Price Effects of Sovereign Debt Auctions in the Euro-zone: The Role of the Crisis

Transcription

1 Price Effects of Sovereign Debt Auctions in the Euro-zone: The Role of the Crisis Massimo Giuliodori (University of Amsterdam and TI) Roel Beetsma (University of Amsterdam and TI) Frank de Jong (Tilburg University) Daniel Widijanto (BNG Vermogensbeheer)

2 Motivation Recent crisis led to a renewed interest on the role of fiscal policy as macroeconomic tool and on debt sustainability Many papers on the determinants of sovereign bond yields stressing the role of credit ratings, news, contagion, etc Little work on the relationship between secondary and primary government bond markets This is surprising given the continuous coverage by the financial press on upcoming (and results of) auctions

3 (20 th April 2012) Italy next in line to test debt market nerve Italy s borrowing costs are under pressure to rise further at a bond sale next Friday because investors are nursing renewed doubts about whether highly-indebted euro zone governments can get their finances under control. Five days is a long time in the euro zone crisis and the market mood may swing sharply, but the auction has been singled out as a central "risk event", watched not only by bond investors but by most people involved in global financial market. If the auction results disappoint, the market could soon test 6 percent again.

4 Structure Related Literature and Conceptual Framework Dataset Description and Auction Statistics Event Study Regression Analysis Summary and Conclusions

5 Related Literature Literature on auctions and bond yields Auction yields relative to yield of almost identical bonds at the moment of the auction (Simon, 1994; Nyborg and Sundaresan, 1996; Goldreich, 2007) Typically underpricing of auctioned bond (around 1 bp) Secondary market yield movements in the days around auctions (Fleming and Rosenberg, 2007; Lou et al., 2013; Forest, 2012) Effects are much larger Yields peak significantly at auction day compared to yields five days before or after auction

6 What We Do in This Paper Construct a new dataset on Treasury auctions for Germany and Italy from January 1999 until mid-february 2013 Explore the secondary market yields movements around (before and after) auctions, distinguishing between the precrisis and the crisis period Study the presence of spillover effects across maturities and from foreign auctions Investigate the role of nonlinear effects related to the size (or volume allotted) and bid-to-cover ratios Assess the potential issuance costs

7 Why Should Yields React to Auctions? Supply effect: auctions increase the supply of bonds P yield Liquidity effect: liquidity of off-the-run (old) bonds P yield Maturity effect: new on-the-run bonds have a longer maturity. Assuming upward-sloping term structure, yield of new bond will be higher than yield on old bond ceteris paribus Limited risk-bearing capacity effect: risk-averse primary dealers participating in auctions need compensation for their positions in the asset and the price risk on their inventories

8 Model of Primary Dealers (Ho and Stoll, 1983) Oligopolistic dealers, maximizing a mean-variance utility function. The price the dealer quotes will be: This implies that: and where and

9 Predictions of Model 1. Upcoming auction of asset i (leading to a positive position in asset i) P i yield i 2. The larger the size of auction (and size of inventory) the larger the above effect: 3. Given high correlation, yield of close-maturity asset j will 4. The larger the size of auction (and size of inventory) the larger the above effect: 5. In a crisis period, the variance of return on asset i will be higher, leading to stronger price/yield effects

10 Dataset Description Sample period: January 1999 until mid-february 2013 Country sample: Italy and Germany Secondary market daily yields for 2, 5 and 10-year Treasury bonds (Bloomberg) Auction data for 2, 5 and 10-year Treasury bonds (Bloomberg and national debt agencies): (i) auction date, (ii) total amount bid, (iii) total amount allotted, and (iv) average accepted yield

11 Secondary and Primary Yields (Germany, 5- year)

12 Secondary and Primary Yields (Italy, 5-year)

13 Auction Statistics

14 Event Study The event study calculates the average of y(t)-y(0), where y(t) is the end-of-day yield of the on-the-run Treasury bond on day t, and y(0) is the end-of-day yield of the same headline maturity bond on the auction day This follows the approach of Lou et al. (2013) We look at 2, 5 and 10-year maturity Caveat: Bloomberg reports the yield of the bond or bill with maturity closest to the headline maturity. As a result the underlying benchmark bond may switch within the remaining part of the window after the auction date

15 Event Study: y(t)-y(0) (full sample)

16 Event Study: y(t)-y(0) (before-crisis)

17 Event Study: y(t)-y(0) (crisis)

18 Event Study: Results Evidence of an inverted V-shaped pattern: secondary market yields increase before and fall after the auction ( auction cycle ) Effects are larger and more significant for Italy than for Germany Yield movements are largely confined to the crisis period, particularly in Italy Secondary market yields with maturity close to the maturity of the newly issues bonds show similar pattern Results are consistent with the theory of limited-risk bearing capacity of dealers

