RDP 2022-02: The Yield and Market Function Effects of the Reserve Bank of Australia's Bond Purchases 3. Event Study of Announcement Effects

As noted, the literature tends to find that in normal markets most of the effect on yields from credible and predictable central bank bond purchase programs occurs when expectations are formed, rather than when purchases are made. For bond purchases that are less credible or predictable, and/or conducted during times of stress and aimed at restoring bond market function, the evidence is more mixed, with a substantial implementation effect from the act of purchasing bonds evident, and a weaker announcement effect as strained markets may fail to immediately incorporate news into bond prices.^[9]

This suggests that an event study of the Reserve Bank's bond purchase program – where key dates in the lead-up to and announcement of the program are identified and the yield change that occurs on those dates is assessed – is one way to measure the overall effect of that program. For purchases to support market function and the yield target, it is likely that the announcement effect will only capture part of the overall effect. This is because bond markets were not functioning well when those policies were announced, the quantum of purchases was not readily predictable in advance, and for the yield target perhaps also because of some degree of scepticism among market participants around the Reserve Bank's level of commitment to the program. In this section we consider each policy announcement in turn.

3.1 Purchases to support market function

At midday on 16 March 2020, amid a serious deterioration in financial market function, the Reserve Bank released a statement stating that the Bank ‘stands ready to purchase Australian government bonds in the secondary market to support the smooth functioning of that market’ (RBA 2020a), while on 19 March 2020 at 2.30 pm following an out-of-cycle Board meeting, the Reserve Bank announced a package of policy measures, including its intention to purchase ‘Government bonds and semi-government securities across the yield curve … to address market dislocations’ (RBA 2020b); purchases began the following day.

To assess the effect on AGS yields stemming from these policy announcements, we sum the change in yields over the Australian trading day (8.30 am until 4.30 pm) on 16 and 19 March 2020. As well as examining the change in AGS yields themselves, we also examine the change in the spreads of AGS yields to term-matched overnight indexed swap (OIS) rates.^[10] The cash rate target was reduced on 19 March 2020, and a number of other policy measures were announced, all of which were likely to have had an effect on government bond yields. OIS rates provide a measure of market expectations for the evolution of the cash rate, and so can be used as at least a partial control to isolate the effect of the announcement of bond purchases to restore market function, separate from the effects of other policy measures or news released at the same time.^[11] For semis, we measure the announcement effect as the change in the spreads of semis yields to AGS yields.

For AGS, we find that the announcement effect varied by term to maturity, with the yields on shorter maturity bonds falling by around 10 basis points in absolute terms (although they would also have been affected by the reduction in the cash rate target and the 3-year yield target, also announced on 19 March 2020), while yields on longer maturity bonds rose by around 10 basis points (Figure 1). Relative to OIS rates, the announcement effect on AGS yields was more pronounced at both ends of the yield curve, with spreads to OIS for the shortest maturity bonds falling by around 20 basis points, while spreads for the longest maturity bonds rose by 20 basis points or more (Figure 2).

Sources: Authors' calculations; Bloomberg

Sources: Authors' calculations; Bloomberg

For semis, spreads to AGS were little changed on average; that is, semis yields broadly tracked AGS yields over the two policy announcement days (Figure 3).

Note: Simple average for the states, Tasmania and the two territories are excluded.

Sources: Authors' calculations; Bloomberg

Taking a longer event window from the open on 16 March 2020 until the close on 20 March 2020 does not change the results for AGS yields or AGS spreads to OIS rates at the short end, although the increases at the long end were larger; for semis, we find that spreads to AGS fell by 5–10 basis points across the yield curve. Conversely, taking a shorter window of just the day of 19 March 2020, the fall in shorter-dated AGS yields was slightly larger, and spreads to OIS for longer-dated AGS did not widen by as much as when using a window of both of the policy announcement days; for semis, the results were little changed.

