Is Transparency Good For You? by Rachel Glennerster, Yongseok Shin

27
1 Is Transparency Good For Is Transparency Good For You? You? by by Rachel Glennerster, Yongseok Shin Rachel Glennerster, Yongseok Shin Discussed by: Campbell R. Harvey Duke University National Bureau of Economic Research IMF Annual Research Conference November 4-5, 2004

description

IMF Annual Research Conference November 4-5, 2004. Is Transparency Good For You? by Rachel Glennerster, Yongseok Shin. Discussed by: Campbell R. Harvey Duke University National Bureau of Economic Research. Motivation. Popular Opinion: - PowerPoint PPT Presentation

Transcript of Is Transparency Good For You? by Rachel Glennerster, Yongseok Shin

Page 1: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

1

Is Transparency Good For You? Is Transparency Good For You? by by

Rachel Glennerster, Yongseok ShinRachel Glennerster, Yongseok Shin

Discussed by:

Campbell R. HarveyDuke University

National Bureau of Economic Research

IMF Annual Research ConferenceNovember 4-5, 2004

Page 2: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

2

Motivation

Popular Opinion:• Emerging markets financial crisis caused or made worse

by lack of transparency

Purpose of paper:• Does increased transparency reduce sovereign spread

(which presumably implies a lower probability of crisis)?

Page 3: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

3

Measures of Transparency

Measures: • Probability of crisis: country credit spreads• Transparency:

– IMF Article IV publication– SDDS adoption– ROSC adoption

Page 4: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

4

Overview of Findings

Results:• Increased transparency credit spreads more than 300bp

in regression (1)

• Larger drop in spreads for:

– Initially less transparent countries

– Countries with smaller debt markets

Page 5: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

5

Theory

Why does transparency affect borrowing costs?• Two models of asymmetric info:

– 1) Moral hazard (behavior of debtor)

– 2) Signaling (nature of random shock)

Page 6: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

6

Theory

Moral Hazard• Transparency increases => reduces asymmetric

information => less difference between informed and uninformed traders increases liquidity and reduces borrowing cost decreases (always)

Signaling• Good shock => voluntary transparency => borrowing cost

decreases

Page 7: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

7

News Results

News Effect of IMF publication• Higher volatility of daily spreads associated with IMF

publication– IMF publication => more informed markets

– But IMF publication also has information about economic outlook. When non-borrowing countries are examined, volatility also increases

Page 8: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

8

Details

Data:• Country Credit Spread = EMBI portfolio yield –

US zero coupon yield• Number of countries: 23 emerging markets• Sample period: 1 January 1999 – 30 June 2002

Page 9: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

9

Estimation

Panel estimation: (p.14)• ln(spread)it = i + t qt + 1 Pubit + 2 ROSCit + 3 SDDSit +

4 Pubit *ROSCit + 5 Pubit * SDDSit + 6 SDDSit * ROSCit + it

• Country effects• Time effects• Transparency dummies

– Pub, ROSC, SDDS

• Interaction terms– Marginal benefit of transparency – increasing/decreasing?

• Std Errors – corrected with Newey West

Page 10: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

10

Estimation

Panel estimation: (p.14)• Other variables:

– Kauffman et al’s rule of law, Transparency International’s Corruption Perception Index

– Size of debt market– Macro variables (inflation, current account/GDP, fiscal

balance/GDP)– PIN (Public Info Notice)

Page 11: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

11

Estimation

Panel estimation: (p.14)• Model (6)

– 16 variables, 23 country effects, 14 time effects– 322 observations

Page 12: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

12

Endogeneity Problem

Adopting the IMF disclosure is a government decision. The decision could be strategic and related to growth opportunities in the economy.

• Hence, any relation between spreads and publishing could be spuriously induced as a result of severe endogeneity

Page 13: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

13

Endogeneity Problem

Author’s approach:

“Any endogeneity bias is corrected for in two ways..”

1. Exclude program countries

2. 2SLS

Page 14: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

14

Endogeneity Problem

2SLS

First stage:(i) Run OLS regression of Pub dummy (e.g.) on:

average time between Article IV discussions, interacted with regional dummies, GDP per capita in 1998, size of debt market, rule of law, voice, corruption, squares of some of the variables, regional dummies

Page 15: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

15

Endogeneity Problem

2SLS

First stage:(ii) Regression (1) has 322 observations,

40 variables and 85% R-square.

(iii) Full 2SLS similar to OLS

Page 16: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

16

Endogeneity Problem

2SLS issues

1. Fitted values lie outside the 0-1 range2. Is it really the case that the 45 variables are uncorrelated

with spreads? “…the precise timing of compliance depends more on the time since the country committed to meet the specifications of the SDDS than on concurrent events.”

3. With enough variables in the first, we would get a 100% R-square. It is no surprise that the results are similar.

Page 17: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

17

Endogeneity ProblemStage 1 Instrument Regression

ln(spread)fop 1.08 sop*ldebt2 -2.83 ***fop*europe 0.34 * sop*lgdp2 16.91 ***fop*latin 0.04 sop*law2 9.83 ***fop*middleeast 0.11 sop*voice2 -2.40 ***fop*ldebt -0.12 sop*corrupt2 34.02 ***fop*law -0.72 *** tsub -0.09 *fop*voice 0.28 tsub*asia -0.05 ***fop*corrupt 0.81 ** tsub*europe -0.06 ***sop 678.13 *** tsub*law 0.04 ***sop*europe -29.06 *** tsub*ldebt 0.01sop*asia -31.67 *** tsub*voice 0.00sop*latin -13.81 *** tsub*corruption -0.09 ***sop*middleeast -15.57 ***sop*law 28.52 *** Quarterly dummies yessop*lgdp -266.05 *** Obs 308sop*ldebt 65.59 *** R-squared 0.54sop*corrupt -35.69 ***sop*voice 12.11 ***

Page 18: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

18

Endogeneity Problem

What to do?

