## 2.1  Traditional determinants of M&A investment

## Source details

**Canonical URL:** [2.1  Traditional determinants of M&A investment](https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019264-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2019/wpiea2019264-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2019/wpiea2019264-print-pdf.pdf.json)

---

### Traditional motives and domestic determinants
- Domestic M&A typically occurs when firm management perceives potential gains from acquiring another entity (Jensen and Ruback (1983); Jarrell et al. (1988); Andrade et al. (2001)).
- Sources of gains listed:
  - Production efficiencies, e.g., reduction in contracting costs across firms.
  - Tax optimization motives.
  - Acquisition to gain market power.
  - Managerial utility-maximizing (potentially value-decreasing) acquisitions.
- Additional domestic determinants highlighted:
  - Importance of intra-industrial flows.
  - Deregulation can increase M&A activity.

### Cross-border determinants and frictions
- Cross-border M&As are affected by additional frictions beyond domestic determinants:
  - Geographic distance, differences in language, currency, legal framework, colonial origin, and time zones.
  - Information asymmetries (valuation of target firms harder for foreign acquirers).
  - Differences in regulations; Chari et al. (2009) find developed-market acquirers benefit more from weaker contracting environments in emerging markets.
  - Currency fluctuations affect profitability of investments independent of fundamentals.
- Empirical strategy will include standard gravity-model variables.

### Financial development and sectoral patterns
- Rajan and Zingales (1998): well-developed financial markets represent a comparative advantage in industries more dependent on external finance.
- Lack of financial development can impede new sector development, affect inflow amounts, sectoral composition, and concentration.
- The analysis will consider sectors in which investments are realized.

### Trade, comparative advantage, and capital flows
- Classical Heckscher-Ohlin-Mundell implications:
  - Exports based on endowments; advanced economies export capital.
  - Trade and capital flows are substitutes; trade integration reduces incentives for capital to flow to capital-scarce countries.
  - De Sousa and Lochard (2011): EMU has increased intra-EMU FDI stocks on average by around 30 percent.
- Recent theory allows trade and capital flows to be complements and emerging economies to export capital (Antras and Caballero (2009); Ju and Wei (2011); Jin (2012)).
- Firm-level motives can drive both exporting and investing abroad (Greenaway and Kneller (2007); Alfaro and Charlton (2009)).
- De la Torre et al. (2015) (cross-country sectoral gravity): advanced economies invest more in sectors where the receiver has comparative advantage; emerging and developing markets invest more where the receiver has a disadvantage.
- This paper integrates comparative advantages of trade into M&A determinants.

### Network determinants of bilateral decision
- Core hypothesis: information frictions are key in international investment (Chaney (2014) framing).
  - Existing contacts reduce search costs and facilitate entry into new partner countries.
  - Empirical implication: if country a invested in country b in yeart, then country a is more likely to invest for the first time in country c in yeart+1 if b had already invested in c.
- Related literature:
  - “Export-platform” literature (Ekholm et al. (2007); Yeaple (2003); Bergstrand and Egger (2007)) — parent invests in host to serve third markets.
  - Head et al. (1995); Head and Mayer (2004) find industry-level agglomeration and adjacent-market potential matter for location choice.
  - Blonigen et al. (2007) use spatial econometrics and find suggestive export-platform FDI evidence in developed Europe.
- This study uses ERGM and TERGM to model extra-dyadic interdependencies arising from an "alliance" network (Cranmer et al. (2012)).

### Methodology: ERGM and TERGM overview
- Rationale:
  - ERGM and TERGM allow examination of higher-level dependencies in an M&A network.
  - Observed M&A network is treated as one realization from the distribution of possible networks.
- Exponential Random Graph Model (ERGM):
  - Probability: Pβ(Y = y | β) = exp(β · S(y)) / Σexp(β · S(y′)).
  - Y is the random network variable; β is vector of model parameters; S(y) is vector of network statistics.
  - Interpretation: β is the log-odds impact of a variable on the appearance of a tie between two countries.
  - Computational challenge: denominator requires summing over all possible networks; addressed using Markov Chain Monte Carlo (MCMC) sampling techniques (Snijders (2002); Handcock et al. (2003)).
  - Convergence remains an issue for many ERGM specifications (Handcock et al. (2003); Hunter et al. (2008)).
  - Consistency challenges: increasing observations does not necessarily increase accuracy. Necessary condition per Jackson (2010): “non-conflicted” condition.
  - Goodness-of-fit comparisons recommended for degree distribution, edgewise shared partners, and geodesic distribution (Hunter et al. (2008)).
- Temporal ERGM (TERGM):
  - Uses a separable TERGM: formation and dissolution of ties occur independently within each time step and are modeled as separate ERGMs.
  - Formation network Y′ at time t+1 conditional on Yt: Pβ′(Y′ = y′ | Yt ; β′) = exp(β′ · S(y′)) / Σexp(β′ · S(y′)).
  - Dissolution network modeled similarly with parameter β́.
  - Cross-sectional network at t+1 constructed by applying formation and dissolution changes to Yt: Yt+1 = Yt ⊖ (Y′ \ Yt) ⊕ (Yt \ Ý).
- Strategies to obtain convergence:
  - Constrain possible networks to those with the same number of ties as the observed network (interpreting bilateral trade coefficient conditional on fixed prevalence of ties).
  - Binarize M&A flows matrix and use a dummy for High Income countries instead of GDPs.
  - Limit nodes to high income and emerging/developing countries using IMF classification; exclude low income economies. Sample reduced to 83 countries, representing nevertheless more than 94 percent of global flows.

### Data and descriptive statistics
- M&A data source: Thompson Reuter’s Security Data Corporation Platinum database for operations realized between 2000 and 2016.
- Data processing:
  - Aggregated by country to obtain bilateral country-level database.
  - Sectors classified by 4-digit Standard Industrial Classification (SIC), based on receiving firm classification.
  - Network representation: countries as nodes; M&A outflows as directed ties; network binarized.
  - Three investment types defined:
    - Primary sector (including agriculture, mining, and oil).
    - Light manufacturing (including food, textiles, and wood).
    - Heavy manufacturing (including chemicals, metals, machinery, and equipment).
  - One network constructed by sector and by year.
- Trade data for NRCA constructed following Vollrath (1991):
  - RCA uses X(i,j,t) = exports of country i in industry j at period t.
  - Dependent variable specified as log(1 + flows) to account for many zero observations.
  - NRCA calculated at bilateral level for each sector and year from UN Comtrade/World Integrated Trade Solution.
  - Dataset covers period from 2000 to 2016 for 205 source and recipient countries.
  - SITC and SIC bridged using Eurostat conversion tables; aggregated into three sectors.
- Control variables:
  - Node-specific and dyadic-specific controls adapted from gravity literature.
  - Sources: GeoDist database (CEPII), The World Factbook (CIA), World Development Indicators (World Bank).
  - Controls include: trade openness (sum of exports and imports), distance, longitude, latitude (all in km), differences in time zones (in hours), common language, common legal origin, colonial history.

