## wp17200

## Source details

**Canonical URL:** [wp17200](https://www.imf.org/-/media/files/publications/wp/2017/wp17200.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2017/wp17200.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2017/wp17200.pdf.json)

---

### I. INTRODUCTION
- Large restructuring of financial sector balance sheets began with the onset of the global financial crisis in 2007.
- Prior mechanisms and literature cited:
  - Pro-cyclicality of leverage among investment banks (Adrian and Shin 2010).
  - Fire-sale mechanism reducing liquidation value when best users are credit constrained (Shleifer and Vishny 1992).
  - Securitized assets shifted across institutions during the crisis: hedge funds reduced holdings while commercial banks increased holdings and leverage (He, Khang, and Krishnamurthy 2010).
  - Theoretical frameworks referenced: Brunnermeier and Pederson (2009), Krishnamurthy (2010), Fostel and Geanakoplos (2008), Allen and Gale (1998, 2005), Acharya and Viswanathan (2010).
- Core empirical focus:
  - How buyer capital position (capital to assets ratio) affects both decision to purchase and transaction value for real assets (property, loan portfolios, branches, other assets).
  - Sectoral differences across financial sub-sectors: deposit-taking institutions, investment banks, broker-dealers, hedge-funds, real estate and insurance companies.

### II. ACCESS TO FUNDING WITHIN THE FINANCIAL SECTOR
- Key institutional funding differences and policy measures:
  - Commercial and savings banks raise (partially) insured deposits; FDIC coverage limit increased from $100,000 to $250,000 in 2008.
  - Temporary Liquidity Guarantee Program (TLGP): October 2008 to December 2010; allowed deposit-taking institutions to issue senior unsecured debt up to three years with FDIC insurance for a fee of 25 to 50 basis points.
  - Fed cut the discount rate for commercial banks several times beginning in August 2007.
  - Investment banks allowed to borrow from the discount window in March of 2008 using a broad range of debt securities as collateral.
  - Hedge funds and broker-dealers lacked government support and rely mostly on repo financing.
- Implication: deposit-taking institutions had greater access to, or a lower cost of, funding, followed by investment banks, then hedge-funds and broker-dealers.

### III. MODELING THE DETERMINANTS OF ASSET TRANSACTION VALUES
- Data and sample:
  - Main dataset: Thomson Reuters SDC Platinum M&A database; assets include real estate portfolios, loan portfolios, bank branches/units, equity investment portfolios, asset-backed securities, information technology systems.
  - Deals analyzed: transactions between financial institutions located in the United States between 2005 and 2011 where buyer is a publicly-traded company.
  - Universe of potential buyers: publicly-traded financial firms in United States with SIC code beginning with digit 6 — sample of 1,116 potential buyers; merged to deals database yielding 402 deals, 402 sellers, and 183 unique buyers.
  - Approximately 85% of assets sold by U.S. institutions to other financial institutions in this period are to other U.S. institutions; 6% to Australian and Canadian firms; remaining to 13 other countries.
- Hypotheses:
  - Main hypothesis: positive relationship between buyer capital to assets ratio and value of asset sale (deal value).
  - Secondary hypothesis: relationship stronger for non-deposit-taking institutions because leverage constraints matter more for them.
- Empirical approach:
  - Two-stage Heckman (1979) selection model to account for sample selection bias.
  - Selection (first-stage) Probit: probability buyer purchases an asset in a given year; main regressors include buyer capital/assets ratio, log(assets), assetgrowth (% change in total assets previous year), markettobook (market assets to book assets), plus year dummies.
  - Main (second-stage) equation: log(deal value) in millions of US dollars as dependent variable; main regressor buyer capital/assets; controls include same city, same state, same subsector indicators, US stock market return over the month prior to announcement, and inverse Mills ratio from selection equation.
  - Notation: i = seller, j = buyer, t = year; E_jt = 1 if institution j buys an asset in year t.

