Merger

International Research Journal of Finance and Economics ISSN 1450-2887 Issue 8 (2007) © EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/finance.htm

Trends and Determinants of Mergers and Acquisitions in Developing Countries in the 1990s Ahmed Kamaly Department of Economics, The American University in Cairo 113 Kasr El Aini P.O. Box 2511, 11511 Cairo, Egypt Email: [email protected] Abstract This study sheds some light on the direction and determinants of the aggregate Mergers and Acquisitions (M&A) activity directed to developing countries in the 1990s using a dataset obtained from the SDC Platinum Worldwide Mergers and Acquisitions Databases. The study adopts a dynamic panel data model to gauge the macroeconomic determinants of aggregate M&A. Results indicate that M&A activity embodies a moderate level of inertia, though much less than the one previously estimated for total FDI. A decrease in the international interest rate or an increase in S&P 500 index positively affect M&A indicating the procyclical nature of M&A directed to developing countries. Openness has a significant effect on M&A, but quantitatively its effect is minimal. Depreciation in the domestic exchange rate strongly and positively affects M&A. This result could explain the apparent stability of aggregate FDI even in face of financial turmoil. Finally and interestingly, higher level of stock market activity and depth in developing countries decrease the amount of M&A directed to them. Keywords: Mergers and Acquisitions, Foreign Direct Investment, Dynamic Panel Models, Developing Countries JEL Classification: C23, G15, F23, F32, C52

I. Introduction Until recently, mergers and acquisitions (M&A) activity was an insignificant part of capital flows to developing countries. It was not until the recent surge in capital flows, and the dominance of foreign direct investment (FDI) in the 1990s, that economists started to take a closer look at M&A. However, such a “look” has often been restricted to a simple reporting of the share of M&A activity in total FDI and its growth over time. However, the growing share of FDI in total capital flows directed to developing countries, coupled with the increase in the share of M&A in FDI flows, should have incited economists to study more carefully the behavior as well as the determinants of such activity. Surprisingly enough, the existing literature focusing on the aggregate M&A activity in developing countries is almost nonexistent1. There are several reasons for this, most of which are related to the availability of data and the way M&A are reported and organized. This paper aims at filling the gap in the literature by empirically examining the determinants of aggregate M&A activity to developing countries in the 1990s using the SDC Platinum Worldwide Mergers and Acquisitions Database. In addition, by drawing on the literature pertaining to the 1

A notable exception is the study by Aguiar and Gopinath (2005).

17

International Research Journal of Finance and Economics - Issue 8 (2007)

determinants of FDI, this study offers a more comprehensive and accurate account of the forces shaping the two components of FDI: greenfield investment and M&A. A number of studies have examined the determinants of FDI in the 1990s. However, a number of questions have remained unanswered. This study should be regarded as another step toward unraveling some of these questions. More specifically, dissecting FDI and looking into one of its ingredients provides valuable insights on how internal and external factors affect M&A, greenfield investment, and ultimately FDI. This study is organized as follows. Section two reviews some background literature. Section three discusses the data, the model, and the estimation procedure. Section four presents and analyzes the results, their implications, and their significance. Finally, section five offers concluding remarks.

II. Background Literature The literature on M&A has roots in both financial economics and macroeconomics. If the focus is on the individual firm, then one has to rely on capital budgeting, information asymmetry, and other areas of corporate finance to explain a firm’s decision to acquire another. On the other hand, if the focus is to explain the “aggregate” behavior of M&A activity, then one has to concentrate on broad macroeconomic and financial indicators used to explain aggregate variables, such as investment. Nevertheless, it is possible that only micro data on M&A produces meaningful empirical results, whereas aggregate M&A activity is just a random variable with no distinct behavioral relation. Shughart and Tollison (1984), using annual U.S data and univariate analysis, found that the aggregate merger level is a white-noise process. However, they emphasized that such a result should not be viewed as evidence against the existence of external determinants of aggregate M&A. Indeed, almost all empirical studies that have examined aggregate M&A activity have found evidence that variables such as the cost of capital, stock prices, and measures of aggregate activity significantly influence M&A activity. Table 1 summarizes the dated empirical literature on aggregate M&A. From this table, one could note the following: First, the data sets used in almost all of these studies are more or less variations of Nelson’s (1959) original dataset which defines M&A as the number of U.S mergers in manufacturing and mining with acquired firm assets of at least one million dollars. A few studies have extended the original period, 1895-1955, up to 1979 (Clark et al., 1988a,b; Benzing, 1991; and Benzing, 1992). The last study appearing in this table, Benzing (1993), used another dataset, the Grimm & Co. dataset, covering the period 1964-1986. Second, all of the studies that included stock prices, with the exception of Benzing (1993), reported its positive significant effect on M&A. This result is in line with the expectation hypothesis (Nelson, 1966) and economic disturbance theory (Gort, 1969). According to the former hypothesis, positive expectations about future, manifested by strong economic growth and high stock prices, are conducive to mergers. When stock prices are buoyant, the cost of capital is reduced and the value of the potential acquired firm rises, making the merger more desirable. Economic booms in turn have two effects that encourage M&A activity. First, the value of the firm usually appreciates as a result of dynamic sales potential. Second, production capacity often falls short of increasing demand, driving firms to acquire others in order to cover this gap. Gort (1969), on the other hand, argued in his economic disturbance theory that during periods of high stock prices, there is greater potential for differences in the valuation of stocks between buyers and sellers, which is a catalyst for mergers.

