## 3. State-Indistry Level Regressions—Benchmark (Schooling)

## Source details

**Canonical URL:** [3. State-Indistry Level Regressions—Benchmark (Schooling)](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08229.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08229.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08229.pdf.json)

---

### Introduction and research question
- Debates the shareholders’ versus stakeholders’ views of corporate governance, focusing on input providers as stakeholders (workers, creditors, suppliers).
- Research goal: assess whether regulatory interventions enhancing stakeholders’ rights can be justified on efficiency grounds.
- Aim: provide evidence on whether changes in legal protection for stakeholder classes affect firm and aggregate performance (does not resolve normative debate).

### Empirical setting and identification advantages
- Uses U.S. state-level variation in legal and regulatory changes over time (1972–1993) to identify effects.
- Key advantages:
  - Natural experiments: substantial state-level changes over the last four decades (bank deregulation, evolving worker protections).
  - Staggered timing across states reduces reverse causality concerns.
  - Joint examination of shareholders’, creditors’, and workers’ rights in one framework reduces omitted variable bias and enables interaction analysis.
  - High and uniform information, contracting, and enforcement across U.S. states support credibility and relevance for U.S.-like environments.

### Data, outcome measures, and methodology
- Outcome measures:
  - Growth of state-level GDP or state-industry level GDP (value added) for the period 1972 to 1993.
  - Growth rates deflated by national CPI.
  - Rationale: captures overall stakeholder contributions, general equilibrium effects, and avoids firm-selection bias.
- Key policy/regulatory variables:
  - Financial deregulation (FinLib): branch deregulation from Jayaratne and Strahan (1996); FinLib defined as M&A dummy or De novo dummy; M&A-based deregulation emphasized.
  - Equity rights (EquRight): anti-takeover index from Bebchuk and Cohen (2003), index runs from 0 to 6 (available 1986–2001).
  - Employment protection (WorkRight): binary indicators from Autor, Donohue III and Schwab (2006) for public policy, good faith, implied contract; earliest = any of the three.
- Industry characteristics:
  - External finance dependence (ExtFinDep): Rajan-Zingales (1998) style ratio.
  - Knowledge dependency (Schooling): average fraction of labor college educated for years 1940 and 2005 (Buera and Kaboski, 2006); alternatives: intangible-asset-to-fixed-asset ratio; sales-to-fixed-asset ratio; R&D-to-sales; staff-costs-to-sales.
- Samples and observations:
  - Sample period: 1972 to 1993.
  - State-level: 1,029 (or 1,008) observations; N Groups = 48.
  - State-industry: about 51,201 (or 50,276; 43,385 in some specs) observations; N Groups examples: 2,487; 2,440; 2,097; 2,092.
  - Delaware and South Dakota dropped; original coverage included all 50 US states and the District of Colombia.
- Identification strategy:
  - Modified Rajan-Zingales (1998) approach using benchmark industry measures to interact policy changes with industry characteristics.
  - State and year fixed effects (state-level); state-industry and year fixed effects (state-industry level).
  - Interactions: FinLib * ExtFinDep; WorkRight * KnowledgeDep (Schooling); double interactions for complementarities.
  - Robustness: dynamic panel GMM (Blundell and Bond AR(1) and AR(2) first-differenced models) treating policy variables as endogenous but predetermined.

### Main empirical findings (state and state-industry levels)
- State-level (gross state product; non-financial sector):
  - Large positive effect of bank branch deregulation (FinLib, M&A dummy):
    - State-level FinLib coefficients examples: 1.621, 1.785, 1.690, 1.575, 1.708 (t-statistics [2.701]***, [2.887]***, [2.809]***, [2.934]***, [2.836]***).
    - Non-financial sector FinLib coefficients examples: 2.089, 2.137, 2.078, 2.045, 2.096 (t-statistics [3.333]***, [3.358]***, [3.220]***, [3.060]***, [3.214]***).
  - Employment protection (WorkRight) effects ambiguous and generally insignificant at the aggregate state level.
  - Equity (anti-takeover) protection coefficients generally negative but never statistically significant in state-level regressions; inclusion reduces observations (~1000 to ~330).
- State-industry (heterogeneity by industry characteristics):
  - FinLib:
    - Positive and statistically significant overall: FinLib coefficients in benchmark state-industry regressions ≈ 1.118 to 1.228 with t-statistics indicating significance at 1% (examples: 1.118 [2.763]***, 1.222 [3.016]***, 1.192 [2.935]***).
    - Interaction FinLib * ExtFinDep:
      - Often insignificant or mixed: examples include -0.058 [-0.797], -0.062 [-0.846], and strongly negative when double interactions included (e.g., -0.276 [-2.698]***).
      - In some specifications (e.g., intangible-based measures) FinLib*ExtFinDep positive and significant (examples: 0.196 [3.619]***, 0.123 [4.598]***).
    - Interpretation: financial liberalization improves overall resource allocation efficiency rather than systematically favoring externally finance-dependent industries in benchmark schooling specs.
  - Employment protection (WorkRight):
    - Direct coefficients at state-industry level uniformly negative and statistically significant:
      - Examples: -1.904 [-2.983]***, -1.788 [-2.832]***, -2.529 [-2.818]*** (Table 3 examples).
      - Quantitative impact range: between -2.5 to -1.6 percent on growth rates (reported across specifications).
    - Interaction WorkRight * Schooling (knowledge intensity) consistently large positive and statistically significant:
      - Examples: 7.393 [5.767]***, 7.332 [5.762]***, 8.047 [6.457]*** in Table 3; GMM examples: 14.661 [8.464]***, 17.506 [9.381]*** (Table 6a).
      - Interpretation: employment protection increases relative growth in more knowledge-dependent industries (consistent with firm-specific human-capital channel).
    - WorkRight interactions with Intangible/Fixed Assets and Sales/Fixed Assets positive and significant (Tables 4 and 5).
  - Complementarities (double interactions):
    - Product of (FinLib * ExtFinDep) and (WorkRight * KnowledgeDep) often positive and highly statistically significant in most specifications (columns 4, 7, 10 and 13), except with the good faith index.
    - Interpretation: simultaneous banking deregulation and stronger employment protection particularly boost growth of industries that are both externally finance dependent and knowledge intensive.
  - Magnitudes (selected examples from benchmark tables):
    - FinLib (state-industry bench): 1.118 [2.763]***; 1.222 [3.016]***; 1.192 [2.935]***.
    - WorkRight (bench): -1.904 [-2.983]***; -2.529 [-2.818]***; -1.661 [-2.461]**.
    - WorkRight*Schooling (bench): 7.393 [5.767]***; 5.283 [3.724]***; 6.719 [4.859]***.
    - GMM AR(1) FinLib: 2.079 [7.180]***; GMM AR(2) FinLib: 2.776 [8.817]***.
    - GMM WorkRight*Schooling: 14.661 [8.464]*** (AR(1)); 14.576 [8.267]*** (AR(2)).

