Sovereign debt levels have risen steadily in recent years, prompting discussions on how to build greater resilience into public balance sheets. While advanced economies are not currently contemplating major structural overhauls to their debt instruments despite high borrowing, the situation has sparked notable policy interest regarding emerging market economies (EMEs). For many emerging markets, acute vulnerability to global economic downturns and external shocks has renewed debates on how sovereign debt contracts could be designed to better absorb shocks.

To lower the likelihood of future debt crises, economists and policymakers have frequently discussed the potential of state-contingent debt instruments (SCDIs), such as GDP-linked bonds. As outlined by Borensztein and Mauro (2004), instruments whose coupon payments or principal are tied to macroeconomic performance offer a built-in insurance mechanism for the issuing economy. When it enters a recession, its debt-service obligations automatically shrink, transferring the risk from borrower to lender, lowering the probability of default, and freeing up fiscal space for countercyclical policy. Writing on VoxEU, Cecchetti and Schoenholtz (2017) provide a primer on how these bonds can smooth economic shocks, while Blanchard et al. (2016) advocate for the introduction of growth-indexed bonds to stave off costly and disruptive debt restructurings.

One standard objection is that investors do not absorb that risk for free, and the resulting premium required to hold this debt might be so large that it wipes out any benefits. But how large is this premium? Until now, direct empirical evidence from normal-times markets has been thin. Estimates from calibrated models or from bonds issued during restructurings are imperfect. Models of risk premia are very sensitive to assumptions (Benford et al. 2016). Furthermore, empirical estimates from debt restructurings (Igan et al. 2021) may be biased either by the use of contingent claims as a ‘sweetener’ to reluctant creditors or by the fact that holders in those settings (often distressed-debt funds) are not representative of normal-times pricing. What has been missing is a clean estimate from a deep capital market in normal times. 

A forgotten experiment in 1956 France

In a recent paper (Mitchener and Pina 2026), we overcome this measurement challenge by turning to a historical episode that allows us to estimate the state-contingent debt premium directly. In April and May 1956, the French Fourth Republic issued two government bonds with virtually identical terms. The Bons PTT (BPTT), issued in April, paid a fixed 5.5% coupon. The Bons d’équipement industriel et agricole (BEIA), issued a month later, paid a minimum 5% coupon plus a bonus indexed to the official industrial production index published by the national statistical agency (INSEE). This bond is the earliest known issuance of a sovereign bond directly linked to a measure of national output. Paul Ramadier, then France’s Minister of Finance and Economic Affairs, aptly described the holders of this bond as “shareholders whose dividend varies according to the national income”. 

Crucially for our empirical strategy, these two bonds were nearly identical in all other respects. Both bonds had the same 15-year maturity, identical face and redemption value, identical lottery-redemption provisions, and identical tax treatments. Both also traded actively on the same Paris secondary markets. This quasi-twin bond setting is the cleanest available laboratory for measuring the state-contingent debt premium during normal times in a well-established debt market.

To conduct our analysis, we hand-collected daily secondary-market prices between 1956 and 1971 for both bonds using the Cote Desfossés. Figure 1 shows that both bonds open near par at 100. Strong industrial growth then drives the price of the indexed BEIA bond well above the conventional BPTT, with the gap widening through the early 1960s, narrowing as both bonds approach redemption, and reopening sharply around the May 1968 general strike that disrupted French industry.

Figure 1 Bond prices in New France

Notes: The figure compares monthly secondary-market prices for two French bonds: the BEIA, a state-contingent bond linked to industrial production, and the BPTT, a conventional bond. Prices are shown as ‘dirty prices’, including accrued interest.

A modest premium, except in a crisis

Translating these prices into yields and taking the state-contingent premium as the difference between the yields of the two bonds delivers the headline result. Figure 2 plots the results for realised values of industrial production while Figure 3 uses contemporaneous forecasts from the Ministry of Finance and the French General Planning Commission. The expected premium at issuance was 77 basis points. Strong postwar growth meant the realised premium was almost twice as large, at 146 basis points. As the French economy continued to outperform forecasts, the spread decreased steadily, reaching roughly zero by 1964. Then, in May-June 1968, an unexpected general strike pushed industrial production sharply below trend; the spread climbed close to 400 basis points before falling back near zero by 1970.

