Archived - Appendix B - Literature Review Summary 

Archived information

Archived information is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please contact us to request a format other than those available.

For a summary and discussion of recent changes to the auction process in the U.S. see Garbade and Ingber (2005), but they do not delve into the academic rationales or evidence for the changes.

For reviews of the general auction theory literature, we guide the reader to Klemperer (1999), and Wolfstetter (1996). Some of the key papers in auction theory are collected in Klemperer (2000). For more recent surveys that include applications, see Milgrom (2004) and Klemperer (2002). We caution that the reader should focus on theories of common value (or affiliated value) auctions which are appropriate for modeling goods with resale markets, such as treasury securities, rather than the more ubiquitous models of private-value auctions which are inappropriate in the context of marketable securities. The seminal paper on the theory of affiliated value auctions is Milgrom and Weber (1982). A good and intuitive starting point is Milgrom (1989).

The theoretical auction literature addresses the conditions under which auctions are most appropriate. Milgrom (2004) argues that auctions are appropriate for homogenous securities. Similarly, Habib and Ziegler (2007) develop a theory that shows that auctions are better than a posted price when there is little information to be gathered. This would be the case for treasuries, especially in the presence of a WI market which publicly reveals much of the price relevant information. In Bulow and Klemperer (1996) it is shown that auctions are usually better than negotiations (which could be taken as a model for syndication), and in those cases in which negotiations are superior, it is only by a small amount. In our interviews, Rocholl reported that in Germany auctions appear to be revenue superior to syndication. Nyborg stressed the importance of distribution and suggested that an auction could be optimal in the presence of numerous bidders and an active secondary market.

Recent literature models the auction participants as being asymmetric. Armantier and Sbai (2006) estimate such a model using French data, and show that there are significant asymmetries. Therefore, they argue that a first-price auction would be revenue superior in France. In a related paper, Armantier and Lafhel (2009) show that there are few such asymmetries in Canadian treasury auctions, and thus recommend a continuation of the multiple-price format.

One theme that repeats throughout the theoretical literature is that auction revenue increases when more value relevant information is publicly revealed in advance or in the auction itself. This is an important justification for the WI market which gives bidders a very close estimate of the secondary market value of the security at the time they bid. The importance of information revelation is central to Milgrom and Weber’s (1982) model of affiliated value auctions. (“Affiliation” means that if one bidder has a higher value of the good, then the other bidders are likely to also have higher values. It is reasonable to assume that this property holds for treasury auctions.) As a result, auction prices are higher when the winning bid incorporates more information from other bids. Milgrom and Weber refer to this as the “Linkage Principle”. A similar concept is the “Publicity Effect” in Milgrom (2004). The public revelation of information prior to the auction also reduces the incentives of participants to make (possibly wasteful) investments in information acquisition. This argument was first put forward by Friedman (1960) and developed by Chari and Weber (1992). Thus, information revelation (again, most notably through the WI market) would improve the auction outcome. Empirically, Berg, Boukai, and Landsberger (1998) use a logistic model show that WI trading affects the auction outcome regardless of the auction format.

In an analysis of Swedish treasury auctions, Nyborg, Rydqvist, and Sundaresan (2002) show that the bids (including the bid prices, the dispersion of bids, and the quantities) are affected by uncertainty at the time of the auction. This suggests that volatility in the WI market or any other source of uncertainty would have a detrimental effect on the auction. Similarly, Goldreich (2007) finds that a wider dispersion of bids leads to more underpricing in U.S. treasury auctions.

In an empirical examination of dealers’ positions around the time of auctions, Fleming and Rosenberg (2007) find that dealers leave much of their auction position unhedged. This suggests that volatile market conditions could increase risk to dealers and would thus be detrimental to the auction.

Literature Review References

Armantier, O. and N. Lafhel, 2009, Comparison of auction formats in Canadian government auctions, working paper, Bank of Canada.

Armantier, O. and E. Sbai, 2006, Estimation and comparison of treasury auction formats when bidders are asymmetric, Journal of Applied Econometrics 21, pp. 745-779.

Back, K. and J.F. Zender, 1993, Auctions of divisible goods: on the rationale for the Treasury experiment. Review of Financial Studies 6, pp. 733–764.

Bartolini, L. and C. Cottarelli, 1997a, Designing effective auctions for treasury securities, Current Issues in Economics and Finance, Federal Reserve Bank of New York, 3(9).

Bartolini, L. and C. Cottarelli, 1997b, Bartolini, Leonardo and Carlo Cottarelli, Treasury Bill Auctions: Issues and Uses, in Mario I. Blejer and Teresa Ter-Minassian, eds., Macroeconomic Dimensions of Public Finance: Essays in Honour of Vito Tanzi, London: Routledge, 1997, pp. 267-336.

Berg, S.A., Boukai, B., Landsberger, M., 1998. Bidding for treasury securities under different auction rules: the Norwegian experience. Working paper, Haifa University.

Bikhchandani, S., P. Edspar, and C Huang, 2000, The Treasury Bill Auction and the When-Issued Market: Some Evidence, Working Paper, MIT.