### Robustness checks and sensitivity analyses (selected exact findings)
- Alternative knowledge-intensity measures:
  - Intangible-to-fixed-asset and sales-to-fixed-asset measures confirm main results: WorkRight interaction with knowledge intensity positive and highly significant; FinLib positive.
  - R&D-to-sales and staff-costs-to-sales broadly confirm patterns at lower significance.
- Threshold effects on human capital:
  - Positive effect of WorkRight arises only for industries in top education quintiles (1st or 2nd quintile of college-educated ratios).
- Union membership and minimum wage:
  - Adding industry unionization rates (1983–1991 average) does not qualitatively change results; union coefficients insignificant.
  - Replacing state minimum wage with federal where applicable and adding real minimum wage yields no significant effects and does not change main coefficients.
- Interstate banking timing and sample periods:
  - Results robust when using sample end points 1990 and 1997.
- Reverse causality and endogeneity:
  - Excluding potentially endogenous reform adopters (Massachusetts, California) does not change results.
  - Dynamic panel GMM (Blundell-Bond AR(1) and AR(2)) largely confirms main results with somewhat larger magnitudes.
  - GMM diagnostics: Hansen-Sargan p-value = 0.000 (reported); m1 p-value = 0.000; m2 p-value examples: 0.378 (AR(1)), 0.247 (AR(2)). Hansen-Sargan rejection suggests overidentifying restrictions not valid; however, core conclusions robust across methods.

### Interpretation and theoretical implications
- Evidence consistent with incomplete-contract theories: firm performance and growth depend on relative bargaining powers of multiple stakeholders (workers, creditors, shareholders).
- Employment protection can generate efficiency gains in knowledge- and skill-intensive industries through incentives for firm-specific human-capital accumulation and better monitoring, despite average negative direct effects at industry level.
- Financial liberalization results imply overly strong stakeholder rights or close bank–borrower relationships prior to deregulation can hurt output growth; deregulation improves allocation and aggregate growth.
- Results highlight omitted variable bias risk in literature that studies creditor or shareholder protection in isolation.

### Policy-relevant conclusions and recommendations
- Financial liberalization (bank branch M&A deregulation) is associated with positive effects on state and state-industry value-added growth (examples: ≈ 1.1–1.6 percent in benchmark FE; up to 2.776 in GMM AR(2)).
- Enhancing employment protection can yield positive efficiency effects in knowledge- and skill-intensive industries (WorkRight*Schooling positive and significant), supporting targeted stakeholder-protecting policies rather than across-the-board reversals.
- Equity (anti-takeover) protection changes appear to have limited direct impact on state or state-industry value added in the studied period and context.
- Policy design should consider complementarities across financial, labor, and shareholder regulations to maximize aggregate and industry-specific performance; simultaneous banking deregulation and stronger employment protection particularly benefit industries that are both externally finance dependent and knowledge intensive.

### Key statistics and diagnostics (selected exact values)
- Sample period for growth measures: 1972 to 1993.
- State-level observations: N = 1,029 (or 1,008); N Groups = 48.
- State-industry observations examples: N = 51,201; 50,276; 41,249; 39,134. N Groups examples: 2,487; 2,440; 2,097; 2,092.
- Typical R-squared ranges: 0.0924–0.0960 (state-industry FE specifications); 0.0943–0.0951 (intangible/sales specifications); 0.256–0.380 (state-level).
- Selected means and standard deviations (Table 1c):
  - Value Added Growth (State level) Mean: 2.45, Std. Dev.: 4.31.
  - Value Added Growth Mean: 2.37, Std. Dev.: 19.17.
  - M&A Mean: 0.51, Std. Dev.: 0.32.
  - De novo Mean: 0.39, Std. Dev.: 0.22.
  - Anti-Takeover Mean: 1.97, Std. Dev.: 0.79.
  - Public Policy Mean: 0.45, Std. Dev.: 0.38.
  - Implied contract Mean: 0.42, Std. Dev.: 0.39.
  - Good faith Mean: 0.10, Std. Dev.: 0.10.
  - Earliest Mean: 0.56, Std. Dev.: 0.41.
  - RZIndex Mean: -0.49, Std. Dev.: 1.99.
  - Sales/Fixed assets Mean: 6.00, Std. Dev.: 5.66.
  - Intangible Assets/Fixed Assets Mean: 0.62, Std. Dev.: 0.60.
  - Schooling Average Mean: 0.29, Std. Dev.: 0.12.
- GMM diagnostics (examples):
  - Hansen-Sargan p-value = 0.000 (reported for GMM specifications).
  - m1 p-value = 0.000; m2 p-value examples: 0.378 (AR(1)), 0.247 (AR(2)).

