Monitoring International Borrowers The IMF s Role in Bank

Monitoring International Borrowers: The IMF’s Role in Bank and Bond Markets Barry Eichengreen, Kenneth Kletzer, and Ashoka Mody June 15, 2004

IMF Program Functions • Monitoring – Reveals country policies and intentions – Hence, reduces variance of expected policy outcomes – But only if country commitment is credible • Lending – Reduces likelihood of default – Hence, lowers probability of costly creditor coordination and restructuring

Research Design • Banks versus Bonds – If banks already provide monitoring, then is IMF monitoring relevant mainly for bonds? • Country Debt Thresholds – Vulnerability to external debt increases non-linearly – Debt thresholds demarcate: • Liquidity risk (30 -55%), Solvency Risk (55%+) • Country policy commitments • Precautionary Programs – “outset” and “turned”

Theoretical Considerations • Country fundamentals and policies: – determine ability and willingness of the government and private debtors to repay debt – beyond debt threshold, market access and spreads worsen • Information asymmetries generate risk premia: – borrower-lender pairs sort into different markets for different types of monitoring; – banks “delegated” monitors: repeated borrowing; – IMF programs more relevant in bond markets – precautionary programs particularly relevant • Banks coordinate creditors and restructuring

Data • Loanware and Bondware • Individual transactions, 1991 -2002 • Characteristics at initiation of transaction – E. g. , Spreads, maturity, amount • Supplemented with: – Global characteristics (US growth, swap rates, volatility of EMBI) – Country features (e. g. , debt/GDP, growth)

Key features of bank and bond markets • 1990 s saw revivial of bond markets • Banks, as delegated monitors, allow: – Lower entry barriers, steadier access – Smaller, more diverse, episodic borrowing • Shift to bonds under IMF program – 22 percent of all loans under IMF program – 33 percent of all bonds during program – IMF programs especially facilitate bonds by mid-level debt countries (30 -50% of GDP).

Table 1: Trends in International Bond and Bank Lending Aggregate Value of Transactions (US$ billions) Number of Transactions Year Bonds Loans Total 1991 81 209 290 10 24 34 1992 177 252 429 21 18 39 1993 357 376 733 45 27 73 1994 307 508 815 39 40 79 1995 369 750 1, 119 48 56 104 1996 522 1, 066 1, 588 81 83 164 1997 555 1, 248 1, 803 100 125 225 1998 234 550 784 52 62 114 1999 334 402 736 65 47 113 2000 284 532 816 59 81 141 2001 290 470 760 78 62 140 2002 219 384 603 63 44 107 3, 729 6, 747 10, 476 661 669 1, 331 Total

Repeated Borrowing • Measure takes value 1 for borrower’s first international debt contract in that market. • Each subsequent borrowing increases the value of R by one. • More repeated borrowing in the bond market: – median number of bond borrowings is 3 (75 th percentile is 8 and 90 th percentile is 27); – for banks, the median is 2 (75 th percentile is 4 and 90 th percentile is 8).

Table 2: Number of Transactions, by Debt Category and IMF Program Type of Credit Low Debt/GDP Range (0 -30 percent) No Program IMF Program None 1, 301 389 Bonds 1, 244 57 Loans 2, 606 99 Medium Debt/GDP Range (30 -55 percent) No Program IMF Program None 973 561 Bonds 958 1, 000 Loans 2, 094 898 High Debt/GDP Range (more than 55 percent) No Program IMF Program None 675 808 Bonds 253 217 Loans 589 461 Full Sample No Program IMF Program None 2, 949 1, 758 Bonds 2, 455 1, 274 Loans 5, 289 1, 458

Table 3: Choice of Loans and Bonds, Relative to No Borrowing (1) (2) Debt/GDP<=0. 3 (3) (4) 0. 3<Debt/GDP<=0. 55 (5) (6) Debt/GDP>0. 55 Bonds Turned Precautionary Loans 0. 308 0. 146 0. 827 0. 369 0. 674 0. 894 [0. 61] [6. 41]** [2. 88]** [3. 79]** [5. 71]** – 0. 335 – 1. 001 – 0. 963 – 0. 758 – 0. 519 – 0. 756 [0. 51] Precautionary Bonds [1. 19] IMF Program Loans Bonds Loans [1. 69] [4. 75]** [3. 89]** [2. 34]* [3. 83]** 0. 340 0. 211 0. 204 0. 314 [1. 93] [1. 20] [0. 80] [1. 42]