### Network-level variables and statistics
- Structural terms included in ERGM/TERGM:
  - Edges: number of links in the network (interpretable as intercept parameter).
  - Transitivity via geometrically weighted edgewise shared partner (GWESP) statistic with parameter α.
    - GWESP measures how frequently two nodes are connected directly and via indirect connections of length 2.
    - A significant positive GWESP coefficient indicates transitivity beyond that explained by nodal characteristics.
    - Shared partner definition: two countries share a partner if both have a tie to the same country; each shared partner forms a triangle if the original pair are tied.
    - GWESP (parametric form): GWESP = e^α Σ_{i=1}^{n-2} [1 − (1 − e^{−α})^i] p_i, where p_i equals number of country pairs connected who share exactly i partners.
    - Adopted α = 0.25 (standard in literature). Alternative α explored in range from 0 to 0.5 with relatively small impact on coefficient estimates or model fit.

### Estimation approach and next steps
- Empirical strategy:
  - Test trade openness and gravity variables in a logit regression.
  - Estimate potential impact of network variables using ERGM and TERGM procedures.
- The Results section will present empirical estimates of M&A determinants, incorporating trade openness, gravity controls, and network effects.

---

### 5.1  Logit estimations

### Overview of regressions and specifications
- Regressions link M&A flows with comparative advantages of source and receiving countries across three sectors: primary, light manufacturing, and heavy manufacturing.
- Specifications:
  - Cross-country regression for 2016 (columns (1)–(3)).
  - Cross-country panel from 2000 to 2016 (columns (4)–(6)).
- Dependent variable: dummy equal to one when the M&A flow between two countries is positive, zero otherwise.
- Total trade measured as the sum of exports and imports.
- Relative comparative advantage (RCA) based on Vollrath (1991).
- Observations: 6,806 (2016 regressions) and 115,702 (2000–2016 regressions) per sector column.
- Number of countries: 83.
- Standard errors clustered by country pairs; regressions control for source- and target-country dummies.
- Data sources: SDC Platinum and Comtrade.

### Key empirical findings (logit)
- Trade openness (coefficients reported exactly as Table 1):
  - 0.718*** (Primary, 2016)
  - 0.766*** (Light Manuf., 2016)
  - 0.885*** (Heavy Manuf., 2016)
  - 0.898*** (Primary, 2000-2016)
  - 0.806*** (Light Manuf., 2000-2016)
  - 0.946*** (Heavy Manuf., 2000-2016)
- Interpretation from source: In 2016, an increase in one unit of the log of trade openness variable (about 2.8 percentage points increase in trade openness) was associated with a higher probability of an M&A transaction of 0.7 percent in the primary sector, 0.8 in the light manufacturing sector and 0.9 percent in the heavy manufacturing sector.
- High-income status (coefficients):
  - 0.859** (Primary, 2016)
  - 0.885*** (Light Manuf., 2016)
  - 0.758*** (Heavy Manuf., 2016)
  - 1.166*** (Primary, 2000-2016)
  - 1.137*** (Light Manuf., 2000-2016)
  - 1.425*** (Heavy Manuf., 2000-2016)
- Net RCA (acquirer and receiver) (coefficients):
  - Net RCAik (acquirer):
    - 0.211*** (Primary, 2016)
    - -0.166 (Light Manuf., 2016)
    - -0.351*** (Heavy Manuf., 2016)
    - 0.0873*** (Primary, 2000-2016)
    - 0.0667 (Light Manuf., 2000-2016)
    - -0.0743* (Heavy Manuf., 2000-2016)
  - Net RCAjk (receiver):
    - 0.169*** (Primary, 2016)
    - 0.383*** (Light Manuf., 2016)
    - 0.669*** (Heavy Manuf., 2016)
    - 0.0598*** (Primary, 2000-2016)
    - 0.368*** (Light Manuf., 2000-2016)
    - 0.790*** (Heavy Manuf., 2000-2016)
- Interpretation from source:
  - Positive relationship between the RCA of the receiver country and M&As across sectors.
  - Acquirers with a net comparative advantage tend to invest abroad in primary sector; acquirers with comparative disadvantage tend to invest in foreign heavy manufacturing.
  - No statistical evidence for light manufacturing acquirer RCA.

### Gravity and bilateral controls (selected coefficients)
- Time difference:
  - 0.0883** (Primary, 2016)
  - -0.0486 (Light Manuf., 2016)
  - 0.00255 (Heavy Manuf., 2016)
  - 0.0215 (Primary, 2000-2016)
  - -0.0912*** (Light Manuf., 2000-2016)
  - -0.0437*** (Heavy Manuf., 2000-2016)
- Common language:
  - 0.814** (Primary, 2016)
  - 0.892*** (Light Manuf., 2016)
  - 0.586** (Heavy Manuf., 2016)
  - 0.991*** (Primary, 2000-2016)
  - 0.820*** (Light Manuf., 2000-2016)
  - 0.762*** (Heavy Manuf., 2000-2016)
- Colonial relationship:
  - 0.452 (Primary, 2016)
  - 0.0638 (Light Manuf., 2016)
  - 0.600** (Heavy Manuf., 2016)
  - 0.807** (Primary, 2000-2016)
  - 0.538** (Light Manuf., 2000-2016)
  - 0.582** (Heavy Manuf., 2000-2016)
- Currency union:
  - -1.133 (Primary, 2016)
  - -1.395*** (Light Manuf., 2016)
  - -0.192 (Heavy Manuf., 2016)
  - -1.386*** (Primary, 2000-2016)
  - -0.637*** (Light Manuf., 2000-2016)
  - -0.699*** (Heavy Manuf., 2000-2016)
- Difference in latitude:
  - -0.00809 (Primary, 2016)
  - -0.0163*** (Light Manuf., 2016)
  - -0.0144*** (Heavy Manuf., 2016)
  - -0.000589 (Primary, 2000-2016)
  - -0.0144*** (Light Manuf., 2000-2016)
  - -0.0115*** (Heavy Manuf., 2000-2016)
- Common legal origin:
  - 1.082*** (Primary, 2016)
  - 1.171*** (Light Manuf., 2016)
  - 1.063*** (Heavy Manuf., 2016)
  - 1.803*** (Primary, 2000-2016)
  - 1.956*** (Light Manuf., 2000-2016)
  - 1.826*** (Heavy Manuf., 2000-2016)
- Common border:
  - 1.612*** (Primary, 2016)
  - 1.222*** (Light Manuf., 2016)
  - 0.871*** (Heavy Manuf., 2016)
  - 1.895*** (Primary, 2000-2016)
  - 1.569*** (Light Manuf., 2000-2016)
  - 1.591*** (Heavy Manuf., 2000-2016)

