Calebe de Roure, Ben Morley and Lena Boneva
In August 2016 the MPC announced a package of easing measures, including the Corporate Bond Purchase Scheme (CBPS). In a recent staff working paper, we explore the announcement impact of the CBPS, using the so called “difference in differences” (or “DID”) approach. Overall – to deliver the punchline to eager readers – this analytical technique suggests that the announcement caused spreads on CBPS eligible bonds to tighten by 13bps, compared with comparable euro or dollar denominated bonds (Charts 1b, 2).
Chart 1a: Yield spreads of corporate bonds eligible for the CBPS and compared to ineligible investment-grade Sterling corporate bonds
Chart 1b: Yield spreads of corporate bonds eligible for the CBPS and compared to USD bonds of eligible issuers
The DID approach explores how the impact of a certain event varies between a ‘treatment group’ and a ‘control group’. By comparing the impact on one group against the other, this approach should strip out the impact of other, exogenous/external, factors driving both groups. This means establishing the correct treatment and control groups is crucial to this analysis. Some more details on the scheme will help support how we defined these groups.
The Bank announced the scheme on 4th August 2016, but the purchases didn’t begin until late September. On the initial announcement, guidance was provided stating that only sterling investment grade non-financial corporate bonds would be purchased. And of those, only bonds issued by companies who make a “material contribution” to the UK economy would be eligible for purchase. This leaves two subsets of sterling investment grade non-financial corporate bonds: those deemed to make a material contribution (eligible), and those deemed not to (ineligible). Although the list of eligible bonds was not published until a few weeks before purchases began, overall the guidance was sufficient to approximate which subset individual bonds would fall into.
Given the sharp move in spreads in response to the scheme, it appears the announcement was somewhat of a surprise. The announcement impact of the CBPS on eligible corporate bonds spreads can be estimated, by setting it as our DID treatment group and using ineligible investment-grade sterling bonds as our control group. This approach reveals spreads on eligible corporate bonds fell by 2bps more than spreads on ineligible bonds (Chart 2). But this figure is quite small, and less than the headline fall in bond spreads. Why? Well, the devil is in the detail of the theoretical channels of asset purchases, outlined in more detail in this earlier Quarterly Bulletin article. Outlined briefly, the announcement of additional demand for sterling corporate bonds should drive prices higher (yields and spreads lower, as they operate as the inverse to bond prices). Higher prices mean some investors would become priced out of these bonds, and instead seek alternative assets. The closest substitute is likely to be the set of ineligible bonds. Investments in these assets should in turn drive their price up, creating a knock on effect into other asset classes, such as high yield bonds and equities. This process is referred to as ‘portfolio rebalancing’.
With portfolio rebalancing in mind, an issue with this 2bps figure may be obvious. Comparing the announcement effect on eligible bonds using sterling ineligible bonds as a control group isn’t a fair test. As already outlined, prices of ineligible investment-grade sterling bonds may have been impacted as a result of portfolio rebalancing. Indeed, Chart 1a shows that the spreads of ineligible sterling bonds fall significantly after the announcement. So there must be a better comparison to act as our counterfactual, not skewed by portfolio rebalancing.
Our next approach was to take the list of corporates who had bonds on the eligible list, and construct a control group of bonds issued by those same corporates in other currencies. These bonds should be just as risky and have similar features, which makes them a much more natural control group. Additionally, investors with a mandate to invest in sterling bonds are not likely to rebalance their portfolio into other currencies, so the impact of the CBPS on foreign bonds should be minimal. Indeed, spreads of USD bonds issued by eligible corporates did not react to the announcement (Chart 1a). Using the same DID approach against the control group of 944 US dollar, and separately 666 euro denominated corporate bonds, indicates a much stronger CBPS impact. This approach suggests eligible sterling corporate bond yields fell by around 13bps compared to each of the control groups (Chart 2). For robustness, we repeated this analysis for different days around the policy announcement (ranging from 5 days before to 5 days after the actual announcement), and found the extraordinary moves in sterling corporate bonds were limited to the announcement day itself.
Chart 2: Change in yield spreads of eligible bonds around the announcement date
This move is large in itself; a 13bp move (in either direction) is in the 99.9th percentile for daily changes in the BAML sterling non-financial corporate bond index over the last 20 years (a larger one-day move has only occurred twice). Additionally, this figure may actually underestimate the impacts, as it’s possible that corporates with debt in sterling as well as euros and/or dollar would benefit from the scheme as a whole. Lower sterling borrowing costs might reduce credit risk and therefore spreads on their non-sterling denominated bonds (compared to a world where the CBPS didn’t exist). If this is true, then the 13bps figure may in fact be an estimation of the lower boundary of the CBPS announcement effect.
Our estimated impact of the CBPS announcement is slightly larger compared to the 10bps figure in a recent Quarterly Bulletin article by Belsham, Rattan and Maher. This can be attributed to differences in methodology: Belsham et al. use an event study methodology to measure the fall in the BAML sterling non-financial corporate bond index on the announcement day itself. In contrast, our DID methodology measures the fall in credit spreads of eligible bonds compared to a control group and over a 2-day window around the announcement day.
In summary, our results suggest that the CBPS announcement decreased the spreads of sterling denominated bonds eligible for the scheme by around 13bps compared with changes in a comparable set of euro and dollar denominated corporate bonds, issued by the same corporates.
Calebe de Roure works at the Reserve Bank of Australia, Lena Boneva works in the Bank’s Sterling Markets Division and this post was written whilst Ben Morley was working in the Bank’s Sterling Markets Division.
If you want to get in touch, please email us at firstname.lastname@example.org or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied.
Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.