*Source: _wp08229 - 3.  State-Indistry Level Regressions—Benchmark (Schooling).*

### 3.  State-Indistry Level Regressions—Benchmark (Schooling)...............................................48

### 3.  State-Indistry Level Regressions—Benchmark (Schooling)...............................................48

### Introduction and research question
- Debates the shareholders’ versus stakeholders’ views of corporate governance, focusing narrowly on input providers as stakeholders (workers, creditors, suppliers).
- Research goal: assess whether regulatory interventions enhancing stakeholders’ rights can be justified on efficiency grounds.
- Does not aim to resolve the normative debate fully but to contribute evidence on whether changes in legal protection for stakeholder classes affect firm and aggregate performance.

### Empirical setting and identification advantages
- Uses U.S. state-level variation in legal and regulatory changes over time to identify effects.
- Key advantages of the U.S. state context:
  - Natural experiments: substantial changes over the last four decades, including banks deregulated gradually and workers becoming more protected over time.
  - Changes occur at the state level and at different points in time, reducing reverse causality concerns from firm-level performance driving legal changes.
  - Ability to examine shareholders’, creditors’, and workers’ rights together in one consistent empirical framework, reducing omitted variable bias and enabling investigation of interactions among stakeholder rights.
  - High and uniform levels of information, contracting, and enforcement across states make identified effects credible and broadly relevant for U.S.-like institutional environments.

### Data, outcome measures, and methodology
- Outcome measures: growth of state-level GDP or state-industry level GDP (value added) for the period 1972 to 1993.
  - Rationale for using aggregate value added:
    - Captures overall contribution of all stakeholders.
    - Accounts for general equilibrium effects of reforms (e.g., redistribution without net change).
    - Avoids selection bias from analyzing only existing firms.
- Identification strategy: modified Rajan-Zingales (1998) methodology using benchmark industry measures of external financing dependency and knowledge-intensity (from U.S. benchmark data from a later period).
- Approach aims to avoid simultaneity biases and to identify channels by which judicial and regulatory changes affect firm performance.

### Main empirical findings
- Overall effects on state-level output growth:
  - Large positive effect of bank branch deregulation.
  - Ambiguous effects of changes in employment protection on state-level output growth.
  - Equity rights protection does not seem to affect state or state-industry level output growth.
- State-industry level findings:
  - Large positive effect of bank branch deregulation.
  - Negligible overall effect of employment protection on state-industry level output growth.
- Channels and heterogeneity:
  - Financial liberalization does not show effects varying with firms’ industry-specific external financing dependency; it improves overall resource allocation efficiency instead.
  - Judicial changes enhancing employment protection promote relatively more growth in more skilled-labor, knowledge- and intangible-asset intensive industries, while having little effect on overall output growth.
  - Positive coefficients on cross-terms between:
    - Bank branch deregulation for external financial dependent industries, and
    - Employment protection for skilled-labor industries.
  - Interpretation: complementarity in stakeholder roles—enhancing employment protection yields efficiency gains for knowledge-intensive industries even if general employment effects are typically considered negative.

### Interpretation and theoretical implications
- Evidence suggests that some enhancements to stakeholders’ rights (especially employment protection) can be justified on efficiency grounds for certain types of industries, consistent with a stakeholders’ view of the firm.
- Findings align with incomplete contract-based theories (e.g., Hart, 1995): firm performance and growth are influenced by the relative bargaining powers of multiple stakeholders.
- Financial liberalization results imply that overly strong stakeholder rights and overly close bank–borrower relationships prior to deregulation can hurt output growth.
- The results suggest omitted variable biases in previous literature that focused primarily on shareholder protection or creditor protection in isolation.

### Policy-relevant conclusions
- Bank branch deregulation has a large positive effect on state and state-industry output growth, pointing to efficiency gains from financial liberalization in the U.S. state context.
- Enhancing employment protection can yield positive efficiency effects in knowledge- and skill-intensive industries, supporting targeted stakeholder-protecting policies rather than blanket reversals of stakeholder-oriented regulations.
- Equity rights protection changes appear to have limited direct impact on state or state-industry value added in the studied period and context.
- Complementarities among stakeholder protections imply that policy design should consider interactions across financial, labor, and shareholder regulations to maximize aggregate and industry-specific performance.

*Source: _wp08229 - 3.  State-Indistry Level Regressions—Benchmark (Schooling)...............................................48*

### Section  3,  the  database  used  and  the  methodology  used  are  described.  Section  4  includes  the

### _wp08229 - Section 3, the database used and the methodology used are described. Section 4 includes the

### Related literature
- Financial liberalization (U.S. branch deregulation history):
  - Unit banking → intrastate banking (early 1970s onward) → 1994: federal interstate banking permission.
  - Jayaratne and Strahan (1996) find states that liberalized experienced faster growth; efficiency channels likely cause increases in growth.
- Employment protection (wrongful discharge categories):
  - Three categories: public policy; good faith; implied contract.
  - Theoretical channels ambiguous: may reduce hiring and output, raise compliance costs, or increase firm-specific human-capital accumulation and monitoring.
- Corporate governance / stakeholder interaction:
  - Literature often studies stakeholders separately; incomplete-contract theories (Allen 2005; Allen, Carletti and Marquez, 2007; Tirole, 2006) argue firm value varies with rights and bargaining powers of multiple stakeholders jointly.
- Empirical gaps:
  - Few studies jointly analyze creditor, shareholder, and worker rights; notable exceptions include Fonseca and Utrero (2007) and Atanassov and Kim (forthcoming).

