Economic Scenario Generation for Nigerian Pension Funds: Challenges and Considerations

Nigerian pension administrators face the same core questions as their counterparts elsewhere: given a mix of exposures on the asset and liability sides of the balance sheet, how likely are funding objectives to be met, and if they are not, how severe could the shortfall be? From a modelling point of view, the challenge is to answer these questions despite the limitations imposed by a short history, through the application of sound judgement where warranted.

Local currency discount curves

A naira-denominated discount curve can be constructed from the yields on Nigerian Treasury Bills and Federal Government of Nigeria (FGN) bonds. A liquid market exists for these securities, with Treasury bills issued at standard tenors of 91, 182, and 364 days, and FGN bonds spanning maturities from 1 to 30 years.

The short end and mid-curve maturities are fairly liquid and produce consistent yield data, but liquidity diminishes significantly at the long end, making it considerably less informative. This is reflected in the discrepancy between long-end yields reported by different data providers, itself a sign of the noise inherent in this part of the curve. For long-term liabilities backed by short- to medium-term assets, this is particularly problematic. Curve construction therefore requires discretion, in particular, in the choice and specification of an ultimate forward rate and how quickly convergence to it occurs.

Inflation without a linker market

The inflation data is volatile and subject to shifts in regime. Data prior to 2017 shows headline year-over-year CPI barely touching 14%. More recent periods show markedly elevated rates on average, though this masks a within-period decline: the CBN's restrictive monetary stance has gradually brought inflation down to around 15% by mid-2026, from a materially higher starting point.

A flat inflation assumption is therefore hard to justify. Without a deep market for inflation-linked securities, the joint dependence between inflation, the discount curve, and the foreign exchange rate is also difficult to capture directly. A core function of the scenario set is to keep these three jointly behaved.

Equities and property

For equities, the NGX All-Share Index (or large-cap alternatives such as the NSE 30) provides a reasonably long daily history from which to construct a total-return factor. Data for private equity portfolios is more limited, so their risk characteristics must instead be inferred from observable listed-market exposures.

Investment property is more difficult to model, as market-based proxies are limited. A basket of listed REIT or property-fund prices on overlapping dates provides a defensible proxy, although thin trading in these instruments reduces the reliability of the observed data. These limitations, and the resulting modelling judgement, should be clearly documented.

Credit

FMDQ has published daily quotations for corporate and state bonds since 2013. The available history is therefore useful, although the depth and liquidity of these markets remain limited. Credit exposure consequently requires proxies based on the more actively traded instruments, with credit spreads specified relative to FGN securities and the resulting assumptions clearly documented.

Short joint history demands more structure

Once rates, credit, inflation, equity, and property are aligned on common dates, the usable joint sample is short: a handful of recent years rather than decades. This makes a purely historical approach unreliable; there is simply not enough history to represent the range of economic conditions that a long-term pension portfolio may experience. The challenge is therefore to supplement the available history with economic structure, sound judgement, and explicit assumptions, while preserving the relationships between the different markets. This is the approach taken by GALM: generating forward-looking scenarios that remain internally consistent even where the historical record is limited.