### Statistical significance and interpretation notes
- Gravity variables generally show expected signs and are often statistically significant (common language, colonial relationship, common legal origin, common border positive; difference in latitude negative for manufacturing regressions).
- Time difference is negative and statistically significant for light and heavy manufacturing in 2000–2016, not significant otherwise.
- The negative sign on currency union is unexpected and noted as requiring further research.

### Descriptive statistics (selected exact figures)
- Table A2: Descriptive statistics for 2016 (Obs. = 6,806)
  - Primary (Obs. 6,806)
    - Mergers and Acquisitions: Mean 0.01 Std. Dev. 0.11 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 17.82 Std. Dev. 2.11 Min 12.42 Max 21.72
    - Net RCA of source country: Mean -0.86 Std. Dev. 2.34 Min -9.53 Max 6.85
    - Net RCA of receiver country: Mean -0.87 Std. Dev. 2.34 Min -10.64 Max 6.85
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.60 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1
  - Light Manufacturing (Obs. 6,806)
    - Mergers and Acquisitions: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 17.82 Std. Dev. 2.11 Min 12.42 Max 21.72
    - Net RCA of source country: Mean -0.52 Std. Dev. 1.13 Min -7.04 Max 1.66
    - Net RCA of receiver country: Mean -0.52 Std. Dev. 1.13 Min -7.04 Max 1.91
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.60 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1
  - Heavy Manufacturing (Obs. 6,806)
    - Mergers and Acquisitions: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 17.82 Std. Dev. 2.11 Min 12.42 Max 21.72
    - Net RCA of source country: Mean -0.98 Std. Dev. 1.31 Min -6.65 Max 4.83
    - Net RCA of receiver country: Mean -0.98 Std. Dev. 1.31 Min -6.65 Max 4.83
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.60 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1
- Table A3: Descriptive statistics for 2000-2016 (Obs. = 115,702)
  - Trade openness (in logs): Mean 12.26 Std. Dev. 3.80 Min 4.85 Max 21.82 (reported for Primary, Light and Heavy Manufacturing)
  - Net RCA of source/receiver country (Primary): Mean -0.35 Std. Dev. 2.51 Min -13.48 Max 15.72
  - Net RCA of source/receiver country (Light Manufacturing): Mean -0.51 Std. Dev. 1.40 Min -11.28 Max 3.59
  - Net RCA of source/receiver country (Heavy Manufacturing): Mean -1.26 Std. Dev. 1.63 Min -11.78 Max 4.86
  - Other dyadic controls: Time difference Mean 4.69 Std. Dev. 3.82 Min 0 Max 18; Difference in latitude Mean 30.26 Std. Dev. 23.59 Min 0 Max 105; Common language Mean 0.07 Std. Dev. 0.26 Min 0 Max 1; Colonial relationship Mean 0.02 Std. Dev. 0.14 Min 0 Max 1; Currency union Mean 0.04 Std. Dev. 0.19 Min 0 Max 1; Common legal origin Mean 0.06 Std. Dev. 0.24 Min 0 Max 1; Common border Mean 0.03 Std. Dev. 0.16 Min 0 Max 1.

### ERGM/TERGM estimation materials referenced
- Goodness-of-fit and MCMC trace plots estimated for statistically significant variables (Figures A1–A3 referenced).
- Yearly ERGM estimation results (2000–2016) and several 5-year rolling specifications reported in Tables A4–A12, including variants restricted to transactions superior to 1 million US dollars.
- Common model elements across ERGM tables:
  - GWESP reported yearly with coefficients typically positive and many significant at *** p < 0.001.
  - Node covariates: High Income i, Trade Openness i, Net RCA ik (source), Net RCA jk (receiver), Time difference, Common language, Colonial relationship, Currency union, Difference in latitude, Common legal origin, Common border, and Edges.
  - Significance levels: *** p < 0.001, ** p < 0.01, * p < 0.05. ‚ indicates years for which the ERGM estimation did not convergence.
- Selected numeric model metadata (examples reported exactly from source):
  - Table A4 (Primary sector, annual ERGM): Observations 6,806; AIC example values 637.26, 563.95, 560.69, 662.5, 574.15, 726.83, 848.09, 1009.74, 1156.77, 1034.64, 1058.67, 1138.24, 959.88, 876.42, 883.95, 788.86, 685.05; BIC example values 726.62, 653.3, 650.05, 751.86, 663.5, 816.19, 937.45, 1099.1, 1246.13, 1124, 1148.02, 1227.6, 1049.23, 965.77, 973.31, 878.21, 774.41; Triangles example values 76, 74, 90, 86, 147, 154, 207, 340, 451, 309, 408, 409, 241, 191, 136, 136, 80; Edges example values: -28.18, -12.72 ***, -12.81 ***, -15.54 ***, -15.38 ***, -10.6 ***, -11.11 ***, -13.68 ***, -10.21 ***, -11.26 ***, -11.72 ***, -11.38 ***, -12.3 ***, -31.47, -15.92 ***, -28.98, -16.87 ***.
  - Table A5 (Light Manufacturing, annual ERGM): Observations 6,806; Triangles example values 282, 119, 142, 112, 109, 130, 186, 201, 173, 65, 71, 112, 90, 140, 118, 150, 147; AIC example values 904.85, 895.87, 890.7, 967.67, 889.63, 872.39, 960.83, 1022.39, 948.14, 723.01, 845.25, 981.85, 931.11, 941.09, 858.57, 838.86, 917.5; Edges example values: -9.905 ***, -11.46 ***, -10.75 ***, -10.51 ***, -11.64 ***, -11.92 ***, -11.26 ***, -9.02 ***, -10.6 ***, -13.6 ***, -11.34 ***, -10.16 ***, -10.79 ***, -15.88 ***, -16.96 ***, -18.47 ***, -16.09 ***.
  - Table A6 (Heavy Manufacturing, annual ERGM): Observations 6,806; Triangles example values 1269, 1112, 826, 801, 654, 1026, 1309, 1726, 1435, 769, 921, 1113, 906, 876, 1288, 1122, 940; AIC example values 1376.85, 1326.86, 1173.76, 1197.49, 1321.86, 1282.44, 1653.93, 1734.35, 1689.91, 1369.9, 1441.92, 1460.81, 1468.77, 1352.7, 1452.03, 1387.89, 1412.81; Edges example values: -11.831 ***, -12.39 ***, -11.52 ***, -12.37 ***, -10.85 ***, -11.81 ***, -11.95 ***, -12.5 ***, -10.47 ***, -11.21 ***, -10.61 ***, -11.96 ***, -11.23 ***, -15.07 ***, -14.51 ***, -16.18 ***, -14.57 ***.
  - 5-year rolling specifications (Tables A7–A12) report Observations = 34,030 and analogous model statistics (examples reported in source).