### IV. RESULTS — DESCRIPTIVE STATISTICS
- Sample counts:
  - Uncensored 5,823 observations; Censored (deal) 402 observations.
- Selected statistics (preserved exactly):
  - Deal value, millions of US dollars: Mean 351, Std Dev 1,551
  - Buyer capital/assets: Uncensored Mean 31.3, Std Dev 26.5; Censored Mean 71.0, Std Dev 31.1
  - Buyer assets, millions of US dollars: Uncensored Mean 20,600, Std Dev 145,000; Censored Mean 67,700, Std Dev 295,000
  - Asset growth (%), previous year: Uncensored Mean 10.6, Std Dev 22.6; Censored Mean 17.5, Std Dev 27.1
  - Buyer market assets to book assets: Uncensored Mean 0.56, Std Dev 1.52; Censored Mean 1.08, Std Dev 0.73
  - US stock market return, previous month: Uncensored Mean 0.54, Std Dev 4.61
  - Same city indicator: Uncensored Mean 0.04, Std Dev 0.20; Same state: 0.28, Std Dev 0.45; Same sub-sector: 0.27, Std Dev 0.02

### IV. RESULTS — HECKMAN SELECTION MODEL (KEY ESTIMATES)
- Selection equation (probability of deal) — notable coefficients and marginal effects:
  - Buyer capital/assets coefficient: 0.026*** (Std Err 0.002)
  - Buyer log(assets): 0.178*** (0.014)
  - Buyer asset growth: 0.002* (0.001)
  - Marginal effects reported:
    - a 1 percentage point increase in the capital ratio (from the mean) increases probability of a deal by 0.8 percent
    - a 1 percent increase in size increases probability by 0.05 percent
    - a 1 percentage point increase in asset growth increases probability by 0.06 percent
- Deal value equation (log of deal value) — notable coefficients:
  - Buyer capital/assets coefficient: 0.032*** (0.008) — interpreted as a 1 percentage point increase in capital ratio associated with a 3.2 percent increase in deal value.
  - Buyer log(assets): 0.695*** (0.062) — a 1 percent increase in buyer size associated with a 0.7 percent increase in deal value.
  - Same sub-sector indicator coefficient: 0.872*** (0.199) — deals between firms in the same financial sector priced higher by US$ 2.4 million on average.
  - lambda (selection correction) = 0.778** (0.337) indicating positive correlation of unobservables and validating selection correction.

### IV. RESULTS — SECTORAL DIFFERENTIATION
- Deposit-taking versus non-deposit-taking (Table 3):
  - Buyer capital/assets main effect: -0.014* (0.008) in probability equation; -0.005 (0.014) in deal value equation.
  - Interaction Buyer capital/assets * non-deposit taking: 0.042*** (0.009) in probability; 0.033*** (0.011) in deal value.
  - Interpretation: positive relationship between buyer capital and both likelihood of buying and deal value driven by non-deposit-taking institutions.
- Sub-sector decomposition (Table 4) — probability equation interactions:
  - Base buyer capital/assets: -0.009 (0.007) probability; 0.001 (0.018) deal value (deposit-taking banks omitted category).
  - Interactions (probability equation):
    - * Inv bank & other credit: 0.016** (0.007)
    - * Hedge fund & broker-dealers: 0.040*** (0.007)
    - * Insurance & real estate: 0.023*** (0.007)
  - Interpretation: significant and positive relationship between capital and deal probability for each non-deposit-taking sub-sector.
  - Heterogeneous size effects on deal value: example interaction * Inv bank & other credit buyer log(assets) * 0.138*** (0.053).

### IV. RESULTS — ROBUSTNESS CHECKS
- Controlling for asset type (Table 5) — subset of 252 coded deals (88 percent properties, 7 percent loan portfolios, 5 percent bank units/branches):
  - Buyer capital/assets: 0.036*** (0.002) in selection; 0.066*** (0.015) in deal value.
  - Asset type: loan portfolio coefficient 0.810* (0.442) in log(deal value) — indicates loan portfolios sell for US$ 2.24 million more than properties on average.
  - lambda = 1.193*** (0.426)
- U.S. sales to buyers in all countries (OLS, Table 6) — 591 observations:
  - Buyer capital/assets: 0.015*** (0.003) in both specifications.
  - Buyer country in crisis coefficient: -1.956* (1.016) in column (1).
  - Interaction Buyer capital/assets * buyer country crisis: -0.007 (0.006) — no significant differential effect of capital when buyer is from a crisis country.
  - Buyer log(assets): 0.563*** (0.039); interaction Buyer log(assets) * buyer country crisis: 0.328*** (0.083) — buyer size effect larger when buyer from crisis country.
  - Adjusted R-squared: 0.36 (col 1), 0.38 (col 2).