### Data and empirical methodology
- Sample period and coverage:
  - Sample restricted to start in 1972 (bank deregulation data starts) and to 1993 (before interstate branch deregulation starts).
  - 1056 state-year balanced panel observations covering both financial liberalization and employment protection.
  - 376 state-year balanced panel observations covering three reforms.
  - About 1,000 state-level observations and about 50,000 state-industry level observations (after exclusions).
  - Delaware and South Dakota dropped; data cover all 50 US states and the District of Colombia originally.
- Key variables and sources:
  - Financial deregulation: branch deregulation from Jayaratne and Strahan (1996); FinLib defined as M&A dummy or De novo dummy.
  - Equity rights: anti-takeover index from Bebchuk and Cohen (2003), index runs from 0 to 6, available 1986-2001.
  - Employment protection: binary indicators from Autor, Donohue III and Schwab (2006) for public policy, good faith, implied contract; earliest = any of the three.
  - Dependent variables: value-added growth (state and state-industry) from BEA, growth rates adjusted for national CPI.
  - Industry characteristics:
    - External finance dependence: Rajan and Zingales (1998) style ratio.
    - Knowledge dependency: average fraction of labor college educated for years 1940 and 2005 (Buera and Kaboski, 2006); alternative measures: intangible-asset-to-fixed-asset ratio; sales-to-fixed-asset ratio; R&D-to-sales; staff-costs-to-sales.
    - Union membership: 1983-1991 average from Hirsh and MacPherson (2003).
- Correlations and descriptive stats:
  - Strong correlation among financial deregulation indexes: 0.78.
  - Employment protection correlations: contract vs public policy = 0.53; other two pairs = 0.29 and 0.14.
  - Correlations financial vs employment protection between 0.16 and 0.32.
  - Correlations between anti-takeover index and all indexes < 0.15.
  - Mean growth ~ 2.4 percentage points (Table 1c).
- Regression specifications:
  - State-level: state and year fixed effects; specification (1) includes lagged growth and FinLib, EquRight, WorkRight.
  - State-industry: state-industry and year fixed effects; benchmark specification (2) includes interactions:
    - FinLib * ExtFinDep
    - WorkRight * KnowledgeDep
    - Double interaction for complementarity
  - Robustness for endogeneity: dynamic panel GMM (Blundell and Bond AR(1) and AR(2) first-differenced models) treating policy variables as endogenous but predetermined.

### State-level regression findings
- Financial liberalization:
  - M&A deregulation dummy effect on state-wide growth ≈ 1.6 percent increase in state level growth rates (with state fixed effects).
  - Without state fixed effects effect ≈ 1.06 percent (almost identical to Jayaratne and Strahan (1996)).
  - De novo deregulation effects are lower and not as important (M&A always precedes de novo).
- Employment protection:
  - Mixed and generally insignificant effects across the three wrongful discharge categories:
    - Public policy protection shows a positive coefficient (not significant).
    - Good faith and implied contract show negative coefficients (not significant).
  - Using earliest adoption of any of the three also not significant.
- Equity (anti-takeover) protection:
  - Coefficients generally negative but never statistically significant in state-level regressions; number of observations drops from ~1000 to ~330 when included (data available from 1986).
- Non-financial sector:
  - Results hold identically when using non-financial sector state GDP growth; effect of financial liberalization becomes stronger; employment protection remains insignificant.

### State-industry benchmark regression findings (channels and heterogeneity)
- Financial liberalization:
  - Positive and statistically significant effect at state-industry level; magnitude ≈ 1.1 to 1.2 percent (smaller than state-level estimate).
  - Interaction FinLib * ExtFinDep: generally not statistically significant positive; when including double interaction, often statistically significant negative (except with good faith index).
  - Interpretation: financial liberalization spurs growth generally via better resource allocation rather than by favoring industries with higher external finance dependence.
- Employment protection:
  - Direct coefficients uniformly negative and statistically significant at state-industry level:
    - Quantitative impact between -2.5 to -1.6 percent on growth rates, largest impact for the good faith protection.
  - Interaction WorkRight * KnowledgeDep:
    - Consistently positive and statistically significant (at the 1% level, except for regressions using the good faith index).
    - Interpretation: greater employment protection increases growth in more knowledge-dependent industries—consistent with incentives for firm- or sector-specific human capital accumulation.
- Complementarity (double interactions):
  - Including the product of (FinLib * ExtFinDep) and (WorkRight * KnowledgeDep) shows positive and highly statistically significant coefficients in most specifications (columns 4, 7, 10 and 13), except with the good faith index.
  - Interpretation: combination of banking deregulation and stronger employment protection particularly boosts growth of industries with both high external financing dependence and high knowledge dependence.
- Robust comparisons:
  - Results robust when controlling for non-financial GDP, excluding key states, or using alternative knowledge-intensity proxies.

### Robustness checks and sensitivity analyses
- Alternative knowledge-intensity measures:
  - Intangible-to-fixed-asset ratio and sales-to-fixed-asset ratio confirm main results: employment protection interaction with knowledge intensity remains positive and highly significant; financial liberalization remains positive.
  - Firm-median R&D-to-sales and staff-costs-to-sales broadly confirm patterns but at lower significance levels.
- Threshold effects on human capital:
  - Positive effect of employment protection arises only for industries in top education quintiles (1st or 2nd quintile of college-educated ratios), consistent with firm-specific human-capital channel.
- Union membership:
  - Adding industry unionization rates (1983-1991 average) does not qualitatively change results; union membership coefficients insignificant.
- Minimum wage:
  - State minimum wage series replaced by federal minimum where federal is higher or state absent; adding real minimum wage (deflated by national CPI) yields no significant effects and does not change main coefficients.
- Interstate banking timing:
  - Re-running benchmarks with sample end points 1990 and 1997 yields qualitatively similar results.
- Reverse causality and endogeneity:
  - Excluding Massachusetts and California (potential endogenous reform adopters) does not change results.
  - Dynamic panel GMM (Blundell-Bond AR(1)/AR(2)) yields main results still holding and somewhat larger magnitudes.
  - Arrelano-Bond diagnostics: m2 test indicates no second-order autocorrelation; m1 indicates first-order autocorrelation (lagged dependent variables are good instruments).
  - Hansen-Sargan test rejects overidentifying restrictions (overidentifying restrictions not valid), suggesting GMM may be imperfect, but main conclusions robust across methods.