*Source: wpiea2019264-print-pdf - 2.1 Traditional determinants of M&A investment; 5.1 Logit estimations — https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019264-print-pdf.pdf*

### 2.1  Traditional determinants of M&A investment

### 2.1  Traditional determinants of M&A investment

### Traditional motives and domestic determinants
- Domestic M&A typically occurs when firm management perceives potential gains from acquiring another entity (references: Jensen and Ruback (1983); Jarrell et al. (1988); Andrade et al. (2001)).
- Sources of gains listed:
  - Production efficiencies, e.g., reduction in contracting costs across firms.
  - Tax optimization motives.
  - Acquisition to gain market power.
  - Managerial utility-maximizing (potentially value-decreasing) acquisitions.
- Additional domestic determinants highlighted:
  - Importance of intra-industrial flows.
  - Deregulation can increase M&A activity.

### Cross-border determinants and frictions
- Cross-border M&As are affected by additional frictions beyond domestic determinants:
  - Geographic distance, differences in language, currency, legal framework, colonial origin, and time zones.
  - Information asymmetries (valuation of target firms harder for foreign acquirers).
  - Differences in regulations; Chari et al. (2009) find developed-market acquirers benefit more from weaker contracting environments in emerging markets.
  - Currency fluctuations affect profitability of investments independent of fundamentals.
- Empirical strategy will include standard gravity-model variables.

### Financial development and sectoral patterns
- Rajan and Zingales (1998) emphasize state of development of financial markets:
  - Well-developed markets represent a comparative advantage in industries more dependent on external finance.
  - Lack of financial development can impede new sector development, affect inflow amounts, sectoral composition, and concentration.
- The analysis will consider sectors in which investments are realized.

### Trade, comparative advantage, and capital flows
- Classical Heckscher-Ohlin-Mundell paradigm implications:
  - Exports based on endowments; advanced economies export capital.
  - Trade and capital flows are substitutes; trade integration reduces incentives for capital to flow to capital-scarce countries.
  - Note: De Sousa and Lochard (2011) find that the Economic and Monetary Union (EMU) has increased intra-EMU FDI stocks on average by around 30 percent.
- Recent theory allows trade and capital flows to be complements and emerging economies to export capital (Antras and Caballero (2009); Ju and Wei (2011); Jin (2012)).
- Firm-level motives can drive both exporting and investing abroad (Greenaway and Kneller (2007); Alfaro and Charlton (2009)).
- De la Torre et al. (2015): using a cross-country sectoral gravity framework, finds:
  - Advanced economies invest more in sectors where the receiver has comparative advantage.
  - Emerging and developing markets invest more where the receiver has a disadvantage.
- This paper integrates comparative advantages of trade into M&A determinants.

### Network determinants of bilateral decision
- Core hypothesis: information frictions are key in international investment (Chaney (2014) framing).
  - Existing contacts reduce search costs and facilitate entry into new partner countries.
  - Empirical implication: if country a invested in country b in yeart, then country a is more likely to invest for the first time in country c in yeart+1 if b had already invested in c.
- Related literature:
  - “Export-platform” literature (Ekholm et al. (2007); Yeaple (2003); Bergstrand and Egger (2007)) — parent invests in host to serve third markets.
  - Head et al. (1995); Head and Mayer (2004) find industry-level agglomeration and adjacent-market potential matter for location choice.
  - Blonigen et al. (2007) use spatial econometrics and find suggestive export-platform FDI evidence in developed Europe.
- This study uses ERGM and TERGM to model extra-dyadic interdependencies arising from an "alliance" network (Cranmer et al. (2012)).

### Methodology: ERGM and TERGM overview
- Rationale:
  - ERGM and TERGM allow examination of higher-level dependencies in an M&A network.
  - Observed M&A network is treated as one realization from the distribution of possible networks.
- Exponential Random Graph Model (ERGM):
  - Probability of observing network y: Pβ(Y = y | β) = exp(β · S(y)) / Σexp(β · S(y′)).
  - Y is the random network variable; β is vector of model parameters; S(y) is vector of network statistics.
  - Interpretation: β is the log-odds impact of a variable on the appearance of a tie between two countries.
  - Computational challenge: denominator requires summing over all possible networks; addressed using Markov Chain Monte Carlo (MCMC) sampling techniques (Snijders (2002); Handcock et al. (2003)).
  - Convergence remains an issue for many ERGM specifications (Handcock et al. (2003); Hunter et al. (2008)).
  - Consistency challenges: increasing observations does not necessarily increase accuracy. Necessary condition per Jackson (2010): “non-conflicted” condition (small neighborhoods must render statistics unconstrained).
  - To improve accuracy, goodness-of-fit comparisons are made for degree distribution, edgewise shared partners, and geodesic distribution (Hunter et al. (2008)).
- Temporal ERGM (TERGM):
  - Uses a separable TERGM: formation and dissolution of ties occur independently within each time step and are modeled as separate ERGMs.
  - Formation network Y′ at time t+1 conditional on Yt: Pβ′(Y′ = y′ | Yt ; β′) = exp(β′ · S(y′)) / Σexp(β′ · S(y′)).
  - Dissolution network Ý is modeled similarly with parameter β́.
  - Cross-sectional network at t+1 constructed by applying formation and dissolution changes to Yt: Yt+1 = Yt ⊖ (Y′ \ Yt) ⊕ (Yt \ Ý) [formal expression in source: Yt+1 = Yt ̄pY` ́Ytq ́pYt ́Y ́q].
- Strategies to obtain convergence:
  - Constrain possible networks to those with the same number of ties as the observed network (interpreting bilateral trade coefficient conditional on fixed prevalence of ties).
  - Binarize M&A flows matrix and use a dummy for High Income countries instead of GDPs.
  - Limit nodes to high income and emerging/developing countries using IMF classification; exclude low income economies. Sample reduced to 83 countries, representing nevertheless more than 94 percent of global flows.