### IV. RESULTS — OVERALL EMPIRICAL CONCLUSION
- Greater capital (lower leverage) increases probability a potential buyer will purchase an asset and increases deal value for most financial firms.
- Effects do not hold for deposit-taking institutions, consistent with their greater access to or cheaper funding during the crisis.

### V. POLICY-RELEVANT INSIGHTS AND INTERPRETATION
- Empirical evidence supports theories where asset demand depends on funding availability and equity capital (leverage constraints).
- Implication for systemic liquidity and asset prices:
  - Credit-constraints of financial intermediaries can interact to reduce asset prices and deepen liquidity crises via fire-sale externalities.
- Implication for government liquidity policies:
  - Government liquidity policies implemented to encourage commercial banks to lend to the real sector may have had the unintended effect of facilitating an accumulation of non-loan assets by deposit-taking institutions (i.e., easing funding for deposit-takers changed cross-institutional asset demand dynamics).

### REFERENCES (selected themes represented in source)
- Acharya, Viral V., and S. Viswanathan, 2011, “Leverage, Moral Hazard, and Liquidity,” Journal of Finance, Vol. 66 (1), pp. 99–138.
- Adrian, Tobias, and Hyun Song Shin, 2010, “Liquidity and Leverage,” Journal of Financial Intermediation, Vol. 19 (3), pp. 418–37.
- Brunnermeier, Markus, and Lesse Pedersen, 2009, “Market Liquidity and Funding Liquidity,” Review of Financial Studies, 22(6), pp. 2201–38.
- He, Zhiguo; In Gu Khang and Arvind Krishnamurthy, 2010, “Balance Sheet Adjustment in the 2008 Crisis,” IMF Economic Review, Vol. 1, pp. 118–156.
- Shleifer, Andrei, and Robert Vishny, 1992, “Liquidation Values and Debt Capacity: A Market Equilibrium Approach,” Journal of Finance, Vol. 47 (4), pp. 1343–66.
- Heckman, James J.,1979, “Sample Selection Bias as a Specification Error,” Econometrica, Vol. 47(1), pp. 153–61.

*Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17200.pdf*

### 1.   Descriptive   Statistics   ........................................................................................

### 1.   Descriptive   Statistics   ............................................................................................................9

### Table of contents — major sections
- 1.   Descriptive   Statistics   ............................................................................................................9
- 2.   Determinants of Deal Probability and Deal Value ............................................................10
- 3.   Determinants of Deal Probability and Deal Value—Deposit Taking Institutions Versus Other Financial Institutions................................................................................................11
- 4.   Determinants of Deal Probability and Deal Value—Sectoral Decomposition ..................12
- 5.   Determinants of Deal Value—Controlling for Type of Asset ...........................................13
- 6.   Determinants of Deal Value—U.S. Sales to Buyers in All Countries ...............................14

*Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17200.pdf*

### References .............................................................................................................

### wp17200 - References .............................................................................................................

### I. INTRODUCTION
- Large restructuring of financial sector balance sheets began with the onset of the global financial crisis in 2007.
- Prior literature and mechanisms referenced:
  - Pro-cyclicality of leverage among investment banks (Adrian and Shin 2010).
  - Fire-sale mechanism reducing liquidation value when best users are credit constrained (Shleifer and Vishny 1992).
  - Securitized assets shifted across institutions during the crisis: hedge funds reduced holdings while commercial banks increased holdings and leverage (He, Khang, and Krishnamurthy 2010).
  - Theoretical frameworks: Brunnermeier and Pederson (2009), Krishnamurthy (2010), Fostel and Geanakoplos (2008), Allen and Gale (1998, 2005), Acharya and Viswanathan (2010).
- Core empirical focus:
  - How buyer capital position (capital to assets ratio) affects both decision to purchase and transaction value for real assets (property, loan portfolios, branches, other assets).
  - Sectoral differences across financial sub-sectors: deposit-taking institutions, investment banks, broker-dealers, hedge-funds, real estate and insurance companies.