### Conclusions and policy-relevant implications
- Main empirical conclusions:
  - Financial liberalization (bank branch M&A deregulation) positively affects state and state-industry value-added growth (≈ 1.1–1.6 percent).
  - Employment protection has ambiguous aggregate effects: negative direct effects at the industry level (-2.5 to -1.6 percent) but positive, significant effects for knowledge-intensive industries via firm-specific human-capital incentives.
  - Financial liberalization primarily operates through improved resource allocation rather than preferentially boosting externally finance-dependent industries.
  - Complementarity: simultaneous banking deregulation and stronger employment protection especially benefit industries that are both externally finance dependent and knowledge intensive.
- Theoretical interpretation:
  - Findings align with incomplete-contract theories: optimal allocation of bargaining powers across stakeholders (workers, creditors, shareholders) can maximize firm value; relative powers matter.
  - The results contradict a perfect factor market assumption where employment protection would unambiguously lower output.
- Broader implications:
  - Evidence from the U.S. suggests stakeholder configurations matter for economic performance; results may inform debates in developed countries where banks and labor play significant governance roles, and in emerging markets with limited capital markets.
- Suggested directions for future research:
  - Cross-country analyses of creditor rights, minority shareholder rights, and employment protection on economic/industry outcomes.
  - Firm-level analyses to identify micro channels and to exploit better measures of knowledge-intensity.
  - Examination of other outcomes (volatility, employment, wages) in relation to stakeholder rights.

*Source: _wp08229 - Section 3, the database used and the methodology used are described. Section 4 includes the (IMF Working Paper content provided).*

### introduction of employment protection.

### _wp08229 - introduction of employment protection.

### Adoption patterns of financial deregulation and employment protection
- Figures show time series counts of states deregulating via M&A, branch restrictions lifted (de novo branching), and the adoption of employment protection (public policy exception, good faith exception, implied contract exception), covering years such as 1972, 1975, 1978, 1981, 1984, 1987, 1990, 1993, 1986, 1989, 1992, 1995, 1998, 2001. (Figure 1a, 1b, 1c)
- Comparative patterns between deregulation events and the adoption of employment protection are summarized as percentage distributions of differences between indicators: n (de novo branching), m (M&A), p (public policy), g (good faith), c (implied contract), and z (combination of p, g and c). (Figure 2a, 2b)

### Correlations among state-level institutional changes
- Table 1a reports pairwise correlations (with observation counts in italics; * denotes significance at 5%):
  - Anti-Takeover with M&A: 0.1078 (1078)
  - Anti-Takeover with de novo: 0.7828* (1078)
  - M&A with de novo: 0.7828* (1078)
  - Public Policy correlations: Anti-Takeover 0.2510* (1056), M&A 0.1573* (1056), de novo 0.1573* (1056)
  - Good Faith correlations: Anti-Takeover 0.1642* (1056), M&A 0.1590* (1056), de novo 0.1590* (1056), Public Policy 0.2916* (1056)
  - Contract correlations: Anti-Takeover 0.3243* (1056), M&A 0.2298* (1056), de novo 0.2298* (1056), Public Policy 0.5304* (1056), Good Faith 0.1411* (1056)
  - Earliest correlations: with Anti-Takeover 0.2764* (1056), with M&A 0.1721* (1056), with de novo 0.1721* (1056), with Public Policy 0.8160* (1056), with Good Faith 0.2907* (1056), with Contract 0.7599* (1056)
  - Anti-Takeover (separate block) correlation matrix elements include 0.1307* (384), 0.1144* (384), 0.025 (376), -0.0665 (376), -0.1414* (376), -0.1081* (376), 1.0000 (392)

### Industry-level characteristics: correlations and descriptive stats
- Table 1b (industry-level correlations; observation counts in italics; * denotes significance at 5%):
  - Ext. Fin. Dep. with Sales/Fixed Assets: -0.5236* (59)
  - Ext. Fin. Dep. with Intangible/Fixed Assets: -0.4360* (59)
  - Sales/Fixed Assets with Intangible/Fixed Assets: 0.7079* (59)
  - Intangible/Fixed Assets with Average Schooling: 0.5279* (49)
  - Sales/Fixed Assets with Average Schooling: 0.3718* (49)
  - Ext. Fin. Dep. with Average Schooling: -0.1039 (49)
- Table 1c (Averages and Standard Deviations of Main Variables; Mean and Std. Dev. shown):
  - Value Added Growth (State level) Mean: 2.45, Std. Dev.: 4.31
  - Value Added Growth Mean: 2.37, Std. Dev.: 19.17
  - M&A Mean: 0.51, Std. Dev.: 0.32
  - De novo Mean: 0.39, Std. Dev.: 0.22
  - Anti-Takeover Mean: 1.97, Std. Dev.: 0.79
  - Public Policy Mean: 0.45, Std. Dev.: 0.38
  - Implied contract Mean: 0.42, Std. Dev.: 0.39
  - Good faith Mean: 0.10, Std. Dev.: 0.10
  - Earliest Mean: 0.56, Std. Dev.: 0.41
  - RZIndex Mean: -0.49, Std. Dev.: 1.99
  - Sales/Fixed assets Mean: 6.00, Std. Dev.: 5.66
  - Intangible Assets/Fixed Assets Mean: 0.62, Std. Dev.: 0.60
  - Schooling Average Mean: 0.29, Std. Dev.: 0.12