### Data and descriptive statistics
- M&A data source: Thompson Reuter’s Security Data Corporation Platinum database for operations realized between 2000 and 2016.
- Data processing:
  - Aggregated by country to obtain bilateral country-level database.
  - Sectors classified by 4-digit Standard Industrial Classification (SIC), based on receiving firm classification.
  - Network representation: countries as nodes; M&A outflows as directed ties; network binarized.
  - Three investment types defined:
    - Primary sector (including agriculture, mining, and oil).
    - Light manufacturing (including food, textiles, and wood).
    - Heavy manufacturing (including chemicals, metals, machinery, and equipment).
  - One network constructed by sector and by year.
- Trade data for NRCA (net relative comparative advantage) constructed following Vollrath (1991):
  - RCAi,j,t formula uses X(i,j,t) = exports of country i in industry j at period t.
  - Dependent variable specified as log(1 + flows) to account for many zero observations.
  - NRCA calculated at bilateral level for each sector and year from UN Comtrade/World Integrated Trade Solution.
  - Dataset covers period from 2000 to 2016 for 205 source and recipient countries.
  - SITC and SIC bridged using Eurostat conversion tables; aggregated into three sectors.
- Control variables:
  - Node-specific and dyadic-specific controls adapted from gravity literature.
  - Sources: GeoDist database of Centre d’Etudes Prospectives et d’Informations Internationales, The World Factbook (CIA), World Development Indicators (World Bank).
  - Controls include: trade openness (sum of exports and imports), distance, longitude, latitude (all in km), differences in time zones (in hours), common language, common legal origin, colonial history.

### Network-level variables and statistics
- Structural terms included in ERGM/TERGM:
  - Edges: number of links in the network (interpretable as intercept parameter).
  - Transitivity via geometrically weighted edgewise shared partner (GWESP) statistic with parameter α.
    - GWESP measures how frequently two nodes are connected directly and via indirect connections of length 2.
    - A significant positive GWESP coefficient indicates transitivity beyond that explained by nodal characteristics.
    - Shared partner definition: two countries share a partner if both have a tie to the same country; each shared partner forms a triangle if the original pair are tied.
    - GWESP (parametric form) gives decreasing marginal impact for each additional shared partner.
    - GWESP formula: GWESP = e^α Σ_{i=1}^{n-2} [1 − (1 − e^{−α})^i] p_i, where p_i equals number of country pairs connected who share exactly i partners.
    - Adopted α = 0.25 (standard in literature).
    - Alternative α explored in range from 0 to 0.5 with relatively small impact on coefficient estimates or model fit.

### Estimation approach and next steps
- Empirical strategy:
  - Test trade openness and gravity variables in a logit regression.
  - Estimate potential impact of network variables using ERGM and TERGM procedures.
- The Results section will present empirical estimates of M&A determinants, incorporating trade openness, gravity controls, and network effects.

*Source: wpiea2019264-print-pdf - 2.1  Traditional determinants of M&A investment — https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019264-print-pdf.pdf*

### 5.1  Logit estimations

### 5.1  Logit estimations

### Overview
- The regressions in Table 1 use logit estimates to link M&A flows with the comparative advantages of source and receiving countries across three sectors: primary, light manufacturing, and heavy manufacturing.
- The regressions include gravity controls and examine a cross-country regression for 2016 (columns (1)–(3)) and a cross-country panel from 2000 to 2016 (columns (4)–(6)).
- The dependent variable is a dummy equal to one when the M&A flow between two countries is positive, and zero otherwise. Total trade is measured as the sum of exports and imports. Relative comparative advantage (RCA) is based on Vollrath (1991).

### Key empirical findings (logit)
- Trade openness:
  - Coefficients (Table 1): 0.718*** (Primary, 2016), 0.766*** (Light Manuf., 2016), 0.885*** (Heavy Manuf., 2016), 0.898*** (Primary, 2000-2016), 0.806*** (Light Manuf., 2000-2016), 0.946*** (Heavy Manuf., 2000-2016).
  - Interpretation: In 2016, an increase in one unit of the log of trade openness variable (about 2.8 percentage points increase in trade openness) was associated with a higher probability of an M&A transaction of 0.7 percent in the primary sector, 0.8 in the light manufacturing sector and 0.9 percent in the heavy manufacturing sector.
- High-income status:
  - Coefficients (Table 1): 0.859** (Primary, 2016), 0.885*** (Light Manuf., 2016), 0.758*** (Heavy Manuf., 2016), 1.166*** (Primary, 2000-2016), 1.137*** (Light Manuf., 2000-2016), 1.425*** (Heavy Manuf., 2000-2016).
  - Finding: High-income countries tend to invest more across all three sectors (high-income variable used as proxy for GDP acquirer).
- Net RCA (acquirer and receiver):
  - Net RCAik (acquirer) coefficients (Table 1): 0.211*** (Primary, 2016), -0.166 (Light Manuf., 2016), -0.351*** (Heavy Manuf., 2016), 0.0873*** (Primary, 2000-2016), 0.0667 (Light Manuf., 2000-2016), -0.0743* (Heavy Manuf., 2000-2016).
  - Net RCAjk (receiver) coefficients (Table 1): 0.169*** (Primary, 2016), 0.383*** (Light Manuf., 2016), 0.669*** (Heavy Manuf., 2016), 0.0598*** (Primary, 2000-2016), 0.368*** (Light Manuf., 2000-2016), 0.790*** (Heavy Manuf., 2000-2016).
  - Interpretation: There is a positive relationship between the relative comparative advantage (RCA) of the receiver country and M&As across sectors. Acquirers with a net comparative advantage tend to invest abroad in primary sector; acquirers with comparative disadvantage tend to invest in foreign heavy manufacturing. No statistical evidence for light manufacturing acquirer RCA.
- Gravity and bilateral controls (selected coefficients, Table 1):
  - Time difference: 0.0883** (Primary, 2016), -0.0486 (Light Manuf., 2016), 0.00255 (Heavy Manuf., 2016), 0.0215 (Primary, 2000-2016), -0.0912*** (Light Manuf., 2000-2016), -0.0437*** (Heavy Manuf., 2000-2016).
  - Common language: 0.814** (Primary, 2016), 0.892*** (Light Manuf., 2016), 0.586** (Heavy Manuf., 2016), 0.991*** (Primary, 2000-2016), 0.820*** (Light Manuf., 2000-2016), 0.762*** (Heavy Manuf., 2000-2016).
  - Colonial relationship: 0.452 (Primary, 2016), 0.0638 (Light Manuf., 2016), 0.600** (Heavy Manuf., 2016), 0.807** (Primary, 2000-2016), 0.538** (Light Manuf., 2000-2016), 0.582** (Heavy Manuf., 2000-2016).
  - Currency union: -1.133 (Primary, 2016), -1.395*** (Light Manuf., 2016), -0.192 (Heavy Manuf., 2016), -1.386*** (Primary, 2000-2016), -0.637*** (Light Manuf., 2000-2016), -0.699*** (Heavy Manuf., 2000-2016).
  - Difference in latitude: -0.00809 (Primary, 2016), -0.0163*** (Light Manuf., 2016), -0.0144*** (Heavy Manuf., 2016), -0.000589 (Primary, 2000-2016), -0.0144*** (Light Manuf., 2000-2016), -0.0115*** (Heavy Manuf., 2000-2016).
  - Common legal origin: 1.082*** (Primary, 2016), 1.171*** (Light Manuf., 2016), 1.063*** (Heavy Manuf., 2016), 1.803*** (Primary, 2000-2016), 1.956*** (Light Manuf., 2000-2016), 1.826*** (Heavy Manuf., 2000-2016).
  - Common border: 1.612*** (Primary, 2016), 1.222*** (Light Manuf., 2016), 0.871*** (Heavy Manuf., 2016), 1.895*** (Primary, 2000-2016), 1.569*** (Light Manuf., 2000-2016), 1.591*** (Heavy Manuf., 2000-2016).
- Statistical significance and signs:
  - Gravity variables generally show expected signs and are often statistically significant (common language, colonial relationship, common legal origin, common border positive; difference in latitude negative for manufacturing regressions).
  - Time difference is negative and statistically significant for light and heavy manufacturing in 2000–2016, not significant otherwise.
  - The negative sign on currency union is unexpected and noted as requiring further research.