International Research Journal of Finance and Economics - Issue 8 (2007)

18

Table 1: Study

Sample used

Nelson (1959)

Number of US mergers in manufacturing and mining with acquired firm assets of at least $1 million over 1895-1956 period Inter-war period

Weston (1961) Nelson (1966) Melicher et al (1983)

Same as Nelson (1953) but over 1919-61 period Same as Nelson (1959) but over the 1947-77 period

Stock Price +

Interest rate

+

Industrial Production +

Unempl oyment

A

Multiple regression analysis was used. Cross correlation and multiple regression analysis were used This analysis used number of mergers, nominal merger values, and real merger values. Evidence for random walk and AR(1) Multiple regression analysis was used but lagged dependent was not included. Results indicated that mergers activity (i) exhibits no deterministic trend but there exists a stochastic pattern; (ii) is not a random process; and (iii) follows an AR(2) process. M&A activity follows an AR(2) process. AR(2) specification. Signs of variables depend on the period examined. AR(2) combined with deterministic trend specification AR(2) specification

+ +

-

A

-

A2

and

Same as Nelson (1953) but extending the end date to 1979. Data for the 1947-79 period is for acquired firm with assets ≥ $ 10 million.

Polonchek and Sushka (1987)

Same restriction as Shughart and Tollison (1984) but for 1956-77 period. Same as Nelson (1959) but changing the period to 1919-79.

+

Same as (1988a) Same as (1988a)

Clark

el

al.

+

Clark

el

al.

+

-3

A

Benzing (1992)

Same as (1988a)

Clark

el

al.

+

-

-

Benzing (1993)

Grimm & Co. data set covering 1964-86 period.

A

A

-

Shughart Tollison (1984)

Clark el (1988a)

al.

Clark el (1988b) Benzing (1991)

al.

Notes

-

A

+ and – signs denote positive and negative effect on M&A. Whereas “A” symbolizes that the corresponding variable has an ambiguous effect on M&A

Another observation pertaining to Table 1 is the significance of the interest rate in most studies. One could think of different ways by which interest rates could affect merger activity. Perhaps the most direct way is through its effect on the cost of capital. A high interest rate raises the cost of capital which in turn, lowers the level of investment and the attractiveness of mergers between firms4. The literature often refers to this channel as capital market condition (see for example Melicher et al., 1983 and Benzing, 1991). A less direct way by which interest rate could affect M&A is through credit channel. Raising the interest rate leads to credit crunch making financial intermediaries reluctant to finance merger deals. Note that it is quite difficult to draw a distinct line between the expectation 2

Change in capacity was used as a proxy for economic condition. Failure rate was also included in the regression, and it was found to have a negative effect on M&A. 3 For the whole period, the sign of the interest rate was found to be negative; however, the sign turned out to be positive prior to 1950. 4 One of the primary explanations of M&A is the existence of investment opportunities outside the firm. This explanation treats M&A as an external investment, as opposed to the usual internal one.

19

International Research Journal of Finance and Economics - Issue 8 (2007)

hypothesis and the capital market condition hypothesis, since any change in interest rates could be transmitted to movements in the stock prices. The last comment concerns the empirical methodology used in these studies. All studies have used cross correlations and/or simple OLS to analyze M&A. It was not until Clark el al. (1988a) that studies recognized the importance of including lagged dependent variables in the regressions. As for explanatory variables, all of these studies have used current values without attempting to correct for possible endogeneity problems. More recently, di Giovanni (2005) using the same database used in this study has provided empirical evidence on the positive and significant effect of financial market deepening in the acquisitions countries on M&A activity. di Giovanni sample consists of a mix of developed and developing countries; however, splitting the sample to developing and developed countries did not appear to change the main result of the study. The literature on the empirical determinants of M&A directed to developing countries is almost nonexistent. The only study that focuses on M&A directed to emerging markets is the study by Aguiar and Gopinath (2005). Aguiar and Gopinath develop a corporate finance-based theoretical model with testable predictions. Using firm-level data for five East Asian countries over the period 1981-2001, the authors verify that the liquidity crunch, faced by domestic firms as a result of the East Asia crisis, increased M&A activity. Other studies such as World Bank Global Development Finance that deal with capital flows, report M&A directed to developing countries as an aggregate percentage of FDI without analyzing the factors shaping its movement.