### State-level regression findings (Gross State Product and Non-Financial Sector)
- Table 2a (State Level Regressions—Gross State Product; dependent variable: real growth rate of gross state product, deflated by national CPI, 1972 to 1993):
  - FinLib coefficients across models: 1.621, 1.785, 1.690, 1.575, 1.708 (t-statistics reported in brackets as [2.701]***, [2.887]***, [2.809]***, [2.934]***, [2.836]***)
  - WorkRight coefficients include 0.816, 0.950, -0.691, -0.643, -1.637, -1.347, -0.174, -0.011 (t-statistics [1.243], [1.459], [-0.984], [-0.916], [-0.889], [-0.767], [-0.267], [-0.017])
  - N = 1029 (or 1008 in some specifications), N Groups = 48, R-squared values reported (e.g., 0.375, 0.370, 0.380, 0.369, 0.379, 0.372, 0.380, 0.368, 0.377)
  - Explanatory variables: FinLib is binary M&A based bank branch deregulation; WorkRight is binary exception for wrongful discharges (public policy, good faith, implied contract, or earliest)
- Table 2b (State Level Regressions—Non-Financial Sector Gross State Product; dependent variable: non-financial sector growth):
  - FinLib coefficients across models: 2.089, 2.137, 2.078, 2.045, 2.096 (t-statistics [3.333]***, [3.358]***, [3.220]***, [3.060]***, [3.214]***)
  - WorkRight coefficients include 0.344, 0.504, -0.715, -0.656, -0.902, -0.525, -0.214, -0.014 (t-statistics [0.419], [0.637], [-0.957], [-0.874], [-0.626], [-0.373], [-0.324], [-0.021])
  - N = 1029 (or 1008), N Groups = 48, R-squared examples: 0.256, 0.254, 0.264, 0.255, 0.265

### State-industry level regressions: heterogeneous effects by industry characteristics
- Table 3 (Benchmark—Schooling; dependent variable: real growth rate of gross state-industry product, 1972 to 1993):
  - FinLib coefficients (various specifications): 1.118, 1.222, 1.152, 1.104, 1.1, 1.192, 1.121, 1.228, 1.161 (t-statistics [2.763]***, [3.016]***, [2.858]***, [2.843]***, [2.836]***, [2.935]***, [2.765]***, [3.056]***, [2.875]***)
  - Interaction FinLib*ExtFinDep coefficients include -0.058, -0.062, -0.276, -0.089, -0.098, -0.068, -0.274, -0.064, -0.278 (t-statistics [-0.797], [-0.846], [-2.698]***, [-1.224], [-1.326], [-0.923], [-2.819]***, [-0.879], [-2.422]**)
  - WorkRight coefficients: consistently negative and significant across specifications (examples: -1.904, -1.788, -1.928, -2.529, -2.312, -2.377, -1.661, -1.602, -1.739, -1.72, -1.58, -1.698 with t-statistics such as [-2.983]***, [-2.832]***, [-3.052]***, [-2.818]***, [-2.482]**, [-2.598]**, [-2.461]**, [-2.401]**, [-2.613]**, [-2.654]**, [-2.421]**, [-2.602]**)
  - WorkRight*Schooling coefficients are large positive: 7.393, 7.332, 8.047, 5.324, 5.31, 5.647, 5.283, 5.212, 5.943, 6.719, 6.665, 7.255 (t-statistics [5.767]***, [5.762]***, [6.457]***, [1.651], [1.665], [1.766]*, [3.724]***, [3.637]***, [4.160]***, [4.859]***, [4.818]***, [5.337]***)
  - N = 51,201 (or 43,385 in some specs), N Groups = 2,487 (or 2,107), R-squared around 0.0924–0.0960
  - Note: ExtFinDep defined per Rajan-Zingales (1998); Schooling = average fraction of college graduates in 1940 and 2005
- Table 4 (Intangible/Fixed Assets):
  - FinLib coefficients: 1.216, 1.158, 1.045, 1.038, 1.175, 1.122, 1.219, 1.168 (t-statistics [2.976]***, [2.849]***, [2.662]**, [2.646]**, [2.854]***, [2.741]***, [2.995]***, [2.871]***)
  - FinLib*ExtFinDep interaction coefficients positive and significant (examples: 0.196, 0.123, 0.177, 0.156 with t-statistics [3.619]***, [4.598]***, [4.323]***, [3.101]***)
  - WorkRight coefficients negative (e.g., -0.727 to -2.109) with t-statistics ranging from [-1.551] to [-2.372]**
  - WorkRight*Intangible/Fixed Assets positive and significant (e.g., 1.711 to 1.93 with t-statistics [5.882]*** to [6.618]***)
  - N = 50,276, N Groups = 2,440, R-squared ~0.0943–0.0949
- Table 5 (Sales/Fixed Assets):
  - FinLib coefficients: 1.234, 1.153, 1.046, 1.034, 1.192, 1.118, 1.238, 1.153 (t-statistics [3.016]***, [2.830]***, [2.659]**, [2.633]**, [2.897]***, [2.734]***, [3.042]***, [2.839]***)
  - FinLib*ExtFinDep interactions small positive but often insignificant (examples: 0.02, 0.016, 0.019, 0.019 with t-statistics [5.494]***, [4.547]***, [5.989]***, [5.225]*** — note: these bracketed t-statistics correspond to the reported significance in table cells)
  - WorkRight coefficients negative and often significant (e.g., -0.722 to -2.375 with t-statistics from [-1.577] to [-2.638]**)
  - WorkRight*Sales/Fixed Assets positive and significant (e.g., 0.175 to 0.271 with t-statistics [5.947]*** to [7.921]***)
  - N = 50,276, N Groups = 2,440, R-squared ~0.0943–0.0951