19 Regression Analysis Drawback of event study: potential presence of confounding factors (e.g. other auctions) during event window For maturity m of country i, we estimate: where AUC is a dummy that takes a value of 1 (0) when there is an (no) auction of the specified country-maturity combination F-test for significance of auction cycle. E.g. for own auction:

20 Regression Analysis: Auction Dummy

21 Regression Analysis: Results Regressions with auction dummies are in line with the results of event study, even when controlling for other auctions Auction cycle is statistically symmetric (not shown) Some evidence of effects of other domestic maturities on yields, namely in Italy during the crisis period Foreign same maturity auctions are generally found to be statistically insignificant The above results are confirmed when the auction dummy is replaced with the auction volume/size Large impact: 3bp/billion (full sample) to 5bp/billion (crisis)

22 Regression Analysis: Auction Volume Germany Italy 2-year 5-year 10-year 2-year 5-year 10-year Full sample period Volume own *** 3.17 *** 2.56 *** 2.04 *** January 1, 1999 June 30, 2007 Volume own *** July 1, 2007 February 12, 2013 Volume own *** *** 4.11 *** 3.54 ***

23 Does the Bid-to-Cover Ratio Play a Role? Bid-to-cover ratio: total amount bid over total amount allotted Our hypothesis: the higher the bid-to-cover ratio, the more successful the auction, and the smaller the auction cycle For maturity m of country i, we estimate: where BC is the bid-to-cover ratio (0) when there is an (no) auction of the specified country-maturity combination F-test for significance of bid-to-cover effect:

24 Role of Bid-to-Cover Ratio in Auction Cycle

25 Role of Bid-to-Cover Ratio: Results Our hypothesis is statistically significant only for the full sample and in Italy, and 10-year bond in Germany Results are robust when only past or known bid-to-cover ratios are included Some evidence that failed auctions drive up yields even more Subsamples too few observations

26 Probing Further into the Effects of Crises Crisis period is characterized by periods of smaller and larger tensions. Are the auction effects on yields constant? 1. We re-estimate the baseline regression for subsamples 2. Link yield movements around auctions to indicators of tensions in the European sovereign debt market: CYCLE = size of full auction cycle CRINT = degree of crisis intensity proxied by KfW-Bund spread (liquidity pricing effect as measure of stress in European sovereign debt market) and ITRAXX Euro 5- year index (credit risk)

27 Probing Further into the Crises : Subsamples Most turbulent period associated with largest effects

28 Indicators of Crisis Intensity

29 Probing Further into the Crises: Crisis Indicators

30 Issuance Costs for Italy: Some Numbers Issuance costs for the 5-year Italian bond for full sample : Average auction size: 2,519 million euros Run-up of secondary market yield: 3.12 basis points Additional cost per year: 785,000 euros Av. duration typical 5-year bond is 4.44 years: 3.50 million euros Effect for total outstanding debt: 1640 bln * 3 bp = 500 mln Effects increase significantly from first to second period: First period: 2,183*0.35bp*4.36 = 330,000 euros Second period: 3,080*7.53bp*4.52 = 10.5 mln Full roll-over: 1640*( ) = 1.2 bln

31 Summary and Conclusions Since the inception of the euro, we find evidence of an inverted V-shaped pattern: secondary market yields increase before and fall after the auction ( auction cycle ) Effects are larger in Italy and concentrated in the period of the crisis since mid-2007 Also bonds with a close maturity follow a similar pattern These results are consistent with the limited risk-bearing capacity model over explanations based on supply, liquidity and maturity effects Back-of-the-envelope calculations show significant additional issuance costs for the Italian Treasury since the crisis

32 Extras

33 Auction Procedure (Germany)

34 Assessing the Potential Issuance Costs First stage: is there a strong relationship between primary and secondary market yields? For maturity m of country i, we estimate: where is the average accepted yield on the auction day is secondary-market beginning auction day t Parameter α 1 is always close to one: movement in issuance cost can be approximated by movement in secondary market during run-up of auction

35 First Stage: Primary vs Secondary Yields

36 Assessing the Potential Issuance Costs Second stage: What is the secondary market yield movement during the run-up to the auction? For maturity m of country i, we estimate:

37 Second Stage: size of run-up to Auction

38 Profit Opportunities Can the yield pattern around auctions be exploited in a trading strategy? Two issues: transaction costs and risk of the strategy Pellizon et al. (2013) report bid-ask spreads for Italian treasury bonds in recent (crisis) period Quoted spread for 5-year bond is 0.378% of the price Translated to yields it is 8.5bp (effective spread is 2.7bp) Volatility of bond yields is somewhat higher on auction days Standard deviation of daily yield change is 12bp Sharpe ratio of sell before and buy at auction day is SR is 0.62 if one can trade at the effective spread Tilburg University