### Estimation details and sample
- Observations (Table 1): 6,806 (2016 regressions) and 115,702 (2000–2016 regressions) per sector column.
- Number of countries: 83.
- R-squared (2016 columns): 0.2494 (Primary), 0.2558 (Light Manuf.), 0.2868 (Heavy Manuf.).
- Standard errors are clustered by country pairs; regressions also control for source- and target-country dummies.
- Data sources: Calculations based on data from SDC Platinum and Comtrade.

*Source: wpiea2019264-print-pdf - 5.1  Logit estimations.*

### References

### wpiea2019264-print-pdf - References

### Key literature cited
- Ahern, K. R., Daminelli, D. and Fracassi, C. (2015). Lost in translation? the effect of cultural values on mergers around the world. Journal of Financial Economics 117: 165–189.
- Alfaro, L. and Charlton, A. (2009). Intra-industry foreign direct investment. The American Economic Review 99: 2096–2119.
- Andrade, G., Mitchell, M. and Stafford, E. (2001). New evidence and perspectives on mergers. The Journal of Economic Perspectives 15: 103.
- Antras, P. and Caballero, R. J. (2009). Trade and capital flows: A financial frictions perspective. Journal of Political Economy 117: 701–744.
- Bergstrand, J. H. and Egger, P. (2007). A knowledge-and-physical-capital model of international trade flows, foreign direct investment, and multinational enterprises. Journal of International Economics 73: 278–308.
- Blonigen, B. A., Davies, R. B., Waddell, G. R. and Naughton, H. T. (2007). Fdi in space: Spatial autoregressive relationships in foreign direct investment. European Economic Review 51: 1303–1325.
- Chaney, T. (2014). The network structure of international trade. American Economic Review 104: 3600–3634.
- Chari, A., Ouimet, P. P. and Tesar, L. L. (2009). The value of control in emerging markets. The Review of Financial Studies 23: 1741–1770.
- Cranmer, S. J., Desmarais, B. A. and Menninga, E. J. (2012). Complex dependencies in the alliance network. Conflict Management and Peace Science 29: 279–313.
- De Sousa, J. and Lochard, J. (2011). Does the single currency affect foreign direct investment? The Scandinavian Journal of Economics 113: 553–578.
- Di Giovanni, J. (2005). What drives capital flows? the case of cross-border m&a activity and financial deepening. Journal of international Economics 65: 127–149.
- Ekholm, K., Forslid, R. and Markusen, J. R. (2007). Export-platform foreign direct investment. Journal of the European Economic Association 5: 776–795.
- Erdös, P. and Rényi, A. (1959). On random graphs, i. Publicationes Mathematicae (Debrecen) 6: 290–297.
- Frank, O. and Strauss, D. (1986). Markov graphs. Journal of the american Statistical association 81: 832–842.
- Goodreau, S. M., Kitts, J. A. and Morris, M. (2009). Birds of a feather, or friend of a friend? using exponential random graph models to investigate adolescent social networks. Demography 46: 103–125.
- Greenaway, D. and Kneller, R. (2007). Firm heterogeneity, exporting and foreign direct investment. The Economic Journal 117.
- Handcock, M. S., Robins, G., Snijders, T., Moody, J. and Besag, J. (2003). Assessing degeneracy in statistical models of social networks. Tech. rep., Citeseer.
- Head, K. and Mayer, T. (2004). Market potential and the location of japanese investment in the european union. Review of Economics and Statistics 86: 959–972.
- Head, K., Ries, J. and Swenson, D. (1995). Agglomeration benefits and location choice: Evidence from japanese manufacturing investments in the united states. Journal of international economics 38: 223–247.
- Herman, P. R. (2017). Identifying multilateral dependencies in the world trade network. USITC Working Paper 73.
- Hunter, D. R. (2007). Curved exponential family models for social networks. Social networks 29: 216–230.
- Hunter, D. R. and Handcock, M. S. (2006). Inference in curved exponential family models for networks. Journal of Computational and Graphical Statistics 15: 565–583.
- Hunter, D. R., Handcock, M. S., Butts, C. T., Goodreau, S. M. and Morris, M. (2008). ergm: A package to fit, simulate and diagnose exponential-family models for networks. Journal of statistical software 24: nihpa54860.
- Jackson, M. O. (2010). An overview of social networks and economic applications. The handbook of social economics 1: 511–85.
- Jarrell, G. A., Brickley, J. A. and Netter, J. M. (1988). The market for corporate control: The empirical evidence since 1980. The Journal of Economic Perspectives 2: 49–68.
- Jensen, M. C. and Ruback, R. S. (1983). The market for corporate control: The scientific evidence. Journal of Financial economics 11: 5–50.
- Jin, K. (2012). Industrial structure and capital flows. The American Economic Review 102: 2111–2146.
- Ju, J. and Wei, S.-J. (2011). When is quality of financial system a source of comparative advantage? Journal of International Economics 84: 178–187.
- Krivitsky, P. N. and Butts, C. T. (2013). Modeling valued networks with statnet. The Statnet Development Team: 2013.
- Krivitsky, P. N. and Handcock, M. S. (2014). A separable model for dynamic networks. Journal of the Royal Statistical Society: Series B (Statistical Methodology) 76: 29–46.
- Rajan, R. G. and Zingales, L. (1998). Financial dependence and growth. The American Economic Review 88: 559–586.
- Romer, P. (1993). Idea gaps and object gaps in economic development. Journal of monetary economics 32: 543–573.
- Snijders, T. A. (2002). Markov chain monte carlo estimation of exponential random graph models. Journal of Social Structure 3: 1–40.
- Snijders, T. A., Pattison, P. E., Robins, G. L. and Handcock, M. S. (2006). New specifications for exponential random graph models. Sociological methodology 36: 99–153.
- Torre, A. De la, Didier, T., Ize, A., Lederman, D. and Schmukler, S. L. (2015). Latin america and the rising south: Changing world, changing priorities.
- Vollrath, T. L. (1991). A theoretical evaluation of alternative trade intensity measures of revealed comparative advantage. Review of World Economics 127: 265–280.
- Yeaple, S. R. (2003). The complex integration strategies of multinationals and cross country dependencies in the structure of foreign direct investment. Journal of International Economics 60: 293–314.

