This article is an unedited excerpt from my upcoming book "Inflation Breakeven Analysis." The book is essentially completed; I am just giving it a last read before passing it on to an editor.
The qualifier “central government” is important in this context. In Canada, the provinces are a major source of conventional bond issuance; they supply a good deal of long-dated paper that few people trust the private sector to supply. The U.S. municipal market is segmented because of its tax treatment, but those issuers are soaking up some of the domestic demand for duration. (Foreigners have little incentive to look at that market, other than using derivatives.) However, these sub-sovereigns are not significant sources of net supply of inflation-linked bonds.
The next qualifier to discuss is the concept of net supply. It is possible that an issuer could emit an inflation-linked bond, and trade some inflation swaps (technically, an asset swap) with dealers to hedge out the inflation risk. I believe that this was the practice for the few inflation-linked bonds issued by Canadian provinces. The supply of inflation protection by the bond is cancelled out by the inflation swap activity, as dealers would probably be forced to buy inflation-linked bonds to hedge their asset swap positions.
Such a pattern of trading probably appears unusual – why bother with swapped issuance if the inflation exposure is just being hedged out by an existing bond? Although I never worked for an investment bank, I believe that it is a straightforward result of market structure. There are dedicated inflation-linked bond funds that are unable to engage in derivatives or leverage. The asset-swapped bond provides them with credit risk exposure that they could not get if they are confined to owning Government of Canada Real Return Bonds. Therefore, this captive audience creates a pricing advantage for such deals – the province can squeeze out paper at a tighter credit spread than it could in the conventional markets.
Nevertheless, so long as the central government remains the main supplier of unhedged inflation-linked bonds, the volume of this asset-swapped activity is constrained by its issuance.
The only other obvious sources of supply are public-private partnerships that have a revenue guarantee that is linked to inflation. In such a case, inflation-linked bonds are a natural hedge for their risks. Nevertheless, the key to such deals is that governments are backstopping the indexation in the contract, and so it is not a purely private sector phenomenon. Since these public-private partnerships are taking over long-term projects for the governments involved, one could argue that they effectively substitute government issuance that would have happened in the absence of the public-private partnership.
The central government is a natural issuer of inflation-linked bonds since tax revenues are set as a percentage of nominal incomes (or sales) in the national economy. An increase in inflation – with real GDP unchanged – implies that nominal revenues would be higher. Although one can debate whether the central government needs to worry about matching revenues to expenses, it is clear that inflation-linked bonds achieve this.
The only risk to the government is these bonds increase the weight of expenditures that are indexed to inflation. Indexed expenditures create a self-reinforcing feedback loop with inflation. Nevertheless, the size of inflation-linked issuance is generally small relative to the size of GDP, so this risk is presumably limited.
If we step away from the central government, even sub-sovereigns have concerns with issuing inflation-linked bonds. For example, Canadian provinces have relatively large governments when compared to states in the United States, and the larger provinces have sophisticated issuance programmes. (For example, they have international issues that they hedge into Canadian dollars with cross-currency basis swaps.) Even so, they have tended to shy away from inflation-linked bonds. Their concern is economic. Unlike the federal government, they do not have a central bank as a subsidiary of their finance ministry. Default is a real risk. Furthermore, the geographical dispersion of Canada shows up in economic performance as well. It is relatively common for one province to be in recession while others are growing. Although an oil-producing province like Alberta is hedged against a spike in oil prices, a province like Ontario is an importer, and a rising oil price is a drag on the provincial economy. Therefore, a rise in national CPI-linked payments could hit at the worst possible time for the provincial government. (The natural fix would be to issue bonds indexed to provincial inflation rates. These would be of less interest to institutional investors with national CPI-linked liabilities, but it might be useful for retail investors, and possibly as a trading vehicle.)
The situation is worse for private sector firms (except for public-private partnerships with explicit indexation of contracts). Most firms have a very narrow focus. Although an oil-producing firm is hedged against oil price rises, it cannot do anything about the cost of tuition. If energy firms issued U.S. CPI-linked bonds at the beginning of the 1990s, they would have been crushed by the divergence of oil prices (which suffered a secular bear market) and the steady rise of inflation, which was largely in the service sector.
Some firms have a wider focus, such as grocery firms or large discount retailers. However, even these firms do not supply college education or most medical services (at least not yet). The tendency for the retail sector to drift towards conglomerates may allow such firms to issue inflation-linked debt in the future.