### II. ACCESS TO FUNDING WITHIN THE FINANCIAL SECTOR
- Key institutional funding differences and policy measures:
  - Commercial and savings banks raise (partially) insured deposits; FDIC coverage limit increased from $100,000 to $250,000 in 2008.
  - Temporary Liquidity Guarantee Program (TLGP): October 2008 to December 2010; allowed deposit-taking institutions to issue senior unsecured debt up to three years with FDIC insurance for a fee of 25 to 50 basis points.
  - Fed cut the discount rate for commercial banks several times beginning in August 2007.
  - Investment banks allowed to borrow from the discount window in March of 2008 using a broad range of debt securities as collateral.
  - Hedge funds and broker-dealers lacked government support and rely mostly on repo financing.
- Implication: deposit-taking institutions had greater access to, or a lower cost of, funding, followed by investment banks, then hedge-funds and broker-dealers.

### III. MODELING THE DETERMINANTS OF ASSET TRANSACTION VALUES
- Data and sample:
  - Main dataset: Thomson Reuters SDC Platinum M&A database; assets include real estate portfolios, loan portfolios, bank branches/units, equity investment portfolios, asset-backed securities, information technology systems.
  - Deals analyzed: transactions between financial institutions located in the United States between 2005 and 2011 where buyer is a publicly-traded company.
  - Universe of potential buyers: publicly-traded financial firms in United States with SIC code beginning with digit 6 — sample of 1,116 potential buyers; merged to deals database yielding 402 deals, 402 sellers, and 183 unique buyers.
  - Approximately 85% of assets sold by U.S. institutions to other financial institutions in this period are to other U.S. institutions; 6% to Australian and Canadian firms; remaining to 13 other countries.
- Hypotheses:
  - Main hypothesis: positive relationship between buyer capital to assets ratio and value of asset sale (deal value).
  - Secondary hypothesis: relationship stronger for non-deposit-taking institutions because leverage constraints matter more for them.
- Empirical approach:
  - Two-stage Heckman (1979) selection model to account for sample selection bias.
  - Selection (first-stage) Probit: probability buyer purchases an asset in a given year; main regressors include buyer capital/assets ratio, log(assets), assetgrowth (% change in total assets previous year), markettobook (market assets to book assets), plus year dummies.
  - Main (second-stage) equation: log(deal value) in millions of US dollars as dependent variable; main regressor buyer capital/assets; controls include same city, same state, same subsector indicators, US stock market return over the month prior to announcement, and inverse Mills ratio from selection equation.
  - Notation: i = seller, j = buyer, t = year; E_jt = 1 if institution j buys an asset in year t.