Bulow, J. and P. Klemperer, 1996, Auctions versus Negotiations, American Economic Review, 86, 180-194.

Cammack, E., 1991, Evidence on Bidding Strategies and the Information in Treasury Bill Auctions, Journal of Political Economy, 99(1), February, 100-130.

Chari, V., and R. Weber, 1992, How the U.S. Treasury Should Auction its Debt Quarterly Reviews, Federal Reserve Bank of Minneapolis.

Fleming, M., 1997, The Round-the-Clock Market for U.S. Treasury Securities. Federal Reserve Bank of New York Economic Policy Review, 3, pp. 9-32.

Fleming, M., 2002, Are larger treasury issues more liquid? Evidence from bill reopenings, Journal of Money Credit and Banking 34, pp. 707-735.

Fleming, M., 2007, Who buys treasury securities at auction? Current Issues in Economics and Finance 13, pp. 1-7.

Fleming, M. and J. Rosenberg, 2007, How do treasury dealers manage their positions? Working paper, Federal Reserve Bank of New York.

Friedman, M., 1960, A Program for Monetary Stability, New York: Fordham University Press.

Garbade, K., and J. Ingber, 2005, The Treasury Auction Process: Objectives, Structure, and Recent Adaptations Federal Reserve Bank of New York: Current Issues in Economics and Finance, 11, pp. 1-11.

Gordy, M., 1999, Hedging Winner's Curse with Multiple Bids: Evidence from the Portuguese Treasury Bill Auction, Review of Economics and Statistics, 448-465.

Goldreich D, 2007 Underpricing in Discriminatory and Uniform-Price Treasury Auctions, Journal of Financial and Quantitative Analysis Vol. 42, No. 2, June, 443-466.

Goldreich, D., B. Hanke, and P. Nath, 2007, The Price of Future Liquidity: Time-Varying Liquidity in the US Treasury Market Review of Finance, 9, pp 1-32.

Goswami, G., T. Noe and M. Rebello 1996, Collusion in Uniform-Price Auctions: Experimental Evidence and Implications for Treasury Auctions, Review of Financial Studies, 9, 757-785.

Habib, M. and A. Ziegler, 2007,  Why Government Bonds are Sold by Auctions and Corporate Bonds by Posted Price, Journal of Financial Intermediation 16, pp. 343-367.

Hortacsu, A., and S. Sareen, 2005. Order Flow and the Formation of Dealer Bids: Information Flows and Strategic Behavior in the Government of Canada Securities Auctions. NBER Working Paper.

Jaganathan, R., and A. Sherman, 2006, Why Do IPO Auctions Fail? working paper, Northwestern University.

Jegadeesh, N., 1993, Treasury Auction Bids and the Salomon Squeeze, The Journal of Finance, Vol. 48, No. 4, pp. 1403-1419.

Klemperer, P., 1999, Auction Theory: A Guide to the Literature, Journal of Economic Literature, 13, pp. 227-286.

Klemperer, P., 2000, The Economic Theory of Auctions, Cheltenham, UK: Edward Elgar.

Klemperer, P., 2002, What Really Matters in Auction Design, Journal of Economic Perspectives, 16, pp 169-189.

Lengwiler, Y. 1998, The multiple unit auction with variable supply, Working Paper, Federal Reserve Board of Governors.

Lu, D., and J. Yang, 2003, Auction Participation and Market Uncertainty: Evidence from Canadian Treasury Auctions, working paper, Industry Canada and Bank of Canada.

McAfee, RP and J McMillan, 1987, Auctions and Bidding, Journal of Economic Literature, Vol. XXV, pp. 699-738.

Merrick J., N. Naik, and P. Yadav, 2005, Strategic trading behavior and price distortion in a manipulated market: anatomy of a squeeze, Journal of Financial Economics, 77, pp 171-208.

Milgrom, P. 1989, Auctions and Bidding: A Primer, The Journal of Economic Perspectives, Vol. 3, No. 3, pp. 3-22

Milgrom, P., 2004, Putting Auction Theory to Work. Cambridge University Press.

Milgrom, P. and R. Weber, 1982. A Theory of Auctions and Competitive Bidding. Econometrica 50: 1089–1122.

Nyborg, K., K. Rydqvist, and S. Sundaresan, 2002, Bidder behavior in multiunit auctions – Evidence from Swedish treasury auctions, Journal of Political Economy 110, 394-424.

Nyborg, K., and S. Sundaresan, 1996, Discriminatory versus Uniform Treasury Auctions: Evidence from When-Issued Transactions, Journal of Financial Economics, Vol. 42.

Pesendorfer W. and J. Swinkels, 1997, The Loser's Curse and Information Aggregation in Common Value Auctions  Econometrica, 65, pp. 1247-1281.

Rocholl J., 2005, Discriminatory Auctions in which the Seller has Discretion, Working Paper, European School of Management and Technology.

Simon, D., 1994, Markups, Quantity Risk, and Bidding Strategies at Treasury Coupon Auctions, Journal of Financial Economics, 35, 43-62.

Tenorio R., 1993, Revenue Equivalence and Bidding Behavior in a Multi-Unit Auction Market: An Empirical Analysis, Review of Economics and Statistics, 75, 312-314.