Figure 2 The state-contingent debt premium

Notes: The figure plots the state-contingent debt premium, computed as the yield of the BEIA bond minus the yield of the BPTT bond, measured in basis points, for the realised values of industrial production.

Figure 3 Expected levels of industrial production

Notes: The dashed lines show different waves of expected levels of industrial production while the solid line shows realised industrial production. The numbers represent the expected state-contingent debt premium measured in basis points at different times: 1956 (issuance), 1958, 1962, and 1965. Next to each premium is the source for expected levels of industrial production as explained in the paper.

Three observations stand out. First, even looking ex post, the average realised premium over the bond’s lifetime is 108 basis points. This falls well within the range obtained from theoretical calibrations (35-150 basis points; Benford et al. 2016) and is an order of magnitude smaller than realised premia for the GDP-linked instruments issued in recent restructurings: roughly 1,250 basis points for Argentina, 425 for Greece, and 665 for Ukraine (Igan et al. 2021). The expected premium, computed using the contemporaneous forecasts of the Third, Fourth, and Fifth Plans, is even smaller, ranging from 29 to 77 basis points. Either way, the cost of issuing this state-contingent bond was lower than for modern counterparts.

Second, the spread moves dynamically with the state of the economy. In good times, when industrial production outperformed expectations, the spread decreased, reaching roughly zero by 1964. However, the insurance mechanism faced a severe stress test in May-June 1968 when an unexpected general strike pushed industrial production below trend. In response, the spread climbed to almost 400 basis points before falling back to near zero by 1970. This reaction is entirely consistent with standard finance theory: risk-averse investors demand a higher premium for holding an asset that fails to pay out precisely when times are bad. 

Third, institutional credibility and investor familiarity appear to have been important. Indexed debt was common at the time in France. Roughly half of all bonds issued by the French state, by state-owned enterprises, and by private firms in 1956 carried some form of indexation (Rozental 1959). Moreover, INSEE was highly regarded as an independent statistical agency, ensuring investors that the underlying macroeconomic data could not be manipulated for debt relief. 

Implications for today’s debate

These results inform the ongoing debate surrounding state-contingent debt instruments (Anthony et al. 2017, Abbas et al. 2017). A primary concern among policymakers is that linking sovereign debt to output may be prohibitively expensive in normal times. Our estimates suggest that when the indexing variable is credibly measured and investors are familiar with contingent contracts, the cost can be smaller than recent estimates. How, then, should we interpret the very large premia observed on Argentine, Greek, and Ukrainian instruments? Those cases provide evidence on state-contingent debt instrument pricing in distressed restructuring settings. Our estimates add a normal-times benchmark for interpreting those premia.

The French episode fills an important gap in the empirical literature because it combines a deep market, a credible statistical agency, and identical ‘twin’ bonds that make it possible to read the state-contingent premium directly off prices rather than infer it from models. Our findings suggest that the cost of state-contingent debt may be considerably lower than the recent restructuring record has led policymakers to believe. But timing matters: issuing state-contingent debt in normal times, when markets are deep and statistical institutions are credible, may look very different from issuing it in the middle of a debt restructuring.

References

Abbas, S M A, D Hardy, J Kim, and A Pienkowski (2017), “State-contingent debt instruments for sovereigns: A balanced view”, VoxEU.org, 6 June.

Anthony, M L, N Balta, T Best, S Nadeem and E Togo (2017), “What history tells us about state-contingent debt instruments”, VoxEU.org, 6 June. 

Benford, J, M Joy, M Kruger and T Best (2016), “Sovereign GDP-linked bonds”, Bank of England Financial Stability Paper 39.

Blanchard, O, P Mauro and J Acalin (2016), “The Case for Growth Indexed Bonds in Advanced Economies Today”, Peterson Institute for International Economics, Policy Brief no. 16–2.

Borensztein, E and P Mauro (2004), “The case for GDP-indexed bonds”, Economic Policy 19(38): 165–216.

Cecchetti, S and K L Schoenholtz (2017), “GDP-linked bonds: A primer”, VoxEU.org, 1 March.

Igan, D O, T Kim and A Levy (2021), “The premia on state-contingent sovereign debt instruments”, IMF Working Papers 2021.282.

Mitchener, K and G Pina (2026), “The State-Contingent Debt Premium: Evidence from French Public Bonds”, CEPR Discussion Paper No. 21425.

Rozental, A A (1959), “Variable-return bonds–The French experience”, The Journal of Finance 14(4): 520–530.