• Problem very similar to Bekaert, Harvey and Lundblad, “Does Financial Liberalization Spur Growth” forthcoming JFE.

• Here regressions of economic growth are run on a liberalization dummy – suffers from same type of endogeneity bias.

Page 19: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

19

Endogeneity Problem

Develop a measure of exogenous growth opportunities

• Assume that a country’s growth opportunities are related to its industrial mix

• Assume that global price to earnings (PE) ratios contain information about growth opportunities

Page 20: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

20

Endogeneity Problem

Develop a measure of exogenous growth opportunities

• Create growth opportunities measure by weighting global PE ratios by a particular country’s industrial weights

• This variable strongly predicts economic growth – but is exogenous.

• This variable has both cross-sectional and time-series variation

Page 21: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

21

Endogeneity Problem

Develop a measure of exogenous growth opportunities

• Suggest adding to the regression to control for growth opportunities

• For more details, see Bekaert, Harvey, Lundblad, Siegel, 2004, “Growth Opportunities and Market Integration”

Page 22: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

22

Specification Issues

1. Much of the significance comes from the interaction of the transparency indicators (which is not reported in the paper)

2. Much of the explanatory power comes from the country fixed effects (which is not reported in the paper). The time effects are not important.

Page 23: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

23

Specification Issues

3. What happens when other measures of country risk are included in the regression?

Page 24: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

24

Specification Issues Regressions of Spreads on Transparency Measures, Interaction Terms and ICRG risk ratings

ln(spread) ln(spread) ln(spread) ln(spread) ln(spread) ln(spread) ln(spread)

pub -0.09 -0.25 ** -0.22 *** -0.30 *** -0.18 *** -0.26 ***(0.07) (0.10) (0.08) (0.08) (0.08) (0.09)

rosc -0.03 -0.21 -0.12 -0.14 -0.26 *** -0.28 **(0.06) (0.17) (0.10) (0.14) (0.12) (0.14)

sdds -0.12 * -0.22 *** -0.21 *** -0.19 *** -0.26 *** -0.21 ***(0.07) (0.07) (0.06) (0.07) (0.06) (0.06)

pub*sdds 0.30 *** 0.23 *** 0.34 *** 0.11 0.19 *(0.11) (0.09) (0.10) (0.10) (0.10)

pub*rosc 0.12 0.03 0.14 0.20 0.25(0.17) (0.13) (0.15) (0.15) (0.17)

sdds*rosc 0.15 0.07 0.09 0.16 0.20 *(0.14) (0.10) (0.11) (0.11) (0.11)

composite -0.07 ***(0.01)

political -0.05 ***(0.01)

finance -0.08 ***(0.01)

economic -0.05 ***(0.01)

country dum Yes Yes Yes Yes Yes Yes Yestime dum Yes Yes Yes Yes Yes Yes YesObs 322 322 322 322 322 322 322R-squared 0.84 0.85 0.85 0.91 0.89 0.90 0.88

Page 25: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

25

Specification Issues

Argentina Bulgaria Colombia Cote D'Ivoire Croatia Mexico Morocco768 0 854 0 604 0 1,023 0 543 0 693 0 743 0685 0 896 0 559 0 1,032 0 650 0 565 0 688 0804 0 903 0 675 0 1,084 0 747 0 624 0 688 0623 0 696 0 524 0 1,218 0 562 0 466 0 488 0551 0 620 0 479 1 1,425 0 336 1 361 0 391 0645 0 775 1 725 1 1,691 0 433 1 402 0 499 0669 0 692 1 701 1 1,785 0 376 1 334 0 430 0808 0 826 1 773 1 2,189 0 372 1 367 0 556 0755 1 734 1 666 1 2,712 0 254 1 396 0 507 0983 1 732 1 601 1 2,175 0 226 1 342 0 484 0

1,489 1 672 1 558 1 2,216 0 200 1 355 0 453 02,972 1 568 1 555 1 2,349 1 211 1 360 1 578 14,808 1 432 1 554 1 2,069 1 167 1 278 1 413 15,467 1 369 1 553 1 1,837 1 133 1 266 1 383 1

- + - + - + +

Difficult to see changes in the data. Here is the “pub” variable.

Spread change

Page 26: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

26

Specification Issues

Nigeria Panama Peru Poland Russia1,128 0 459 0 664 0 259 0 5,014 01,094 0 430 0 567 0 246 0 3,909 01,232 0 511 0 640 0 268 0 3,273 01,151 0 462 0 541 0 241 0 3,207 01,170 0 384 0 446 0 225 0 2,112 02,079 0 444 0 553 0 274 1 1,204 02,151 0 425 0 545 0 265 1 932 01,992 0 485 0 726 0 252 1 1,109 11,654 0 468 0 649 0 220 1 1,064 11,339 0 416 1 729 1 179 1 952 11,346 1 412 1 635 1 210 1 902 11,253 1 452 1 592 1 226 1 818 1

997 1 375 1 458 1 187 1 569 1999 1 387 1 501 1 179 1 482 1

+ - + + +

Spread change

Page 27: Is Transparency Good For You?  by  Rachel Glennerster, Yongseok Shin

27

Conclusions

• I believe the story (transparency leads to lower spreads)

• There are some issues that need to be cleaned up