The lack of a beneficial announcement effect for longer-dated AGS yields and semis spreads is perhaps related to uncertainty by market participants regarding which bonds the Reserve Bank would purchase – while the Reserve Bank's 19 March 2020 statement noted that purchases to support market function would be of bonds across the yield curve, that text came under the rubric of ‘A target for the yield on 3-year Australian Government bonds of around 0.25 per cent’, and some market participants may have assumed that purchases would be concentrated in shorter-dated bonds.^[12] Further, the distressed nature of financial markets and strong desire of market participants to raise cash by selling bonds meant that announcements by themselves would not be enough to unclog bond dealers' balance sheets. Rather, that would require actual purchases by the Reserve Bank (Finlay, Seibold and Xiang 2020). In any case, these purchases were not designed to reduce yields, but to restore good market function, which they achieved, as discussed in Section 7 below.

3.2 The yield target

At 2.30 pm on 19 March 2020 the Reserve Bank announced the introduction of a target for the 3-year AGS yield of around 25 basis points, and at the same time announced a reduction in the cash rate target from 50 basis points to 25 basis points. To assess the announcement effect associated with the yield target we consider the change in the yield on the April 2023 AGS – the 3-year AGS at the time – over a 30-minute window following the announcement, as well over the entire day of the announcement. Note that the 3-year yield target provided a form of forward guidance concerning the cash rate, and so attempting to abstract from changes in cash rate expectations by considering the spread of the April 2023 AGS yield to the 3-year OIS rate would be inappropriate, as it would exclude part of the policy's intended effect.^[13]

The announcement effect was substantial, at around 25 basis points if measured over a 30-minute event window, or around 15 basis points if measured over the course of the day, although the observed effect is confounded with the reduction in the cash rate target, announced at the same time (Figure 4). However, the effect was insufficient to reduce the 3-year AGS yield all the way to the target; as the Reserve Bank began to implement bond purchases over subsequent days, the yield declined further to become consistent with the target.

Sources: Tullett Prebon; Yieldbroker

3.3 The bond purchase program

The Reserve Bank's bond purchase program was announced at 2.30 pm on 3 November 2020, but there was market speculation leading up to this date that the Bank would announce a QE program. To conduct an event study of the announcement effect we identify nine events in the two months leading up to the initial announcement. We then sum the cumulative change over those dates in AGS yields, the spreads of AGS yields to OIS rates, and the spreads of semis yields to AGS yields.

To identify events, we examine end-of-day market summary reports written by bond traders and market economists over September and October 2020, and select those days where a piece of news was widely cited as relevant to the potential for a Reserve Bank bond purchase program. In total we identify nine such events, which include speeches by Reserve Bank Governor Lowe and Deputy Governor Debelle, the October and November 2020 Reserve Bank Board announcements, three newspaper articles, and two market economist reports (Table 1). We use a one-day interval to measure the change in yields for each event – either ‘open-to-close’ for events that occurred during trading hours, or ‘previous close-to-close’ for events that occurred before the market opened (a two-day event window gives similar results, as does controlling for offshore events by measuring AGS yields as a spread to US Treasury yields).

AGS yields declined across the curve in response to the identified events, with the cumulative change largest at the 10-year point at around 30 basis points (Figure 5). Although the 10-year AGS yield declined for each of the identified events using a one-day window, for a few of the events this masks substantial intraday yield retracements, as markets digested the news associated with the event – our approach thus captures both increases and decreases in expectations for a Reserve Bank bond purchase program, at least on the dates identified. To the extent that we have correctly identified the key dates when market participants reassessed the likelihood of the Reserve Bank conducting a bond purchase program, and no other major news occurred on those dates to move yields for other reasons, this suggests that the bond purchase program led to a fall in the 10-year AGS yield of around 30 basis points.^[14]

Sources: Authors' calculations; Bloomberg

Next, we examine how the spreads of AGS yields relative to OIS rates changed over the event study days. As noted earlier, OIS rates provide a measure of market expectations for the evolution of the cash rate, and so can be used as at least a partial control for any other macroeconomic or financial market news that was unrelated to bond purchases but affected cash rate expectations. This will, however, also ‘control’ for any signalling effect of QE, which will therefore not be captured.^[15] The results of this analysis are presented in Figure 6, and show that shorter-dated OIS rates fell by a similar magnitude to AGS yields. This in turn suggests that, for shorter-dated maturities out to around five years, most of the observed fall in AGS yields was due to lower cash rate expectations (or that lower term premia on AGS yields flowed through to lower term premia in OIS rates).^[16] For bonds with residual maturity of around 10 years, the fall in the spread of AGS yields to OIS rates is very similar to the fall in actual AGS yields, at around 30 basis points. This suggests that the fall in the 10-year AGS yield was for the most part driven by falls in AGS term and liquidity premia, and most likely the former (because outside of periods of market dysfunction, liquidity premia are typically low in the AGS market).^[17]