Weber, R., 1983, Multiple Object Auctions In Auctions Bidding and Contracting: Uses and Theory, R. Englebrecht-Wiggans, M. Shubick, and R. Stark, eds. New York, NYU Press.

Wilson, R., 1979, Auctions of Shares. Quarterly Journal of Economics 93, pp. 675–698.

Wolfstetter, E., 1996, Auctions: an Introduction, Journal of Economic Surveys, 10, pp. 367-420.

Zheng, Charles Z., 2001, High Bids and Broke Winners, Journal of Economic Theory, vol. 100(1), pages 129-171.

Appendix C - Data and Market Analysis Summary 

One focal point for this evaluation is the ‘effectiveness’ of auctions. One measure of auction effectiveness is the auction Tail (for multiple-price auctions, the difference between the highest accepted yield and the average yield of the auction). Tail is generally below 1 basis point for most products covered in this report, reflecting the relatively high liquidity of Government of Canada debt obligations and indicating an overall high level of effectiveness of these auctions. However, during unstable markets the tail for nominal bonds or treasury bill auctions can increase significantly above 1 basis point. We calculated a ‘tail’ for real-return bond auctions, which are single-price auctions, using the same methodology as for multiple-price auctions. Due to the behaviour of bidders in a single-price auction, this tail calculation will naturally be higher because bidders will bid aggressively to have their order filled knowing that they will only have to pay the cut-off yield of the auction.

A second measure of auction effectiveness, the Bid-Cover ratio, defined as the ratio of total bids submitted to auction size, is usually over two times for all auctions as seen in the chart below on the right. In addition, in some sections of this report, we calculate an adjusted bid-cover ratio which is defined as the ratio of total bids, less the Bank of Canada allotment, to total auction size.

Average Auction Tail/Bid Cover

Average Auction Tail (bps)
  Avg Std Dev Min Max
3mo 0.6 0.5 0.1 4.1
6mo 0.6 0.4 0.2 3.1
1yr 0.7 0.4 0.2 2.1

2Y 0.4 0.2 0.1 0.8
5Y 0.4 0.2 0.0 0.7
10Y 0.4 0.2 0.1 0.7
30Y 0.4 0.2 0.1 0.7

RRB 2.8 1.0 1.4 5.6

 

Average Bid-Cover (times)
  Avg Std Dev Min Max
3mo 2.2 0.2 1.8 2.6
6mo 2.2 0.2 1.8 2.7
1yr 2.1 0.2 1.6 2.6

2Y 2.5 0.1 2.3 2.7
5Y 2.6 0.1 2.4 3.0
10Y 2.5 0.1 2.2 2.8
30Y 2.5 0.2 2.3 2.7

RRB 2.8 0.3 2.2 3.3

Financial market conditions varied significantly over the data period. The period begins in 2003 the last time credit spreads were wide and ends in 2008 well after the current credit crisis began in mid 2007. We measure financial market conditions using the following three factors and examine their effect on the auction Tail and Bid-Cover ratio jointly. Individual effects were also analysed but the evidence does not alter any conclusions drawn from the analysis of joint effects.

Bond yield volatility:

  • specific to the term of each auction, measured by the standard deviation of the per cent change in the yield in the previous 20 days
  • tends to decline with term-to-maturity of the bond (this is due to higher duration risk with longer term-to-maturity bonds) and increases with term-to-maturity of treasury bill

Credit conditions:

  • measured by the Baa Credit Spread

Liquidity:

  • measured by the monthly trading volume (billions of dollars, except RRBs in millions)
  • tends to decline with the term to maturity of the bond or treasury bill, RRBs have the lowest liquidity

Financial Market Conditions

Summary statistics for the measures of financial market conditions are listed in the tables below. Yield volatility was calculated for each corresponding point on the yield curve. While yield volatility increases with the term-to-maturity of treasury bills, it decreases with the term-to-maturity of bonds. This is due to the higher duration risk associated with longer term bonds, i.e. price volatility increases with maturity but yield volatility decreases. As mentioned previously, liquidity is measured as the average monthly trading volume of various instruments. Due to data limitations, total volume is combined for treasury bills since it is not reported separately by term. The table below shows that the liquidity of 30-year nominal bonds and RRBs is relatively lower than other instruments. Their longer term-to-maturity and higher duration risk means they experience a greater price change for a given change in yield. Finally, the Baa spread is not distinguished by term in this analysis since credit conditions reflect an overall market variable. Other measures of volatility and credit conditions could be used, but we suspect that results will not differ significantly for other measures. There is generally little statistical evidence of the impact of market conditions on auction characteristics.