### A1 Country sample and descriptive statistics
- Country sample (as listed): Algeria; Antigua & Barbuda; Argentina; Australia; Austria; Azerbaijan; Bahamas, The; Barbados; Belgium; Belize; Bosnia & Herzegovina; Botswana; Brazil; Brunei; Bulgaria; Canada; Chile; China; Colombia; Costa Rica; Croatia; Cyprus; Czech Rep.; Denmark; Dominican Rep.; Ecuador; Estonia; Fiji; Finland; France; Gabon; Germany; Greece; Guyana; Hong Kong SAR; Hungary; Iceland; Iran; Iraq; Ireland; Israel; Italy; Jamaica; Japan; Kazakhstan; Korea; Kuwait; Latvia; Lebanon; Lithuania; Luxembourg; Malaysia; Malta; Mauritius; Mexico; Netherlands; New Zealand; Norway; Panama; Peru; Poland; Portugal; Russia; Saudi Arabia; Seychelles; Singapore; Slovak Rep.; Slovenia; South Africa; Spain; St. Kit. & Nev.; St. Lucia; Sweden; Switzerland; Thailand; Trin. & Tob.; Turkey; UAE; United Kingdom; United States; Uruguay.

- Table A2: Descriptive statistics for 2016 (Primary, Light Manufacturing, Heavy Manufacturing; Obs. = 6,806)
  - Primary (Obs. 6,806)
    - Mergers and Acquisitions: Mean 0.01 Std. Dev. 0.11 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 17.82 Std. Dev. 2.11 Min 12.42 Max 21.72
    - Net RCA of source country: Mean -0.86 Std. Dev. 2.34 Min -9.53 Max 6.85
    - Net RCA of receiver country: Mean -0.87 Std. Dev. 2.34 Min -10.64 Max 6.85
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.60 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1
  - Light Manufacturing (Obs. 6,806)
    - Mergers and Acquisitions: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 17.82 Std. Dev. 2.11 Min 12.42 Max 21.72
    - Net RCA of source country: Mean -0.52 Std. Dev. 1.13 Min -7.04 Max 1.66
    - Net RCA of receiver country: Mean -0.52 Std. Dev. 1.13 Min -7.04 Max 1.91
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.60 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1
  - Heavy Manufacturing (Obs. 6,806)
    - Mergers and Acquisitions: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 17.82 Std. Dev. 2.11 Min 12.42 Max 21.72
    - Net RCA of source country: Mean -0.98 Std. Dev. 1.31 Min -6.65 Max 4.83
    - Net RCA of receiver country: Mean -0.98 Std. Dev. 1.31 Min -6.65 Max 4.83
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.60 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1

- Table A3: Descriptive statistics for 2000-2016 (Obs. = 115,702; Light and Heavy Manufacturing also reported)
  - Primary (Obs. 115,702)
    - Mergers and Acquisitions: Mean 0.02 Std. Dev. 0.13 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 12.26 Std. Dev. 3.80 Min 4.85 Max 21.82
    - Net RCA of source country: Mean -0.35 Std. Dev. 2.51 Min -13.48 Max 15.72
    - Net RCA of receiver country: Mean -0.35 Std. Dev. 2.51 Min -13.48 Max 15.72
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.59 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1
  - Light Manufacturing (Obs. 115,702)
    - Mergers and Acquisitions: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 12.26 Std. Dev. 3.80 Min 4.85 Max 21.82
    - Net RCA of source country: Mean -0.51 Std. Dev. 1.40 Min -11.28 Max 3.59
    - Net RCA of receiver country: Mean -0.51 Std. Dev. 1.40 Min -11.28 Max 3.59
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.59 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1
  - Heavy Manufacturing (Obs. 115,702)
    - Mergers and Acquisitions: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - High income: Mean 0.42 Std. Dev. 0.49 Min 0 Max 1
    - Trade openness (in logs): Mean 12.26 Std. Dev. 3.80 Min 4.85 Max 21.82
    - Net RCA of source country: Mean -1.26 Std. Dev. 1.63 Min -11.78 Max 4.86
    - Net RCA of receiver country: Mean -1.26 Std. Dev. 1.63 Min -11.78 Max 4.86
    - Time difference: Mean 4.69 Std. Dev. 3.82 Min 0 Max 18
    - Common language: Mean 0.07 Std. Dev. 0.26 Min 0 Max 1
    - Colonial relationship: Mean 0.02 Std. Dev. 0.14 Min 0 Max 1
    - Currency union: Mean 0.04 Std. Dev. 0.19 Min 0 Max 1
    - Difference in latitude: Mean 30.26 Std. Dev. 23.59 Min 0 Max 105
    - Common legal origin: Mean 0.06 Std. Dev. 0.24 Min 0 Max 1
    - Common border: Mean 0.03 Std. Dev. 0.16 Min 0 Max 1