Sources: Authors' calculations; Bloomberg

Additionally, the bond purchase program led to a larger fall in semis yields than in AGS yields, with the spreads of semis yields to AGS yields at the relevant maturities narrowing by around 5 basis points when measured over a one-day event window (Figure 7), and by around 10 basis points when measured over a two-day window.^[18] AGS yields act as the benchmark yield curve in Australia, with other fixed income securities typically priced relative to AGS yields or swap rates. If the Reserve Bank had elected to purchase only AGS as part of its bond purchase program, it is likely that semis yields would have fallen by roughly the same amount as AGS yields, leaving the spreads between semis and AGS little changed. The evidence suggests that the inclusion of semis in the program put additional downward pressure on semis yields, resulting in a narrowing in spreads.

Note: Simple average for the states, Tasmania and the two territories are excluded.

Sources: Authors' calculations; Bloomberg

While the Reserve Bank announced a $100 billion bond purchase program on 3 November 2020, many market participants were likely to have expected from the outset that further extensions to this program would be announced in time. This implies that the 30 basis point fall in longer-term AGS yields that was observed may be better explained by a larger total expected stock of purchases than the size of the initial announcement. It is hard to be precise about how large the total expected stock of purchases might have been, but it seems reasonable to think that expectations might have been in the order of $200 billion to $300 billion, and later forecasts made by market economists over the first half of 2021 tended to fall within that range. This would imply that each $10 billion (or roughly ½ percentage point of GDP) of expected purchases over the life of the bond purchase program resulted in a fall in longer-term AGS yields of around 1 to 1½ basis points.

Following the initial announcement, two further announcements by the Reserve Bank had relevance for the total expected stock of bond purchases. First, there was the announcement on 2 February 2021 of the first extension to the bond purchase program of a further $100 billion of bond purchases. Second, there was the announcement on 6 July 2021 that, from September, purchases would proceed at a reduced pace of $4 billion per month, down from $5 billion, until at least mid-November 2021. We briefly consider these two subsequent announcements in turn.

A poll by Reuters ahead of the February 2021 Board meeting found that market economists expected the Reserve Bank to announce a further QE program of around $80 billion on average, although the modal expectation was for a $100 billion extension. In the event, the Reserve Bank announced that it would purchase an additional $100 billion of bonds once the original $100 billion of bond purchases was completed in mid-April 2021. AGS yields fell around 1½ basis points in subsequent trading. If one assumes that market economist expectations are representative of wider market expectations as embedded in bond yields, then the $20 billion upward surprise to the total expected stock of bond purchases resulted in a fall in AGS yields of around 1½ basis points, or around ¾ basis points per $10 billion. This is a little lower than the earlier estimate, but given the uncertainties inherent in the estimate it is reasonably close.

On 6 July 2021, the Reserve Bank announced that from early September 2021 it would reduce the pace of bond purchases from $5 billion per week to $4 billion per week, with the new pace to be maintained until at least mid-November 2021. Around half of surveyed market economists had expected the pace of purchases to continue at $5 billion per week and around half had expected a small reduction in the pace. Market economists expected total future purchases to be in the order of $100 billion. If we apply the roughly 10 per cent downward surprise on the pace of purchases to the $100 billion figure for total future purchases, then we arrive at an estimate that the announcement led to a revision lower in the total expected stock of purchases in the order of $10 billion. The yields on AGS eligible to be purchased under the bond purchase program rose by around 2 basis points in the 30 minutes immediately following the 2.30 pm announcement, where they finished the trading day. This change of 2 basis points following a $10 billion surprise is a little larger than the earlier estimate of 1 to 1½ basis points per $10 billion, but again given the uncertainties inherent in the estimate, it is reasonably close.