Yield Volatility
  Avg Std Dev Min Max
3mo 0.7% 0.4% 0.1% 3.2%
6mo 0.9% 0.5% 0.2% 2.5%
1yr 1.2% 0.6% 0.3% 2.8%

2Y 1.5% 0.6% 0.5% 2.5%
5Y 1.2% 0.4% 0.5% 2.0%
10Y 0.9% 0.3% 0.5% 1.5%
30Y 0.8% 0.2% 0.6% 1.1%

RRB 0.7% 0.1% 0.5% 1.0%

 

Liquidity ($ billions)
  Avg Std Dev Min Max
3mo/6mo/1yr 24.0 3.9 16.8 33.4

2Y 38.0 6.3 26.9 49.7
5Y 51.8 12.1 35.1 83.0
10Y 50.2 9.2 35.1 69.0
30Y 10.4 1.7 7.9 13.0

RRB 0.5 0.1 0.3 0.7
Baa Spread        
All 1.5% 0.2% 1.1% 2.6%

Customer participation in debt auctions depends on the product area and term. Customers tend to participate more in RRB auctions, treasury bill auctions and 30-year nominal bond auctions. As a result, over this period, average customer allotment for RRBs is over 40 per cent, treasury bills vary between 5-10 per cent depending on term, and 7 per cent of 30-year nominal bonds are allotted to customers.

The data suggest that the buyback program does not impact auction effectiveness for 2 and 5 year auctions. For 10-year auctions, larger buybacks are associated with wider tails for switch buybacks, but there is no other evidence of an impact on effectiveness for 10-year auctions. For the impact on Bid Cover, the only statistically significant relationship was for the 30-year bond auctions with buybacks on a cash basis. The relationship has an R2 of 40 per cent and the coefficient is negative suggesting that Bid-Cover declines (effectiveness decreases) as buyback amounts increase.

For the impact on auction Tail, the relationships vary in significance with R2 from 0 per cent to 37 per cent and all the coefficients are positive, providing weak evidence that as buyback amounts increase, the Tail increases (effectiveness decreases). Note that the buybacks are intended to increase auction size and liquidity of the auctioned bonds, which might be expected to contribute to more effective auctions.

The Bank of Canada participates up to 15 per cent in nominal bond auctions and 25 per cent in treasury bill auctions. The general evidence is that the Bank of Canada’s participation does not impact auction effectiveness for any of the types of auction securities.

For nominal bonds, the multiple-regressions below suggest that the three factors jointly impact nominal bond auction effectiveness for maturities up to 10 years (R2 in the range of 25 per cent to 43 per cent). Most individual coefficients have marginal statistical significance in the multiple regression but individual regression results in Appendix C offer further supportive evidence.

There is little evidence that RRB auction effectiveness is influenced by changes in financial market conditions.

Multiple regressions for treasury bills support the conclusion that higher volatility, wider credit spreads, and lower liquidity are related to wider auction Tails, i.e. less effective auctions. The coefficients are statistically significant and consistent across all three maturities of treasury bill auctions, although the impact of credit spreads is somewhat weaker for 3-month treasury bills. The effect of the financial market condition factors on Bid-Cover is greatest for the 1-year treasury bill auction. Higher Yield Vol is associated with lower Bid-Cover for treasury bill auctions (in particular the 1-year treasury bill auction). Wider Baa spread increases Bid-Cover, likely reflecting a flight-to-quality phenomenon, i.e. as credit conditions worsen, market participants bid more aggressively for high quality liquid assets such as treasury bills (this is consistent with the evidence for short term bonds). Finally, the negative relationship between Bid-Cover and Liquidity suggests that when liquidity is high there is confidence in the secondary market, and thus little incentive to bid aggressively at auction.

Overall, treasury bill and nominal bond auction Tails ranged from 0.1 to 4.1 bps and 0.1 to 0.8 bps, respectively. Bid-Cover ratios for most treasury bill and nominal bond auctions are between 1.8 and 3.0. Under normal market conditions prior to March 2007 treasury bills have slightly wider Tails and lower Bid-Cover than nominal bonds. The Tails and Bid-Cover are also more variable for treasury bills than for nominal bonds. This same general pattern is also observed after March 2007 with the Tail and Bid-Cover for treasury bills increasing significantly (see p-values in table) from the period prior to March 2007. There is a significant increase in Tail for treasury bills after March 2007. For example, average tail for 3-month treasury bills increased from .4 to .9 bps, suggesting that the auctions were less effective in the later period. However, Bid-Cover was higher for treasury bills after March 2007, so tails were wider during that period, the auctions were effective in terms of covering the auction size.

For 2-year and 5-year auctions, in aggregate and individually, primary dealers generally have net long positions prior to auctions where the bond is a reopening of an existing issue and have neutral or net short positions prior to auctions where the bond is a new issue. While this general pattern is observed for 2- and 5-year auctions, it is not for 10- and 30-year auctions for which they tend to maintain net short positions for reopening as well as new issues. This may reflect the higher duration risk and relatively lower liquidity associated with longer term bonds, particularly for new issues, and the desire by dealers not to be long when auction supply is anticipated.

Dealers appear to position themselves for RRB auctions in a similar fashion as for 30-year nominal bonds, i.e. they tend to be short or neutral but more recently have shifted to being slightly long which has also coincided with an increase in auction size. The dealer net position in RRBs does not appear to be related to changes in real yields.