### IV. RESULTS
- Descriptive statistics (Uncensored 5,823 obs; Censored: deal 402 obs) — selected figures preserved exactly:
  - Deal value, millions of US dollars: Mean 351, Std Dev 1,551
  - Buyer capital/assets: Uncensored Mean 31.3, Std Dev 26.5; Censored Mean 71.0, Std Dev 31.1
  - Buyer assets, millions of US dollars: Uncensored Mean 20,600, Std Dev 145,000; Censored Mean 67,700, Std Dev 295,000
  - Asset growth (%), previous year: Uncensored Mean 10.6, Std Dev 22.6; Censored Mean 17.5, Std Dev 27.1
  - Buyer market assets to book assets: Uncensored Mean 0.56, Std Dev 1.52; Censored Mean 1.08, Std Dev 0.73
  - US stock market return, previous month: Uncensored Mean 0.54, Std Dev 4.61
  - Same city indicator: Uncensored Mean 0.04, Std Dev 0.20; Same state: 0.28, Std Dev 0.45; Same sub-sector: 0.27, Std Dev 0.02
- Heckman selection model (Table 2) — key coefficient estimates and marginal effects:
  - Selection equation (probability of deal):
    - Buyer capital/assets coefficient: 0.026*** (Std Err 0.002)
    - Buyer log(assets): 0.178*** (0.014)
    - Buyer asset growth: 0.002* (0.001)
    - Marginal effects reported: a 1 percentage point increase in the capital ratio (from the mean) increases probability of a deal by 0.8 percent; a 1 percent increase in size increases probability by 0.05 percent; a 1 percentage point increase in asset growth increases probability by 0.06 percent.
  - Deal value equation (log of deal value):
    - Buyer capital/assets coefficient: 0.032*** (0.008) — interpreted as a 1 percentage point increase in capital ratio associated with a 3.2 percent increase in deal value.
    - Buyer log(assets): 0.695*** (0.062) — a 1 percent increase in buyer size associated with a 0.7 percent increase in deal value.
    - Same sub-sector indicator coefficient: 0.872*** (0.199) — deals between firms in the same financial sector priced higher by US$ 2.4 million on average.
    - lambda (selection correction) = 0.778** (0.337) indicating positive correlation of unobservables and validating selection correction.
- Sectoral differentiation (Table 3 and Table 4):
  - Deposit-taking vs non-deposit-taking (Table 3):
    - Buyer capital/assets main effect: -0.014* (0.008) in probability equation; -0.005 (0.014) in deal value equation.
    - Interaction Buyer capital/assets * non-deposit taking: 0.042*** (0.009) in probability; 0.033*** (0.011) in deal value.
    - Interpretation: positive relationship between buyer capital and both likelihood of buying and deal value driven by non-deposit-taking institutions.
  - Sub-sector decomposition (Table 4):
    - Base buyer capital/assets: -0.009 (0.007) probability; 0.001 (0.018) deal value (deposit-taking banks omitted category).
    - Interactions (probability equation):
      - * Inv bank & other credit: 0.016** (0.007)
      - * Hedge fund & broker-dealers: 0.040*** (0.007)
      - * Insurance & real estate: 0.023*** (0.007)
    - Interpretation: significant and positive relationship between capital and deal probability for each non-deposit-taking sub-sector.
    - Some interactions of buyer log(assets) with sub-sectors show heterogeneous size effects on deal value (e.g., * Inv bank & other credit buyer log(assets) * 0.138*** (0.053)).
- Robustness checks (Tables 5 and 6):
  - Controlling for asset type (Table 5): subset of 252 coded deals (88 percent properties, 7 percent loan portfolios, 5 percent bank units/branches).
    - Buyer capital/assets: 0.036*** (0.002) in selection; 0.066*** (0.015) in deal value.
    - Asset type: loan portfolio coefficient 0.810* (0.442) in log(deal value) — indicates loan portfolios sell for US$ 2.24 million more than properties on average (as noted in text).
    - lambda = 1.193*** (0.426)
  - U.S. sales to buyers in all countries (OLS, Table 6) — 591 observations:
    - Buyer capital/assets: 0.015*** (0.003) in both specifications.
    - Buyer country in crisis coefficient: -1.956* (1.016) in column (1).
    - Interaction Buyer capital/assets * buyer country crisis: -0.007 (0.006) — no significant differential effect of capital when buyer is from a crisis country.
    - Buyer log(assets): 0.563*** (0.039); interaction Buyer log(assets) * buyer country crisis: 0.328*** (0.083) — buyer size effect larger when buyer from crisis country.
    - Adjusted R-squared: 0.36 (col 1), 0.38 (col 2).
- Overall empirical conclusion:
  - Greater capital (lower leverage) increases probability a potential buyer will purchase an asset and increases deal value for most financial firms.
  - Effects do not hold for deposit-taking institutions, consistent with their greater access to or cheaper funding during the crisis.

### V. POLICY-RELEVANT INSIGHTS AND INTERPRETATION
- Empirical evidence supports theories where asset demand depends on funding availability and equity capital (leverage constraints).
- Implication for systemic liquidity and asset prices:
  - Credit-constraints of financial intermediaries can interact to reduce asset prices and deepen liquidity crises via fire-sale externalities.
- Implication for government liquidity policies:
  - Government liquidity policies implemented to encourage commercial banks to lend to the real sector may have had the unintended effect of facilitating an accumulation of non-loan assets by deposit-taking institutions (i.e., easing funding for deposit-takers changed cross-institutional asset demand dynamics).

