John Bayliss writes about his secondment from the Government Actuary’s Department to UK Government Investments as head of modelling – particularly in relation to the government’s student loan portfolio
A relatively little-known organisation, UK Government Investments (UKGI) operates as a government company that is wholly owned by HM Treasury. It comprises around 120 people, including a number of direct employees – many of whom have investment banking experience – and some secondees from the private sector and civil service. Its main role is to provide expertise in corporate finance and corporate governance to the UK government, working with arm’s length bodies such as Channel 4 and the Post Office. It also helps the government with asset sales such as Green Investment Bank and Lloyds Bank.
Student loan sales
One of the government’s assets is the portfolio of income-contingent student loans. ‘Income-contingent’ means that borrowers only make a loan payment if they earn above a threshold. The cash flows from selected cohorts of these are sold to investors as a multi-tranche securitisation, meaning that the loans are packaged together and their related cash flows are sold to investors as bond-like products referred to as notes, with different levels of seniority.
There have been two sales, and the intention was to sell up to six further tranches. The chancellor has discontinued sales in his latest Budget, citing the broad stability of public sector debt pre-COVID-19 as the reason. Each completed sale has comprised one unrated and three rated notes, with a range of specific characteristics and tenors. These appeal to different types of investors – from banks and traditional fund managers, to life insurers and pension schemes, to hedge funds.
“Unlike with traditional debt, the borrower cannot default, so there can potentially be a period of many years of non-payment before payments start again”
Why do we need a model?
This is a new and unique capital market product. There have been student loans sales before, both in the UK and abroad, but these were predominantly mortgage-style loans. The income-contingent nature of the current student loans portfolio is the crucial differentiating aspect. No one has monetised this type of profile before, so a specialised approach was required.
A key part was understanding and assigning a value on the cash flows expected to emerge. Because of this, the government took the unusual step of developing a model to assess cash flows and releasing this to potential investors to help them estimate and stress test their expected cash flows. This clearly puts pressure on the team to make sure the model is clearly laid out and extremely well documented.
Our approach is to provide a base case projection via a secure portal, and then provide tools and instructions to help investors amend key assumptions. Our formal documentation also contains a range of scenarios that help investors assess their risk.
Is the model complex?
In its essence, it is a simple model that projects earnings, works out if a borrower has earnings above the threshold and, if so, generates a cash flow, which itself is paid to investors in accordance with the capital structure. First, each borrower’s earnings are projected year by year. Next, the earnings are converted into repayments based on each loan’s terms and the amount outstanding. Finally, the payments to investors are determined from the total projected collections and the package loan structure.
Unlike with traditional debt, the borrower cannot default, so there can potentially be a period of many years of non-payment before payments start again when earnings go above the threshold. When a projected borrower reaches the age at which they are no longer liable to repay, their loan is extinguished.
In practice, there are some complexities. Firstly, and most importantly, we allow borrowers to switch between earnings bands, and from inactivity (not earning) to activity (earning) and vice versa (see Figure 1). While individual paths are random, collectively borrowers can be expected to follow a distribution of paths that we are simulating. Our assumption for this distribution is driven by the data we have. Further complexity arises because of borrowers having combinations of PAYE and self-assessed earnings, moving overseas and making voluntary repayments.
Even though the sale programme is cancelled, the model remains a vital component of the servicing of investors, and will remain so until the last loans in the sold cohorts are extinguished.
A significant part of the job is managing the myriad stakeholders. These include the Department for Education, the Treasury, our data providers, the Student Loans Company, HMRC, auditors, rating agencies, sale advisers and investors. The model is audited to give investors additional confidence, and we liaise with rating agencies to understand their stress tests and help them understand the nuances of the model.
We work closely with the transaction team that presents the business case for the sale, decides the capital structure, determines the market capacity and sentiment, and ensures (with the help of our model) that the government’s value for money requirements are met.
We also have various governance processes to manage. In particular, the model, being data-driven, is continually evolving to improve accuracy and reliability, so we need to go through formal quality assurance and sign-off processes when proposing a change.
Other stakeholders include the Office for Budget Responsibility, National Audit Office and Government Internal Audit Agency.
Although the sale programme has been discontinued, we have a legal responsibility to provide annual updates to investors that have already invested in previous sales, in the form of a report for each prior sale. This happens in July each year and requires a significant amount of planning and preparation, including auditing the model. In between these primary deliverables, we need to liaise with data providers, set assumptions and do model development work.
For the completed sales, the audited and rated model had to be ready to share at least two months in advance of the final sale date (so far this has been early December), and the team did an investor roadshow, culminating in a ‘book build’ pricing process. The first two sales raised £3.6bn.
The team and relationship with GAD
There are 10-12 people in the modelling team, comprising civil servant analysts (statisticians, operational researchers and data scientists), actuaries and a project manager.
GAD gives ongoing support and advice to UKGI throughout the sale process on the appropriate use of the model and the significance of the assumptions and judgments on which it is built. It also helps UKGI to assess the ‘value for money’ of the sale.
GAD’s working partnership with UKGI has evolved and strengthened through this support, and through secondments of GAD staff to UKGI. Secondments at both junior analyst and senior modelling team lead levels have helped to embed the model within UKGI, and to transfer knowledge – as well as providing development opportunities for staff at both organisations.
The work has been interesting and variable, as well as intensive at times. Our work feels important, and is subject to a great deal of scrutiny. We enjoy the challenges – and we try to have fun.
John Bayliss is a senior actuary at the Government Actuary’s Department