainty appears to have depressed the nationalization option: capital markets in Zurich were already pricing rising war risk by 1938,²⁰ and for a share like the SNB’s, a wartime intervention could as easily have meant confiscation, capital controls, or forced conversion below fair value as an orderly premium buyout. With the nationalization option weighted toward adverse scenarios, the premium above DCF value was smaller than usual.

• 1964-1975: rising nationalization probability and a decade-long floor. From early 1964 to mid-1975, the z-spread was persistently below its norm. As Swiss rates rose from ~4% to ~7.5% through the oil shock, the DCF value fell toward CHF 200 while the expected nationalization payout, anchored to book value, stayed high. Meanwhile the probability of nationalization genuinely rose: Switzerland was drafting a new Nationalbankgesetz throughout this period (enacted 1978), the Federal Council was intervening in SNB operations via emergency decrees, and capital inflow controls (including a 1972 ban on non-residents purchasing CHF bonds²¹) signalled creeping government assertion over the monetary system. Switzerland floated the franc in January 1973,²² briefly intensifying the distortion. The anomaly unwound from 1975 as SNB tightening brought rates back down and the new law clarified governance without triggering a buyout.

• 1997: special dividend priced as imminent, incorrectly. In the first half of 1997, the SNB share price roughly quadrupled and the z-spread fell far below its historical norm. The trigger was a 5 March 1997 speech by Federal President Arnold Koller proposing a Stiftung solidarische Schweiz (Swiss Foundation for Solidarity), to be funded by revaluing the SNB’s 2,590 tonnes of gold from the frozen statutory parity of CHF 4,595/kg to 60% of market price (a revaluation gain of approximately CHF 14 billion). Investment newsletters argued shareholders would receive a windfall; in model terms, they were pricing nationalization as near-certain, with a large expected payout. This was incorrect: the SNB’s dividend is capped at CHF 15 per share by law regardless of profits; any revaluation gain above the statutory dividend goes to the federal government and cantons. The share price gave back most of the gain within the year.²³ Notably, the book value shown in the model does not jump in 1997: the gold remained on the SNB balance sheet at the old parity until the constitutional change was approved in April 1999 and gold sales began in May 2000, at which point the book value does show a ~70% step up.

• 2010-2019: lower rates, book losses, and a balance sheet explosion. Three forces pushed the z-spread below its historical norm simultaneously. First, the SNB’s foreign exchange interventions expanded the balance sheet from ~CHF 100B (2008) to over CHF 800B (2019), lifting book value per share from ~CHF 400K to ~CHF 1.4M and raising the model’s expected nationalization payout. Second, Swiss long rates collapsed from ~4% toward zero, mechanically compressing the z-spread as the DCF value rose. Third, the large foreign-currency portfolio exposed the SNB to mark-to-market losses: the three missed dividends in 2013, 2022, and 2023 were a direct consequence, and by the mid-2010s markets were beginning to price in recurring interruption risk, pushing the z-spread further toward zero. Against this backdrop, a speculative overshoot occurred in 2017-2018, driven by the same error of pricing nationalization as near-certain: the price reached an all-time high of CHF 9,760 on 5 April 2018, roughly double the composite fair value. This coincided with aggressive buy recommendations from German investment newsletters, most notably Hans Bernecker’s Actien-Brief, explicitly invoking a BIS-style premium buyout, and with the exceptionally thin float (~25,300 privately held shares, daily volumes often below 60 shares) amplifying any order flow. The NZZ warned of bubble risk as early as September 2017.²⁴ At the April 2018 annual general meeting, Bank Council President Jean Studer gave an unusual speech explicitly warning that SNB shares are “less a conventional investment than a means for shareholders to express solidarity with the institution.”²⁵ The price fell 58% over the following eight months, returning closer to fair value.