### A2 Goodness-of-fit and ERGM estimation overview
- After running ERGM, the goodness-of-fit and MCMC trace plots were estimated for statistically significant variables (Figures A1–A3 referenced for Primary, Light manufacturing, Heavy manufacturing).
- Tables A4–A12 present yearly ERGM estimation results (2000–2016) and several 5-year rolling specifications, including variants restricted to transactions superior to 1 million US dollars.
- Common model elements across tables:
  - GWESP (geometrically weighted edgewise shared partner distribution) reported yearly with coefficients typically positive and many significant at *** p < 0.001.
  - Node covariates reported include High Income i, Trade Openness i, Net RCA ik (source), Net RCA jk (receiver), Time difference, Common language, Colonial relationship, Currency union, Difference in latitude, Common legal origin, Common border, and Edges.
  - Significance levels correspond to *** p < 0.001, ** p < 0.01, * p < 0.05. ‚ indicates years for which the ERGM estimation did not convergence.
  - Relative comparative advantage (RCA) is based on Vollrath (1991).
  - All regressions include gravity control variables: differences in latitude, differences in time zones, common language, common legal origin, and colonial relationship.
  - Sources: Calculations based on data from SDC Platinum and Comtrade.

- Selected numeric model metadata (examples taken from tables exactly as reported)
  - Table A4 (Primary sector, annual ERGM): Observations 6,806; AIC criteria (example years) 637.26, 563.95, 560.69, 662.5, 574.15, 726.83, 848.09, 1009.74, 1156.77, 1034.64, 1058.67, 1138.24, 959.88, 876.42, 883.95, 788.86, 685.05; BIC criteria (example years) 726.62, 653.3, 650.05, 751.86, 663.5, 816.19, 937.45, 1099.1, 1246.13, 1124, 1148.02, 1227.6, 1049.23, 965.77, 973.31, 878.21, 774.41; Triangles (example years) 76, 74, 90, 86, 147, 154, 207, 340, 451, 309, 408, 409, 241, 191, 136, 136, 80; Edges (example year values): -28.18, -12.72 ***, -12.81 ***, -15.54 ***, -15.38 ***, -10.6 ***, -11.11 ***, -13.68 ***, -10.21 ***, -11.26 ***, -11.72 ***, -11.38 ***, -12.3 ***, -31.47, -15.92 ***, -28.98, -16.87 ***.
  - Table A5 (Light Manufacturing, annual ERGM): Observations 6,806; Triangles (example years) 282, 119, 142, 112, 109, 130, 186, 201, 173, 65, 71, 112, 90, 140, 118, 150, 147; AIC criteria (example years) 904.85, 895.87, 890.7, 967.67, 889.63, 872.39, 960.83, 1022.39, 948.14, 723.01, 845.25, 981.85, 931.11, 941.09, 858.57, 838.86, 917.5; Edges (example year values): -9.905 ***, -11.46 ***, -10.75 ***, -10.51 ***, -11.64 ***, -11.92 ***, -11.26 ***, -9.02 ***, -10.6 ***, -13.6 ***, -11.34 ***, -10.16 ***, -10.79 ***, -15.88 ***, -16.96 ***, -18.47 ***, -16.09 ***.
  - Table A6 (Heavy Manufacturing, annual ERGM): Observations 6,806; Triangles (example years) 1269, 1112, 826, 801, 654, 1026, 1309, 1726, 1435, 769, 921, 1113, 906, 876, 1288, 1122, 940; AIC criteria (example years) 1376.85, 1326.86, 1173.76, 1197.49, 1321.86, 1282.44, 1653.93, 1734.35, 1689.91, 1369.9, 1441.92, 1460.81, 1468.77, 1352.7, 1452.03, 1387.89, 1412.81; Edges (example year values): -11.831 ***, -12.39 ***, -11.52 ***, -12.37 ***, -10.85 ***, -11.81 ***, -11.95 ***, -12.5 ***, -10.47 ***, -11.21 ***, -10.61 ***, -11.96 ***, -11.23 ***, -15.07 ***, -14.51 ***, -16.18 ***, -14.57 ***.

- 5-year rolling specifications (Tables A7–A12) report larger sample Observations = 34,030 and include analogous model statistics (AIC, BIC, Triangles) and the same covariates. Example entries:
  - Table A7 (5-year rolling Primary): Observations 34,030; AIC (example years) 1432.83, 1424.8, 1398.56, 1422.22, 1462.97, 1527.32, 1616.46, 1810.72, 1981.37, 2022.81, 2146.32, 2276.36, 2246.85, 2106.68, 2092.6, 2011.78, 1891.06; Triangles (example years) 640, 579, 558, 570, 694, 848, 932, 1266, 1570, 1822, 2104, 2468, 2482, 2161, 1950, 1692, 1459.
  - Table A10 (5-year rolling M&A superior to 1 million US dollars in Primary): Observations 34,030; AIC (example years) 1334.08, 1335.76, 1302.63, 1337.27, 1358.11, 1456.73, 1535.56, 1747.56, 1895.53, 1934.79, 2049.11, 2157.93, 2153.75, 2032.7, 2009.19, 1923.41, 1831.43; Triangles (example years) 575, 532, 494, 510, 553, 695, 781, 1145, 1452, 1681, 1970, 2267, 2295, 1996, 1820, 1571, 1350.

### Notes on model interpretation (as provided in source text)
- The GWESP indicator measures the likeliness of a common receiver country for two countries linked with an M&A (geometrically weighted edgewise shared partner distribution).
- The dependent variable varies by table: M&A flow occurrence (annual), equal to one if an M&A flow occurred in the last 5 years, or equal to one if an M&A transaction superior to 1 million US dollars occurred in the last 5 years, with sectoral breakdowns for Primary, Light manufacturing, and Heavy manufacturing.
- Total trade is measured as the sum of exports and imports.
- All regressions include gravity control variables listed above.
- ‚ indicates years for which the ERGM estimation did not convergence.
- Standard deviations are not reported in the estimation tables.
- Sources for calculations: SDC Platinum and Comtrade.

*Italic: References and appendices content extracted from wpiea2019264-print-pdf - References.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2019/wpiea2019264-print-pdf.pdf_