Our recent Policy Brief focuses on the benefits of growth-indexed bonds for a country's solvency, by reporting the evolution of the debt-to-GDP ratio. In a world of perfect markets, only solvency would matter, but in practice, the time pattern of cash flows is likely to be crucial, for two reasons. First, debt crises are often triggered by refinancing difficulties. Second, smoother cash flows are likely to give more room for governments to use countercyclical fiscal policy. These considerations raise several interesting issues for designing growth-indexed bonds, particularly in an environment of low nominal interest rates.
Consider first the tradeoffs involved in choosing the maturity of growth-indexed bonds when only the principal is indexed (the coupons may be zero or unindexed), so that all the relief in terms of lower payments when growth is weak comes at the end of the contract. In that case, from the standpoint of smoothing net cash flow, shorter maturities are clearly preferable to longer maturities. From the standpoint of solvency, however, a shorter maturity—say, one year—may not provide as much insurance as a longer maturity. If growth declines unexpectedly, a one-year bond with indexed principal will provide relief at the end of the year. But if it becomes clear that the decline reflects a prolonged recession, growth expectations will soon be adjusted downward, causing investors to request a higher return (in expected terms) for the following year, thus reducing the insurance benefits to the government. Instead, a longer-maturity bond—by setting the expected growth rate for the whole multiyear period at the beginning of the contract—will better protect the government from the standpoint of solvency (the debt-to-GDP ratio at the end of the multiyear period) in the event of a prolonged recession.
This problem seems to have a simple solution, namely making the coupons dependent on the growth rate, so that low growth leads to lower interest payments in the same year. But the current environment of low nominal rates poses another difficulty. Suppose that bonds are issued at par, and coupons are equal to ri = k+g, where ri is the contractually established coupon on growth-indexed bonds, k is a constant, and g is the actual growth rate (which will determine the actual coupon payment). If the corresponding nominal rate on ordinary bonds is low, riwill also be low in expected value. For countries where growth is volatile, k+g may well turn negative. As it is unrealistic to expect investors to make net payments to the government during the life of the contract, this limits the cash flow relief that the government will be able to receive from investors during years when growth is especially weak.
A possible fix is to allow the government to "accumulate a credit" vis-à-vis investors for the negative portion of k+g, and to pay a correspondingly lower principal at the end of the contract, but this will delay receiving that portion of cash flow relief. Another potential solution (probably my favorite) may be to increase the share of repayment occurring through the nonindexed portion of the coupons (that is, to increase k) and correspondingly reduce the value of the principal repayment. A higher k reduces the likelihood of k+g hitting the "zero lower bound," thus ensuring that cash flow relief is likely to be obtained in a timely manner even for severe adverse shocks to growth against the current background of low global interest rates. For a given amount borrowed, boosting k while meeting market conditions entails a lower principal repayment in the contract design. Equivalently, the bond would have to sell above par. The tradeoff in this case is then between preserving the cash flow benefits of indexation during the life of the contract and accepting an above-par issuance price as well as shorter bond duration.