### Dynamic panel (Panel GMM) evidence: AR(1) and AR(2) specifications
- Table 6a (Panel GMM AR(1)):
  - L.Real GDP growth coefficients small and near zero (e.g., 0.007, 0.006, 0.005, 0.003, 0.002; t-statistics [1.220], [1.263], [1.044], [0.848], [0.619], [0.385], etc.)
  - FinLib coefficients robustly positive and significant: 2.079, 1.995, 1.714, 1.893, 1.661, 1.725, 1.835, 1.911 (t-statistics [7.180]***, [6.932]***, [6.072]***, [6.834]***, [5.804]***, [6.067]***, [6.291]***, [6.645]***)
  - WorkRight coefficients strongly negative and significant across all specifications (examples: -2.896 to -6.842; t-statistics [-4.820]*** to [-7.139]***)
  - WorkRight*Schooling coefficients large positive and highly significant (examples: 14.661 to 17.506; t-statistics [8.464]*** to [9.381]***)
  - FinLib*ExtFinDep interactions: 0.418, 0.451, 0.338, 0.044 with t-statistics [1.828]*, [2.503]**, [1.904]*, [0.231]
  - N = 41,249 (N Groups = 2,097). Hansen-Sargan p-value = 0.000; m1 p-value = 0.000; m2 p-value examples: 0.378, 0.389, 0.404, etc.
- Table 6b (Panel GMM AR(2)):
  - L.Real GDP growth and L2.Real GDP growth coefficients small; one L2 coefficient is -0.012 (t-statistic [-2.204]**)
  - FinLib coefficients even larger and significant: 2.776, 2.781, 2.571, 2.741, 2.349, 2.427, 2.65, 2.757 (t-statistics [8.817]*** to [9.044]***)
  - WorkRight coefficients negative and significant (e.g., -2.177 to -4.457; t-statistics [-3.579]*** to [-5.663]***)
  - WorkRight*Schooling again large positive (e.g., 14.576 to 15.255; t-statistics [8.267]*** to [8.949]***)
  - FinLib*ExtFinDep interactions: 0.265, 0.375, 0.335, 0.018 with t-statistics [1.105], [1.991]**, [1.821]*, [0.092]
  - N = 39,134 (N Groups = 2,092). Hansen-Sargan p-value = 0.000; m1 p-value = 0.000; m2 p-value examples: 0.247, 0.315, 0.343, etc.

### Substantive empirical findings distilled from reported estimates
- Financial liberalization (FinLib, defined as M&A based bank branch deregulation) is associated with positive and statistically significant effects on real growth rates at the state level and state-industry level:
  - Examples: FinLib coefficients in state-level regressions range from 1.621 to 2.137 with t-statistics indicating significance at 1% in several specifications.
  - In state-industry panel GMM, FinLib coefficients range up to 2.776 (AR(2) specification), with t-statistics such as [8.817]***.
- Employment protection (WorkRight, binary indicator of wrongful discharge exceptions) is associated with negative coefficients on growth measures across multiple specifications:
  - Examples: WorkRight coefficients in state-industry tables are often negative and statistically significant (e.g., -1.904 with t-statistic [-2.983]*** in Table 3; -2.896 with t-statistic [-4.820]*** in Table 6a).
- Interaction effects indicate heterogeneity by industry characteristics:
  - FinLib*ExtFinDep: positive interaction coefficients in several specifications (e.g., 0.418 with [1.828]* in Table 6a; 0.265 with [1.105] in Table 6b), suggesting financial liberalization benefits are larger in industries more dependent on external finance in some specifications.
  - WorkRight*Schooling: consistently large positive coefficients (e.g., 7.393 [5.767]*** in Table 3; 14.661 [8.464]*** in Table 6a), indicating the negative baseline effect of WorkRight on growth is mitigated or reversed in industries with higher average schooling (greater reliance on knowledge).
  - WorkRight interactions with Intangible/Fixed Assets and Sales/Fixed Assets are positive and significant (Table 4 and Table 5), indicating the employment protection-growth relationship differs for industries with higher intangible intensity or higher sales-to-fixed-assets ratios.

### Key statistics and model diagnostics (selected exact values)
- Sample period: 1972 to 1993 (dependent variable timeframe for growth measures)
- Observation and grouping counts:
  - State-level N: 1029 (or 1008 in some specs); N Groups = 48
  - State-industry level N examples: 51,201; 50,276; 41,249; 39,134
  - N Groups examples: 2,487; 2,440; 2,097; 2,092
- R-squared ranges in cross-section/fixed-effect specifications: examples 0.0924–0.0960 (state-industry), ~0.0943–0.0951 (intangible/sales specifications), 0.256–0.380 (state-level)
- GMM diagnostics: Hansen-Sargan p-value = 0.000 reported for all GMM specifications; m1 p-value = 0.000; m2 p-value examples: 0.378 (AR(1)), 0.247 (AR(2))

### Interpretive implications (based on reported estimates)
- The reported estimates imply that M&A-based bank branch deregulation (FinLib) is consistently associated with higher real growth rates across state and state-industry specifications.
- Employment protection statutes/exceptions (WorkRight) are associated with lower growth on average, but their negative association is attenuated or reversed in industries characterized by higher schooling, higher intangible intensity, or higher sales-to-fixed-assets (i.e., more knowledge-intensive or externally finance-dependent industries).
- Interaction terms (FinLib*ExtFinDep, WorkRight*Schooling, WorkRight*Intangible/Fixed Assets, WorkRight*Sales/Fixed Assets) point to meaningful cross-industry heterogeneity in the growth effects of both financial liberalization and employment protection.