We measure dealer Willingness to Bid as the percentage (of total auction size) of dealer bid amounts with bid yield less than the auction average yield. We find no significant relationships between this Willingness to Bid and Yield Volatility, Baa credit spread and Auction Size for any securities. For RRBs, dealer willingness to bid is positively related to auction size but must be carefully interpreted.

Customer participation, which ranges from 0 to 8 per cent of nominal bond auctions, has little impact on auction effectiveness for 2- 10- and 30-year auctions. The effectiveness of 5-year auctions seems to be impacted favourably by customer participation and is statistically significant; however, this relationship is influenced by two extreme data points with high percentages of customer bids. Likely due to significantly larger customer participation in RRB auctions (representing 10 to 40 per cent of total bids), Bid-Cover and Tail are both favourably affected by increased customer participation. Further, the R2 is higher for auctions that occur near coupon payment dates.

Customer participation in treasury bill auctions ranges from 0 to 25 per cent of total bids. There is a significant positive relationship between Bid-Cover and Customer bids per cent for 6-month and 1-year treasury bill auctions (R2 greater than 30 per cent) and a weaker positive relationship with 3-month auctions (R2 = 8 per cent). There is no significant relationship between auction Tail and Customer bids per cent.

Appendix D - Primary Dealers’ Role 

Primary Dealers dominate nominal bond and treasury bill auctions, and Customer participation is greatest for RRB auctions.

Nominal bond auctions:

Primary Dealers dominate bidding and generally capture 60 per cent of the auction. Other government securities distributors have less than 5 per cent participation. Customer participation is low for shorter term auctions, and higher for 30-year auctions, sometimes as high as 10 per cent of these auctions (see chart below).

30Y: % Allotment (excl. BOC)

30Y: % share of Total Auction Bids (excl. BOC)

Real Return Bond auctions:

Customers play an integral role in RRB auctions, taking up to 83 per cent of some auction allotments. This reflects the nature of the buy-and-hold demand for RRBs among customers and the relatively lower liquidity of these bonds in the secondary market. Customers bid through government securities distributors and their share of RRB auction bids ranges from 10 per cent to 40 per cent.

RRB: % Allotment

RRB: % share of Total Auction Bids

Treasury bill auctions:

Similar to nominal bond auctions, Primary Dealers dominate bidding and allotments in treasury bill auctions but there is a very gradual decline in Primary Dealer participation offset by a gradual increase in customer participation.

Appendix E - Endnotes 


1 The 2009 study of debt issuance procedures in OECD countries provides a summary of changes to issuance procedures that occurred during the crisis. http://www.oecd.org/dataoecd/2/31/43002712.pdf

2 http://laws.justice.gc.ca/en/F-11/index.html

3 Approved Orders in Council are available on the Privy Council Office website at http://www.pco-bcp.gc.ca/oic-ddc.asp?lang=eng&Page=secretariats

4 http://www.fin.gc.ca/pub/dms-sgd/index-eng.asp. As circumstances warrant over the course of the fiscal year, the Minister approves significant initiatives or major changes to strategies.

5 http://www.fin.gc.ca/pub/dmr-rgd/index-eng.asp

6 http://laws.justice.gc.ca/en/B-2/

7 http://www.fin.gc.ca/treas/Goveev/TMGF_1-eng.asp

8 http://www.fin.gc.ca/access/fininst-eng.asp#treasury

9 http://www.bankofcanada.ca/cars/bd_auction_schedule.html

10 http://www.bankofcanada.ca/en/markets/markets_auct.html

11 Broadly speaking, the Canadian government debt auction process and those of the other sovereigns we surveyed are similar.  However, there are differences worth noting.  First, although regular auctions are the primary issuance method, some sovereigns use non-auction methods when there is reason to believe that price discovery will be difficult.  There are different approaches to the trade-off between predictability and transparency, especially in times of market turmoil, and the frequency of new issues compared to re-openings varies.  Every sovereign is aware of the possibility of abusive practices by bidders but they approach this problem differently.  Participation and non-competitive bidding also vary. Transparency and communication with participants is at least as effective in Canada as it is in other sovereigns.  Canada succeeds at rapidly releasing relevant information to participants after the auction, and the information released is similar to that of other sovereigns.  Modifying one or more aspects of the process should be evaluated in the context of the other aspects, since the various pieces work together as part of the broad process.  In 2009, the Survey of the OECD Working Party on Debt Management was conducted (endnote 1). That study notes that “…the principal issuing procedures are auction and tap. The responses show that all OECD countries use auctions for issuing new long-term debt, while 26 countries also use auctions for issuing short-term debt.  Syndication is used by a number of (mostly smaller) countries (usually from the euro-zone) for selling benchmark bonds.  It enables achieving very rapidly a high initial outstanding volume, thereby boosting liquidity and achieving greater placing certainty with lower borrowing costs. In addition issuers (also from larger countries) are using syndication for the first-time issuance of new instruments such as linkers or ultra-long bonds.”