Otherwise, there are no natural payers of inflation. Asset managers have inflation-linked liabilities, nobody wants to have “inverse inflation protection” for their pension. Textbooks may quaintly suggest that there are “speculators” that might pay inflation in the inflation swap market, but such firms have a very limited balance sheet capacity. No derivatives dealer is going to rely on a firm with $100 million in capital as a long-term hedge on a 30-year $1 billion inflation swap. Even if the failure of the counter-party is covered by collateral, its demise opens up the dealer’s inflation exposure. Since the counter-party’s failure would most likely be linked to higher-than-expected inflation, being able to find a new counter-party for a swap of that size might be a challenge. Since the potential gains on receiving inflation dwarf that of paying inflation (the CPI can only go to zero!), the tail risk for inflation swaps all goes in the same direction. The fact that there are no intermediate cash flows in a zero-coupon swap to reset the value of the position makes the counter-party risk situation worse.
The general lack of other sources of net supply limits the growth of the inflation-linked market, and hangs over the questions about the long-term risk premium. Many institutional investors would like to match the inflation exposure of their liabilities, creating an obvious source of demand that is not completely price sensitive, whereas the net supply is largely the unilateral decision of the central government. It would not be difficult to make an a priori argument that inflation-linked bonds should be expensive because of this fundamental supply/demand argument (and I believe that I did). That said, this factor is largely dependent upon the views of government debt managers. Although I never paid much attention to central government issuance patterns for conventional bonds as a fundamental pricing factor, this does not apply to inflation-linked debt. For example, if the government pushes for fundamentally incompatible policies – regulators demanding inflation-risk matching, while the debt office cuts inflation-linked issuance – the result could be absolutely ridiculous pricing (in either direction).* Since everyone involved knows that this is the case, there are many parties making sure such incompatible policy stances do not happen.
Footnote:
* The yield curve in the United Kingdom was a horror show in the 1990s because of liability-matching regulations. That case demonstrates that we cannot completely ignore supply and demand factors, even though I generally downplay its effects.
(c) Brian Romanchuk 2018
This comment has been removed by a blog administrator.
ReplyDeleteVery helpful post.
ReplyDeleteTwo questions. First, probably obvious, but just wanted to spell it out: one implication of this is that we cannot take breakeven rates as measures of "market expectations" of future inflation, right? If an inflation-indexed bond trades at a price equal to a conventional bond with a yield two points lower, that doesn't mean the representative market participant expects two points of inflation, it means that that's what the marginal buyer thinks hedging their inflation risk is worth. Since, as you say, the markt is quit small, the bonds are presumably held by the small subset of the market that is most worried about inflation for whatever reason. So would it be safe to say that the representative bond owner, to the extent there is such a thing, expects inflation lower than that implied by the breakeven rate?
Second, do you have any sense how much of the market for these things is institutions like pensions that wnat to hendge their inflation risk, versus speculators betting on higher inflation? There's a nontrivial population of rich people convinced that runaway inflation is jsut around the corner. Presumably that's where most of the investors in those dedicated funds are drawn from.
Sorry for the late reply; for some reason, I didn’t get a notification about the comment.
Delete(As one technical point. In bond market jargon, a “point” is $1 in price, which translates into varying amounts of yield based on the duration, a “basis point” is 1/100 of 1% in yield.)
The spread between the nominal yield and corresponding “real” yield is the standard breakeven inflation (nominal - real). Pretty much my whole report revolves around the question whether the breakeven matches a “market forecast”. On the whole, I think it is a pretty good approximation. As you note, inflation-linked bond buyers are a self-selecting group. However, there are enough fixed income investors that trade them on a relative value basis that they cannot get too stupidly expensive. At present, I’d agree that they tend to overstate inflation expectations. This was not always the case; in the early days of the market, they were cheap as investors were skeptical about them.
I never worked for an investment dealer, so I am only guessing about trading patterns. But I think that the bulk of *holdings* are buy-and-hold by institutional investors. In Canada, pension and insurance funds do a lot of liability matching; this was less true in the U.S. Inflation-linked funds are the other major holder. However, trading volumes were probably dominated by “fast money” and relative value trading by standard bond funds.
Although rich individuals worried about inflation may be a factor, they probable have very little influence on pricing relative to imstitutional fixed income investors. (They don’t move their holdings around fast enough to match the flows that institutions can produce.) This is different than something like the US municipal market, which is much more “retail” focussed.