*Source: wp17200 - References (IMF working paper content provided).*

### REFERENCES

### REFERENCES

### Liquidity, Leverage, and Banking
- Acharya, Viral V., and S. Viswanathan, 2011, “Leverage, Moral Hazard, and Liquidity,” Journal of Finance, Vol. 66 (1), pp. 99–138.
- Adrian, Tobias, and Hyun Song Shin, 2010, “Liquidity and Leverage,” Journal of Financial Intermediation, Vol. 19 (3), pp. 418–37.
- Brunnermeier, Markus, and Lesse Pedersen, 2009, “Market Liquidity and Funding Liquidity,” Review of Financial Studies, 22(6), pp. 2201–38.
- He, Zhiguo; In Gu Khang and Arvind Krishnamurthy, 2010, “Balance Sheet Adjustment in the 2008 Crisis,” IMF Economic Review, Vol. 1, pp. 118–156.
- Greenwood, R., A. Landier, and D. Thesmar, 2015, “Vulnerable banks,” Journal of Financial Economics, Vol. 115 No. 3, pp. 471–485.
- Krishnamurthy, Arvind, 2010, “How Debt Markets Have Malfunctioned in the Crisis,” Journal of Economic Perspectives, Vol. 24, No. 1, Winter 2010, pp. 3–28.
- Allen, Franklin and Douglas Gale, 2005, “From Cash-in-the-Market Pricing to Financial Fragility,” Journal of the European Economic Association, Vol. 3 (2–3), pp. 535–546.
- Acharya, Viral V., Gujral, Irvind and Shin, Hyun Song, 2009, “Dividends and Bank Capital in the Financial Crisis of 2007-2009” (March 18, 2009). Available at SSRN: https://ssrn.com/abstract=1362299

### Fire Sales, Spillovers, and Systemic Risk
- Caballero, Ricardo J. and Alp Simsek, 2013, “Fire Sales in a Model of Complexity,” The Journal of Finance, Vol. 68, pp 2549-2587.
- Duarte, Fernando and Thomas M. Eisenbach, 2015, “Fire-Sale Spillovers and Systemic Risk,” Federal Reserve Bank of New York Staff Reports, no. 645, February 2015.
- Shleifer, Andrei, and Robert Vishny, 2011, “Fire Sales in Finance and Macroeconomics,” Journal of Economic Perspectives, Vol. 25(1), pp. 29–48.
- Shleifer, Andrei, and Robert W. Vishny, 1992, “Liquidation Values and Debt Capacity: A Market Equilibrium Approach,” Journal of Finance, Vol. 47 (4), pp. 1343–66.

### Leverage Cycles and Behavioral Models
- Fostel, Ana and John Geanakoplos, 2008, “Leverage Cycles and the Anxious Economy,” American Economic Review, Vol. 98 (4), pp. 1211–1244.
- Allen, Franklin and Douglas Gale, 1994, “Limited Market Participation and Volatility of Asset Prices,” American Economic Review, Vol. 84 (4), pp. 933–955.
- Allen, Franklin and Douglas Gale, 1998, “Optimal Financial Crises,” Journal of Finance, Vol. 53, No. 4, pp. 1245-1284.
- Kyle, A., and W. Xiong, 2001, “Contagion as a Wealth Effect,” Journal of Finance, Vol. 56, pp. 1401–1440.

### Corporate Finance, Capital Structure, and Debt
- Myers, S., 1977, “Determinants of Corporate Borrowing,” Journal of Financial Economics, Vol. 5, pp. 147-75.
- Myers, S., Majluf, N., 1984, “Corporate financing and investment decisions when firms have information that investors do not have,” Journal of Financial Economics, Vol. 13, pp. 187– 221.

### Econometric Methods
- Heckman, James J.,1979, “Sample Selection Bias as a Specification Error,” Econometrica, Vol. 47(1), pp. 153–61.

*Source: wp17200 - REFERENCES*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2017/wp17200.pdf_