*Italicized attribution: Source: _wp08229 - introduction of employment protection. (PDF chapter/section).*

### Chapter 6, The Role of Government in East Asian Economic Development, ed. by

### Chapter 6, The Role of Government in East Asian Economic Development, ed. by

### References cited in excerpt
- Masahiko Aoki, Hyung-Ki Kim, and Masahiro Okuno-Fujiwara (New York: Oxford University Press).
- Hirsch, Barry T., and David A. MacPherson, 2003, Union Membership and Coverage Database from the Current Population Survey: Note, Industrial and Labor Relations Review, Vol. 56, No. 2, pp. 349-54.
- Jayaratne, Jith, and Philip E. Strahan, 1996, “The Finance-Growth Nexus: Evidence from Bank Branch Deregulation,” Quarterly Journal of Economics, Vol. 111, No. 3, pp. 639-70.
- Jensen, Michael C., and William H. Meckling, 1976, “Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure,” Journal of Financial Economics, Vol. 3, pp. 305-60.
- Joh, Sung Wook, 2003, “Corporate Governance and Firm Profitability: Evidence from Korea Before the Economic Crisis,? Journal of Financial Economics, Vol. 68, No. 2, pp. 287-322.
- Johnson, Simon, Peter Boone, Alasdair Breach, and Eric Friedman, 2000, “Corporate Governance in the Asian Financial Crisis,” Journal of Financial Economics, Vol. 58, No.1-2, pp. 141-86.
- Kaboski, Joseph P., 2007, “The Rise of the Service Economy, with Appendix A: Skill-Intensity and Labor Income Growth of Detailed Service Industries,” mimeo, Northwestern University http://faculty.wcas.northwestern.edu/~fjb913/research.html.
- Karpoff, J. and Malatesta, P., 1989, “The Wealth Effects of Second-Generation State Takeover Legislation.” Journal of Financial Economics, Vol. 25, pp. 291–322
- ———, 1995, “State Takeover Legislation and Share Values: The Wealth Effects of Pennsylvania’s Act 36.” Journal of Corporate Finance, Vol. 1, pp. 367–82.
- Kim, E. Han, and Paige Quimet, 2007, “Employee Capitalism or Corporate Socialism? Broad-Based Employee Stock Ownership,” Working Paper, Ross School of Business, University of Michigan.
- La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert Vishny, 1997, “Legal Determinants of External Finance,” Journal of Finance, Vol. 52, No. 3, pp. 131-150.
- ———, 1998, Law and Finance, Journal of Political Economy, Vol. 106, No. 6, pp. 1113-155.
- Levine, Ross, 2005, “Finance and Growth: Theory and Evidence,” in Philippe Aghion and Steven Durlauf, eds. Handbook of Economic Growth. The Netherlands: Elsevier Science.
- Miles, Thomas J., 2000, “Common Law Exceptions to Employment at Will and U.S. Labor Markets,” Journal of Law, Economics, and Organizations, Vol. 16, No. 1, pp. 74-101.
- Mitton, Todd, 2002, “A Cross-Firm Analysis of the Impact of Corporate Governance on the East Asian Financial Crisis,” Journal of Financial Economics, Vol. 64, No. 2, pp. 15-41.
- Mukoyama, Toshihiko, and Aysegul Sahin, 2006, “Specialization and Human Capital in Search Equilibrium,” Journal of the European Economic Association, Vol. 4, No. 2-3, pp. 503-12.
- Murphy, Kevin M., 1986, Specialization and Human Capital, Ph.D. dissertation, University of Chicago.
- Pagano, Marco, and Paolo F. Volpin, 2005, “The Political Economy of Corporate Governance, American Economic Review, Vol. 95, No. 4, pp. 1005-1005.
- Parente, Stephen L., and Edward C. Prescott, 2000, Barriers to Riches (Cambridge: MIT Press).
- Partridge, Mark D., and Jamie S. Partridge, 1999, “Do Minimum Wage Hikes Reduce Employment? State-Level Evidence from the Low-Wage Retail Sector,” Journal of Labor Research, Vol. XX, No. 3, pp. 393-413.
- Pissarides, Christopher A., 2000, Equilibrium Unemployment Theory (Cambridge: MIT Press).
- Rajan, Raghuram, and Luigi Zingales, 1998, “Financial Dependence and Growth,” American Economic Review, Vol. 88, No. 3, pp. 559–86.
- Stiroh, Kevin J., and Philip E. Strahan, 2003, “Competitive Dynamics of Deregulation: Evidence from U.S. Banking,” Journal of Money, Credit, and Banking, Vol. 35, No. 5, pp. 801-28.
- Szewczyk, S., Tsetsekos, G., 1992, “State Intervention in the Market for Corporate Control: The Case of Pennsylvania Senate Bill 1310,” Journal of Financial Economics, Vol. 31, pp. 3-23.
- Takizawa, Hajime, 2003, “Job-Specific Investment and the Cost of Dismissal Restrictions—The Case of Portugal,” IMF Working Paper 03/75 (Washington: International Monetary Fund).
- Tirole, 2006, The Theory of Corporate Finance. Princeton: Princeton University Press.
- Ueda, Kenichi, 2006, “Banks as Coordinators of Economic Growth,” IMF Working Paper, No. 06/264 (Washignton: international Monetary Fund).
- Wald, John K., and Michael S. Long, 2007, “The Effect of State Laws on Capital Structure,” Journal of Financial Economics, Vol. 83, pp. 297-319.

*Content excerpted from the chapter bibliography.*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08229.pdf_