• Theo Siegert. The largest private shareholder through this period was Theo Siegert, a German entrepreneur. He co-authored a 2001 NZZ op-ed criticising the BIS buyout price as unfair, suggesting he held BIS shares at the time²⁶, and the speculative thesis behind his SNBN position was a BIS-style repeat: the Bank for International Settlements had bought out its private shareholders in January 2001 at CHF 16,000 per share, a 95-155% premium to market but well below the BIS’s own net asset value of ~CHF 34,000 per share. Siegert and Weber’s critique proved partially correct: the Hague Arbitral Tribunal ruled in 2003 that the buyout price was insufficient and ordered the BIS to pay an additional ~CHF 9,053 per share (plus interest), bringing the final compensation to ~CHF 25,053 per share.²⁷ According to SNB annual reports, Siegert first appeared as a disclosed shareholder at end-2008 with 4,850 shares (4.85%), accumulated steadily to a peak of 6,720 shares (6.72%) at end-2016, then sold ~1,700 shares in 2017-2018 near the price peak, and has held approximately 5,010 shares (5.01%) since 2022.²⁸²⁹

Conclusion

The SNB share is best understood as a perpetual bond trading at a persistent premium. The dividend is fixed at CHF 15 per share by statute; the equity behind it is CHF 1.4 million per share and largely beyond shareholders’ reach. The premium exists because the market assigns a small but non-zero probability to a government-triggered event that would force a distribution above the CHF 15 annual coupon. That probability is what the z-spread measures: when it falls, the market is pricing in more; when it rises, less. History suggests the market gets this wrong at the extremes (most dramatically in 1997 and 2017-2018) and right on average. Given current legislation, the market appears to overestimate the likelihood of generous compensation in a nationalization scenario: the statutory framework caps shareholder entitlements well below the per-share book value, and nothing in the political landscape suggests an imminent change.

Epilogue

The author thanks the SNB archive in Zurich for responding to a request for historical share price data.

This article is for informational purposes only and does not constitute financial, investment, or legal advice. The models and estimates presented are illustrative and carry significant uncertainty. Nothing herein should be construed as a recommendation to buy, sell, or hold any security.

---

1. Bundesgesetz über die Schweizerische Nationalbank (BBl 1905 V 297).↩︎

2. SNB, The history of the SNB. https://www.snb.ch/en/the-snb/organisation/history↩︎

3. Landmann, J. (1906). The Swiss National Bank. The Quarterly Journal of Economics. https://doi.org/10.2307/1882859↩︎

4. When the SNB holds large foreign-currency reserves, CHF appreciation reduces the value of those assets in franc terms and can produce book losses, suspending dividend payments.↩︎

5. Bank of England Act 1946 (9 & 10 Geo. 6. c. 27).↩︎

6. Loi no. 45-015. Loi no. 46-626.↩︎

7. Reserve Bank of India (Transfer to Public Ownership) Act, 1948. Reserve Bank of India, Report of the Central Board of Directors, for the year ended 30th June 1948.↩︎

8. Decreto 8503/1946. Banco Central de la Republica Argentina, Memoria Anual, 1945.↩︎

9. Bank of Canada Act Amendment Act, 1938. Commercial and Central Banks, League of Nations, 1939.↩︎

10. Ley no. 13 de 1948.↩︎

11. Founded 1860.↩︎

12. Central Bank of China founded in 1928. Various other government-controlled currency-issuing banks also existed.↩︎

13. 71st Annual Report, Bank for International Settlements. Bank for International Settlements, Note to Holders of BIS Shares (2000), https://www.bis.org/about/notesholdg.pdf.↩︎

14. Bundesgesetz über die Schweizerische Nationalbank (NBG), 3 October 2003, in force 1 May 2004. Art. 32. https://www.fedlex.admin.ch/eli/cc/2004/221/de↩︎

15. The count of ~56 nationalization events is the author’s estimate covering central banks that at some point had private shareholders and were subsequently nationalized or had shares mandatorily repurchased. Central banks founded as state-owned institutions (Soviet State Bank, Central Bank of China, Cuba’s central bank) are excluded from both numerator and denominator. The ~15 average number of privately held central banks at any given time is a rough estimate; the true time-average was likely higher before the 1940s wave of nationalizations and lower after. The ~1% hazard rate is an unconditional population average across all private central banks over the full period; it is not a SNB-specific estimate. The SNB’s structural features — ~75% cantonal and cantonal-bank ownership, quasi-constitutional governance under the Nationalbankgesetz, and 118 years of uninterrupted operation — plausibly place it below the population mean. Conversely, a bank that has survived 118 years under a 1% annual hazard has roughly a 30% prior survival probability, so the SNB’s longevity is statistically consistent with the model rather than evidence against it. The ~1% figure should be read as an order-of-magnitude prior, not a precise actuarial estimate of the SNB’s specific risk. No published source provides a comprehensive global count; the closest references are J. Rossouw and A. Breytenbach, “One hundred years of private shareholding in the South African Reserve Bank,” South African Journal of Economic History, 2021, and J. Singleton, Central Banking in the Twentieth Century, Cambridge University Press, 2011.↩︎