Testing for Effects on Spreads log (spread) = βX + u 1 (1) B' = γZ + u 2 (2) u 1 ~ N(0, σ), u 2 ~ N(0, 1), corr (u 1 , u 2 ) = ρ

Table 4: Pricing of Loans and Bonds (1) (2) (3) (4) Loans Debt/GDP range Low Medium (5) (6) Bonds High Low Medium High Spread Equation IMF Program 0. 152 – 0. 130 0. 018 – 0. 096 – 0. 233 [5. 98]** [4. 33]** [2. 13]* [0. 20] [1. 88] [3. 03]** – 0. 192 – 0. 079 – 0. 162 – 0. 081 0. 003 [14. 57]** Log of Repeat Borrowing 0. 529 [6. 06]** [5. 60]** [3. 36]** [0. 16] [0. 13] Selection Equation IMF Program – 0. 067 0. 148 0. 372 0. 168 0. 421 0. 186 [0. 64] [3. 72]** [6. 81]** [1. 35] [9. 20]** [2. 65]** No. of Transactions 2, 477 2, 426 2, 057 1, 771 1, 556 1, 355 Observations 4, 022 4, 729 2, 523 2, 635 3, 711 1, 998 Robust z statistics in brackets, * significant at 5%; ** significant at 1%.

Table 5 A: Loans: Impact of IMF Programs and Repeat Borrowing (1) Debt Range (% of GDP) 25 -45 (2) (3) (4) (5) 30 -50 35 -55 40 -60 45 -65 (6) (7) (8) (9) 50 -70 55 -75 60 -80 65 -85 Spread Equation IMF Program 0. 279 [7. 39]** – 0. 058 0. 135 [3. 10]** – 0. 064 – 0. 041 [0. 77] – 0. 149 – 0. 054 [0. 85] – 0. 162 – 0. 018 [0. 29] – 0. 158 – 0. 046 [0. 57] – 0. 174 – 0. 019 [0. 20] – 0. 144 0. 095 [1. 08] – 0. 111 [5. 22]** Repeat Borrowing 0. 259 [6. 88]** – 0. 073 [4. 02]** [3. 59]** [6. 81]** [7. 10]** [6. 38]** [5. 54]** [3. 64]** [2. 41]* 0. 449 [5. 09]** 0. 408 [4. 26]** – 0. 103 [0. 83] Selection Equation IMF Program 0. 191 [2. 79]** 0. 115 [1. 66] 0. 287 [4. 03]** 0. 299 [4. 24]** 0. 415 [5. 51]** 0. 408 [5. 39]** No. of Transactions 2, 477 2, 426 2, 057 1, 771 1, 556 1, 355 887 571 358 Observations 3, 941 3, 804 3, 354 3, 102 2, 899 2, 647 1, 970 1, 471 949 Absolute value of z statistics in brackets, * significant at 5%; ** significant at 1% .

Table 5 B: Bonds: Impact of IMF Programs and Repeat Borrowing (1) Debt Range (% of GDP) (2) (3) (4) (5) (6) (7) (8) (9) 25 -45 30 -50 35 -55 40 -60 45 -65 50 -70 55 -75 60 -80 65 -85 Spread Equation IMF Program – 0. 080 [1. 75] – 0. 008 – 0. 161 [2. 91]** 0. 001 – 0. 315 [4. 41]** 0. 027 – 0. 367 [4. 67]** 0. 048 – 0. 262 [2. 70]** 0. 049 – 0. 149 [1. 45] – 0. 018 – 0. 02 [0. 33] 0. 011 0. 038 [0. 48] 0. 006 [0. 79] Repeat Borrowing – 0. 045 [1. 09] – 0. 012 [0. 55] [0. 04] [1. 28] [2. 08]* [1. 39] [0. 51] [0. 36] [0. 16] 0. 258 [2. 67]** 0. 131 [1. 26] – 0. 13 [1. 02] Selection Equation IMF Program 0. 748 [10. 49]** 0. 610 [8. 58]** 0. 329 [4. 35]** No. of Transactions 1, 497 1, 539 1, 116 Observations 3, 068 3, 038 2, 537 0. 315 [4. 02]** 0. 270 [3. 24]** 0. 073 [0. 88] 899 707 580 352 272 200 2, 351 2, 170 1, 973 1513 1, 212 814 Absolute value of z statistics in brackets, * significant at 5%; ** significant at 1%.