12 While the ultimate decision-making authority rests with the Minister of Finance, the design of key strategies, policies and oversight of operations, and the coordination of funding, investment and liquidity management activities are largely delegated to several key committees, working groups and teams. Ministerial decisions are sought on issues such as strategic plans, investment policies and levels, and debt structure targets. The Funds Management Coordinating Committee (FMCC) meets regularly (typically once per month) or as needed to discuss key issues, develop policy advice and provide ongoing coordination of work on wholesale debt, cash management, and issues that involve both the domestic debt program and foreign currency management. In addition, the Domestic Debt and Cash Management Working Group meets regularly (typically once a month) to provide a forum for discussion and updates on a wide range of current issues associated with the implementation of funds management activities. The working group is responsible for reporting on ongoing activities and developing policy analysis and recommendations for the FMCC, in the lead-up to the FMC meetings.

13 We understand that Germany has a process that is similar to Canada’s current approach, but the sovereigns selected for this study in conjunction with the Department of Finance and Bank of Canada did not include Germany.

14 In an empirical study of newly issued 26-week US treasury bills with 26-week bills that are re-openings of old 52-week bills, re-openings cause lower prices. This indicates that the supply effect more than offsets the liquidity effect and in this instance, re-openings can be inferior to new issues.

15 A detailed analysis of the buyback process is beyond the scope of this analysis. The focus here is on the impact on auction effectiveness.

16 In Australia all communication and execution of auctions was previously conducted using a Bloomberg platform. It is now conducted through the same trading system that is used in the secondary market so all market participants are aware of auction details and results simultaneously and efficiently.

17 The literature notes that while the information revealed in the WI market can affect the auction outcome, the auction also affects the WI market. Research documented in the literature suggests that traders hesitate to trade in the WI market prior to auction for fear of revealing information to their competitors, and when-issued trading volume decreases and bid-ask spreads widen in the few minutes prior to the auction. This may be due to a reluctance to reveal information that could be potentially valuable in the auction itself. The potential for a short squeeze is a negative consequence of the existence of the when-issued market. Issuer flexibility, such as the ability to reject bids, to reduce supply, and especially to re-open the issue can prevent short squeezes.

18 The empirical literature examining treasury auction outcomes does not directly focus on the process, but it does have some relevant results. Bid prices, the dispersion of bids, and the quantities are detrimentally affected by uncertainty and volatility at the time of the auction. There is empirical evidence that wider dispersion of bids leads to more underpricing in US treasury auctions.

19 The US encourages broad access, even having a system that allows direct customer participation. At the other extreme, Australia limits participation to a specific list of institutions. The remaining countries are in the middle. All of the sovereigns we interviewed have a category of bidder similar to primary dealers. Other than in Australia, primary dealers have auction participation requirements to maintain primary dealer status.

21 See Endnote 1.

22 In Australia, volatility in futures spreads to bonds has the most impact on willingness to bid since it affects hedging ability. A recent development (September 2009) in the Australian market has affected willingness to bid and coverage ratios. The commonwealth government announced a guarantee to state government debt for a fee. As a result, investors can buy state debt at a higher yield with a sovereign guarantee. This has affected auction results. For example coverage ratios have been around 1.9 compared to averages of 3 or higher beforehand. The government had previously provided guarantees to banks but this did not impact auctions since the guarantee was limited to debt up to 5 years and banks were issuing offshore.

23 In both Canada and the US, the system for accepting bids is proprietary to the issuing authority. In France, regular bidders use TELESAT and occasional bidders use SWIFT. In the UK, a Bloomberg system has been used for gilts since 2007, but bills are auctioned using a telephone system, with an anticipated move to an electronic system for bills later in 2009. Australia used a Bloomberg system for accepting bids, but switched to an internet-based system in March 2009 which is integrated with the bond trading platform used by all primary dealers. This is possible and effective since Australia allows only primary dealers to participate in auctions.

24 The UK is the extreme of the sovereigns we surveyed in this dimension, requiring that dealers submit records of all bond trades for the week prior to the auction as well as the pre-auction net position. There are no pre-auction reporting requirements for bills in the UK. France requires ex post reporting if certain auction limits are met. Australia has no requirements regarding position reporting and, while it feels confident it could resolve any problems that might arise from excessive post-auction concentration of ownership, has never needed to act. Australia does not believe that reporting requirements would improve auction effectiveness in their market. The US requires pre-auction reporting only if the 35 per cent limit will be violated and so is less resource intensive than, for example, the pre-auction reporting requirements in the UK for gilts.

25 For example, in June 2009, the UK announced a post-auction option facility open from 12 to 2pm (starts an hour after the auction closes) where dealers can take down an additional 10 percent of the auction amount at the auction price i.e. they make a profit if the secondary market price rises. The option does not apply if the auction is not covered. The post-auction option facility was the result of a 2008/2009 UK consultation on supplementary issuance methods, which looked at a variety of alternative issuance methods.

26 We measure the relative participation of dealers compared to investors as relative bid, defined as the natural logarithm of the ratio of total dealer bids to customer bids. The log of the ratio is used because there are many instances where investor participation is very small relative to dealer participation.