16. Return regressions at multi-year horizons use overlapping monthly observations, which creates artificial serial correlation and understates standard errors. All return-regression p-values are corrected using Newey-West standard errors with a lag equal to the window length (24 months and 120 months respectively), which accounts for the overlap. The effective independent sample sizes are approximately 56 (2-year) and 10 (10-year).↩︎

17. Historical Statistics of Switzerland (HSSO), Table O.18c: “Policy rates of SNB, bank interest rates, and yield on Confederation bonds, 1905-1992.” https://hsso.ch/en/2012/o/18b. Yield data originally from: Statistisches Handbuch des schweizerischen Geld- und Kapitalmarktes (1944).↩︎

18. Ernst Baltensperger and Peter Kugler, “The historical origins of the safe haven status of the Swiss franc,” Aussenwirtschaft 67(2), 2016. Argues that Switzerland’s structural low-yield “interest rate island” emerged only after WWI; before 1914 Swiss rates were higher than in France, the Netherlands, or the UK.↩︎

19. Federal Reserve Bulletin, “Gold Reserves of Principal Countries, 1913-1925.” FRASER (Federal Reserve Archival System for Economic Research). https://fraser.stlouisfed.org/files/docs/publications/FRB/pages/1925-1929/26253_1925-1929.pdf↩︎

20. Bruno S. Frey and Marcel Kucher, “World War II as Reflected on Capital Markets,” Economics Letters, 2000. The paper studies prices of government bonds (Germany, Austria, France, Belgium, and Switzerland) traded on the Zurich bourse from the early 1930s to 1946, showing that war-risk events were consistently and quickly priced into these assets. https://www.bsfrey.ch/wp-content/uploads/2021/08/world-war-ii-as-reflected-on-capital-markets.pdf↩︎

21. SNB, Geld- und währungspolitische Chronik. https://www.snb.ch/en/the-snb/organisation/history/geld-und-waehrungspolitische-chronik↩︎

22. SWI Swissinfo, When the franc had to learn to float, 2023. https://www.swissinfo.ch/eng/politics/when-the-franc-had-to-learn-to-float/48301476↩︎

23. Jean-Pierre Roth (SNB), Demonetisation of gold in Switzerland, BIS Review, June 1999. https://www.bis.org/review/r990622b.pdf↩︎

24. NZZ, Auch bei der Nationalbank-Aktie besteht ein Blasenrisiko, 8 September 2017. https://www.nzz.ch/finanzen/starker-kursanstieg-auch-bei-der-nationalbank-aktie-besteht-ein-blasenrisiko-ld.1315013↩︎

25. Jean Studer (SNB Bank Council President), Spotlight on SNB profits and shares, SNB AGM speech, 27 April 2018. https://www.snb.ch/en/publications/communication/speeches/2018/ref_20180427_stj↩︎

26. Theo Siegert and Bruno Weber, Fünfzig Rappen für einen Franken, NZZ, 7 April 2001. https://www.nzz.ch/article77IEO-1.485138↩︎

27. Hague Arbitral Tribunal, Reineccius, First Eagle and Mathieu v. Bank for International Settlements, Partial Award 22 November 2002 and Final Award 19 September 2003. BIS press releases: https://www.bis.org/press/p021125.htm and https://www.bis.org/press/p030922.htm. The BIS extended the additional payment to all former private shareholders, not only the claimants.↩︎

28. SNB Annual Financial Statements, “Principal shareholders” table, 2008-2024. https://www.snb.ch/en/publications/core-publications/annual-report↩︎

29. Bloomberg, SNB’s Elusive German Investor May Be Clue to Its Big Share Gain, 22 March 2018. https://www.bloomberg.com/news/articles/2018-03-22/snb-s-elusive-german-investor-may-be-clue-to-its-big-share-gain↩︎

◑