Table 6: Does the Amount of IMF Lending Matter? (1) (2) (3) (4) Loans (5) (6) Bonds Debt/GDP range Low Medium High Low Spread Equation Medium High IMF Program 0. 479 [0. 45] 0. 589 [2. 68]** – 0. 362 [1. 18] – 2. 101 [1. 17] 0. 311 [1. 91] – 2. 041 [4. 11]** IMF*Debt/GDP – 2. 759 [0. 83] – 1. 688 [3. 72]** 0. 245 [0. 54] 7. 088 [1. 18] – 1. 049 [2. 70]** 2. 927 [4. 03]** IMF Amount/Debt 26. 396 4. 874 2. 146 5. 139 0. 012 – 3. 505 [3. 63]** [6. 61]** [1. 73] [1. 34] [0. 02] [2. 08]* – 0. 164 0. 070 – 0. 167 – 0. 156 – 0. 088 – 0. 243 [5. 75]** [1. 14] [0. 87] [3. 28]** [1. 38] [1. 75] – 0. 153 – 0. 400 0. 006 0. 433 0. 245 0. 370 [1. 01] [2. 47]* [0. 02] [2. 03]* [1. 71] [1. 90] Log of Repeat Borrowing Repeat*Debt/GDP Selection Equation IMF Program 1. 066 2. 842 2. 077 [2. 58]** [2. 51]* [1. 14] [12. 16]** [6. 15]** – 3. 158 1. 019 – 0. 811 – 3. 605 – 5. 883 – 3. 225 [2. 29]* [1. 78] [0. 99] [10. 51]** [6. 77]** – 5. 336 4. 794 3. 610 1. 047 0. 605 7. 200 [4. 51]** Observations 0. 787 [0. 93] IMF Amount/Debt – 0. 516 [1. 09] IMF*Debt/GDP 0. 991 [8. 88]** [3. 70]** [0. 44] [0. 75] [5. 68]** 4, 022 4, 729 2, 523 2, 635 3, 711 1, 998 Robust z statistics in brackets, * significant at 5%; ** significant at 1%.

Table 7: Is Precaution Valuable? (1) Debt/GDP range IMF Program Low – 0. 191 High Spread Equation – 0. 021 [0. 28] – 0. 202 [2. 13]* – 0. 149 [1. 60] – 0. 166 [6. 25]** Turned Precautionary Log of Repeat Borrowing (3) 0. 283 [6. 41]** – 0. 374 [5. 36]** – 0. 100 [1. 78] – 0. 081 [14. 55]** Precautionary 0. 552 [5. 95]** – 0. 389 [2. 64]** (2) Loans Medium [5. 73]** (4) Low 0. 038 [0. 39] – 0. 293 [2. 51]* (5) Bonds Medium (6) High – 0. 082 0. 006 [0. 12] – 0. 172 [3. 11]** – 0. 212 [4. 97]** 0. 005 – 0. 119 [1. 42 ] – 0. 205 [1. 51] – 0. 270 [1. 52] – 0. 001 [3. 39]** [0. 30] [0. 03] Selection Equation IMF Program Precautionary 0. 003 [0. 03] – 0. 446 [1. 83] 0. 236 [4. 52]** – 0. 377 [5. 38]** – 0. 010 [0. 17] 0. 433 [6. 27]** – 0. 415 [4. 38]** 0. 218 [2. 61]** 0. 191 [1. 46] – 0. 141 [0. 43] 0. 403 [7. 70]** – 0. 419 [5. 35]** 0. 261 [4. 42]** 0. 206 [2. 52]* – 0. 196 [1. 83] 0. 188 [1. 50] 4 , 022 4 , 729 2 , 523 2 , 635 3 , 711 1 , 998 Turned Precautionary Observations Robust z statistics in brackets, * significant at 5%; ** significant at 1% , .

Conclusions • Evidence consistent with segmentation of bank and bond markets • Repeated borrowing more valuable for banks but not for bonds • IMF monitoring more valuable in bond markets: – Especially in mid, liquidity risk zone – Lending amount more important when solvency at stake • IMF redressed disadvantage of bond markets, helping it grow in 1990 s: – But contributed to moral hazard and debtor coordination problems?