27 The number of bidders at nominal bond auctions ranges from 16 to 23 and does not vary significantly over time. There is little effect of market conditions, except that as yield volatility has increased recently there has been a small decline in the number of bidders. There is a weak positive relationship between Number of Bidders and Baa Credit Spread for all but the 5-year auction. This might reflect a flight-to-quality effect increasing the number of bidders at auctions for government securities as credit conditions worsen. Consistent with greater customer participation in these auctions, there is a relatively higher number of bidders in RRB auctions. The number of bidders ranges from 20 to 30 compared to 18 to 21 for 30-year nominal bond auctions. Similar to nominal bond auctions however, there is no significant relationship between the number of bidders and market conditions. The number of bidders at treasury bill auctions has gradually declined from about 20 bidders in 2003 to 16 in 2008. There is little impact of market conditions on the number of bidders.

28 With regard to short squeezes, there are tradeoffs involved in re-openings. While re-opening can alleviate a squeeze, there are difficulties. Because it is difficult to measure exactly when a squeeze is occurring, a re-opening would have to be done at the discretion of the Bank of Canada rather than by a fixed formula (which is the approach in Australia). However, this possibility adds uncertainty to the market (which may impact auction prices). The number of primary dealers in itself does not appear to be a concern with respect to potential irregular practices. Given the number of auction participants, some sovereigns prevent concentration of ownership through rules, while others take a less rules-based approach and would deal with a potential concentration of ownership by re-opening an issue at any time (e.g. Australia).

29 http://www.bankofcanada.ca/en/markets/markets_auct.html

30 According to the Bank of Canada, there is no operational issue internally. See also http://www.ccma-acmc.ca/en/key_priorities/national24101.html

31 In Canada, as long as dealers are willing to live up to their obligations as primary dealers, the bidding rules guarantee the Bid Cover is at least one, and the chance of a failed auction is remote. Likely causes are extreme market events, natural disaster, failure of major financial institutions, or similar extreme conditions.

32 Execution risk is measured primarily as the likelihood of a Bid-Cover ratio less than one. The actual Bid-Cover for each auction is monitored by Debt Management and by Risk Management and reported to senior management and the Department of Finance. Execution risk is controlled through primary dealer selection criteria, minimum bidding requirements, consultation with market participants, transparency of the Debt Management Strategy, and selection of auction size and terms of the calls for tender.

33 Enforcement of net position reporting is through various channels. If customers do not sign off on their compliance certificate, they are removed from the CARS system and no longer permitted to bid. The compliance framework in the Debt Management Group is used to impose consequences for dealers. At the time of the auction, if a large net position is reported, the CARS system limits or stops access to an auction automatically. If there is suspicion of a squeeze on an issue, IIROC could request a net position report (rule 2800, formerly IDA rule 5, governs irregular trading practices). However, the threshold that constitutes a squeeze is vague. The Bank of Canada uses 25 per cent of amount outstanding. The Bank of Canada, IIROC, and Department of Finance have a right to ask for net positions at any time, not just at auction or on formal review dates.

34 Bonds and bills settle three days after the auction (T+3), and 2-year auctions are T+2. All auctions are settled through the third-party CDSX settlement system. The CDSX system has controls that require irrevocable simultaneous exchange of cash for assets. The CDSX settlement system is key to the reduction of settlement risk. Each sovereign in our study uses similar standardized third-party settlement systems. None of the sovereigns surveyed expressed any concern regarding settlement risk, and no concerns were expressed by any Canadian market participants or by the Bank of Canada. Dealers and customers said that the settlement procedure is good overall.

35 Specific to the auction process, the CARS system is proprietary to the Bank of Canada and uses dedicated computers at each dealer site. This minimizes the risk of failure due to external systems related failures, but is still subject to its own structural integrity, and disaster. In contrast to some other sovereigns (e.g. Australia), the systems supporting the auction do not depend on third-party systems, such as Bloomberg, or trading systems.

36 Note that detailed evaluation of alternatives to an auction is not within the scope of this study. This question is addressed at a very high level. Government Evaluation Policy requires us to examine issues around program relevance, which is an assessment of the persistence of the issue/problem that the program has been established to address.

37 3-year re-introduced in 2009 to provide added flexibility re increased funding requirements and smooth debt maturity profile.


Appendix F – Management Response and Action Plan 

PROLOGUE

The Department is pleased to note that the report submitted by Twist Financial Corp concludes that the current debt auction process structure works well, and supports the transparency, effectiveness and efficiency of debt auctions. Twist reports that the auction process has been successful in its immediate, intermediate, and ultimate goals of raising necessary funding at a low cost, and that it has helped sustain a liquid and efficient secondary market for Government of Canada debt.

The report finds that the major global issuers of debt tend to use broadly similar issuance procedures and debt management policies and that the many similarities between Government of Canada debt auctions and those of other sovereigns suggest limited ways in which other sovereigns’ practices might be used to improve Canada’s auctions.

Twist concludes that the debt auction process can continue to achieve its objectives in its current form. That said, the report suggests marginal changes which could be made to improve the process.

Recommendation Management Response Planned Action Lead Target Date
Measures to improve communications and reporting between debt managers at the Department of Finance and the Bank of Canada:
  • For potential improvements in efficiency, consider changing the interaction between the Department of Finance and the Bank of Canada occurring ahead of bond auctions to a reporting function rather than an approval process.

The Department disagrees with this recommendation.

Comment: With respect to the suggestion that consideration be given to changing the interaction between the Department of Finance and the Bank of Canada ahead of bond auctions, the current framework has proven to be effective. The Department believes that the active involvement of both the Bank of Canada and the Department of Finance in decisions surrounding each bond operation ensures that due consideration is given to the evolving borrowing requirements of the Government within the context of the overall debt strategy.

None None None
  • Replace authorization letters for debt auction operations with more informative, timely reporting and accountability to senior management.

The Department agrees with this recommendation.

Comment: The delegation of borrowing authority and how that authority is recorded and communicated to the Bank of Canada needs to be improved. A small task force composed of individuals from the Debt Management Section, Internal Audit, Public Debt Reporting, Justice and the Bank of Canada will be charged with moving the issue forward.

A system for providing internal quarterly debt reporting is currently under development. Under this new reporting system, authorization letters would no longer be required. Department of Finance By the end of 2010-11

Measures to improve communications with market participants:

  • Increase the frequency of consultations with a larger sample of current and potential customers utilizing a mix of communication formats (including more frequent telephone conference calls and multi-lateral discussions in addition to the traditional bi-lateral visits with individual dealers).

The Department agrees with this recommendation.

Measures have already been taken to enhance the effectiveness of the on-going dialogue with market participants. For example, a broader array of participants and customers were included during 2009-10 consultations. More frequent consultation meetings have been conducted since 2008-09, with some on specific issues (e.g., treasury bill program, long bonds). Multi-lateral consultations were found to be especially useful during the financial crisis, with special meetings held with primary dealers in November 2008 and June 2009.

Department of Finance and the Bank of Canada

On-going

  • Obtain detailed and more consistent feedback from a larger number of market participants, including large and small customers, prior to and following auctions of Real Return bonds and 30-year nominal bonds.

The Department agrees with this recommendation.

The Bank has had a long-standing process in place to obtain feedback from market participants, including customers, prior to and following auctions, however, the procedure will be reviewed.

Bank of Canada

By the end of 2010-11

  • Release auction results simultaneously on the Bank of Canada pages on Reuters and Bloomberg and through CanDeal and other relevant trading systems used by Canadian market participants.

The Department agrees with this recommendation. However, the Department notes that timely auction results are already available on some trading systems.

Comment: The Bank of Canada offers email notifications whenever the Bank releases new information, such as auctions results (http://www.bank-banque-canada.ca/en/messenger/index.html).

The process used to disseminate information regarding auction operations will be reviewed.

Bank of Canada

By the end of 2010-11

Measures to improve participation at auctions:

  • Encourage all market participants to play a more active role in the design of the debt strategy, such as through an annual workshop in addition to the standard consultation process.

The Department agrees with this recommendation.

The Bank of Canada and the Department of Finance will review the possibility and benefits of holding workshops with market participants as part of the debt strategy consultations exercise.

Department of Finance and the Bank of Canada

By the end of 2011-12

  • Promote greater participation of customers already holding a bidder identification number, and identify and actively solicit potential new customers. Large customers should be offered direct access to the Communication, Auction and Reporting System (CARS).

The Department agrees that any measures taken to encourage greater participation by customers would be beneficial to the federal debt program. The Department will need time to assess how this recommendation fits within the broader debt distribution framework.

The Department, in conjunction with the Bank of Canada, will continue to promote participation at auctions. It is agreed that any measures taken to encourage greater participation by customers would be beneficial to the federal debt program.

Department of Finance and the Bank of Canada

By the end of 2011-12

  • Relax the upper limit on bids and emphasize the government’s right to re-open a security at any time through something like the United Kingdom mini-tender facility, or similar to the Australian approach of re-opening at its discretion, which will both support participation and counteract the possibility of a participant attempting to exert undue influence on the price of a security.

The Department disagrees with this recommendation at this time.

Comment: Debt management practices in Canada have evolved over time to respond to the specific needs of our domestic marketplace; initiatives used in other countries may or may not be applicable for Canada. In view of the overall results of the treasury evaluation of the Debt Auction Process and the success of debt operations throughout the financial crisis, this recommendation will not be pursued at this time.

None

None

None

  • Encourage Government Securities Distributors to strive to become Primary Dealers (who are subject to more stringent minimum bidding requirements) by attaching more visibility and prestige to the primary dealer status (e.g., as in France).

The Department agrees with this recommendation.

Government Securities Distributors and Primary Dealers in particular, are essential to the continued success of Canada’s debt distribution framework. The report’s suggestions on how to better promote Primary Dealer status, including establishing an annual multi-lateral forum, will be examined carefully.

Department of Finance and the Bank of Canada

By the end of 2011-12

Measures to support the transfer of corporate knowledge:

  • Review and update annually detailed guidelines at the management level at both the Bank of Canada and Department of Finance.

The Department agrees with this recommendation.

Departmental practices and documentation will continue to be strengthened to support the transfer of corporate knowledge and reduce potential operational risks related to employee turnover.

Department of Finance and the Bank of Canada

By the end of 2011-12