III. Data, Model, and Estimation Procedure A. Data Description and Trend The main problem hampering the advancement of the empirical literature on M&A is data availability. This explains the limited number of existing studies that deal with this important issue, and their sole focus on the U.S. This study makes use of a relatively new data source: The SDC Platinum Worldwide Mergers and Acquisitions Database. This database is quite comprehensive as it records actual and potential M&A transactions worldwide. Each record of a transaction usually includes general information about the two sides of the transaction, such as the country, primary SIC code, and balance sheet information as well as general information about the deal such as the value, effective date, and percentage acquired from the target firm. In this study, I focus only on completed transactions5 in which the target nation is a developing country with population greater than one million. Some observations are dropped from the sample due to the unavailability of values for these transactions6. Hence, one should expect that the total value of M&A calculated from this database is biased downwards. The data obtained from this database covers approximately 60 developing countries from 1990 till 1999. However, not every country has observations spanning over the whole period. By aggregating the values of all these transactions according to the previous criteria, one can get the total value of M&A for each year for different developing countries. Figure 1 depicts the total value of M&A to sample developing countries over the 1990s. One can easily discern the strong upward trend of M&A during the 1990s, which tracks more or less the trend of FDI. As shown in the graph, M&A reaches its peak in 1998, before retracting a little bit in 1999. From 1993 till 1998, aggregate M&A activity witnesses a seemingly exponential growth.

5 6

The criterion for a completed transaction is that it has an effective date. As argued by di Giovanni (2005), the distribution and number of deals with no values appear random; hence one should not expect that the missing values would bias the estimation results.

International Research Journal of Finance and Economics - Issue 8 (2007)

20

Figure 1:

$ million

The Trend of M&A over the 1990s 90000 80000 70000 60000 50000 40000 30000 20000 10000 0

Total M&A

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Year Source: Compiled by the author

Figure 2 shows the composition of M&A over the 1990s. M&A transactions are grouped into four categories according to the SIC code of the target company. These categories are: (i) transportation, communication and public utilities; (ii) trade, finance, real estate, and insurance; (iii) services, public administration, and construction; (iv) manufacturing and mining. A few comments are in order. First, the share of trade and the financial sector in total M&A has been growing throughout most of the 1990s. Interestingly, this share grows following both major financial crises of this period; growing in 1995 following the Tequila crisis, and in 1998 following the Asian crisis. Second, on average, the share of manufacturing and mining declines over the sample period, and seems to shrink right after major financial and currency crises. Third, the total share of the tradable sector has been shriveling over the sample period relative to nontradable sector. Again, this movement appears to be more pronounced right after major financial and currency crises.

21

International Research Journal of Finance and Economics - Issue 8 (2007) Figure 2: 100%

80% Transportation, Comm & Public utilities

60%

40%

Trade, Finance, Real estate & Insurance Services, Public adm. & Construction Manufacturing & Mining

20%

0% 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Source: Compiled by the author

A couple of reasons could be offered to explain these changes in the composition of M&A activity and their apparent sensitivity to financial and currency crises. First, as countries develop, the share of services and finance, which are a part of the nontradable sector, expand relative to the goods sector, which are mostly tradable. This change in the composition of profitable investment opportunities encourages foreign firms to increase their holdings of domestic assets belonging to the nontradable sector. Second, in times of financial distress and currency crash, the tradable sector is more able than the nontradable sector to weather the crisis. This is simply because, by definition, companies operating in the tradable sector can turn to the rest of the world to exhaust the accumulated production resulting from the fall in domestic demand. In fact, Lehmann (2002) finds that in Mexico during the Tequila crisis, and in Korea, Thailand, Malaysia and Indonesia during the Asian crisis, the revitalization in the export market and the rapid direction of sales away from depressed local markets have eased, and in some cases, mitigated the huge drop in local sales; thus, helping these firms to avoid bankruptcy and the liquidation of their assets. Firms operating in the nontradable sector, on the other hand, are not as fortunate. The massive drop in local demand coupled with credit rationing, which usually characterizes financial crises, take their toll on nontradable sector firms pushing them toward liquidation and bankruptcy. This, coupled with the collapse of the local currency, drives the market value of these firms to an abyss. To multinational corporations, these bargains are not to be passed up! This situation leads to what is known now in the literature as “Fire-sale” FDI (Krugman, 1998). This could explain the observed turn in the composition of M&A toward the nontradable sector following the Tequila crisis of 1994 and the Asian crisis of 19977. B. Model Following most of the empirical studies that attempt to examine the macro determinants of FDI, (see for example Edwards, 1992; Calvo and Reinhart, 1998; Mercereau, 2005), this study adopts a reduced 7

This observation somewhat contradicts one of the findings of Aguiar and Gopinath (2005) where they find evidence that during crisis period, the effect of liquidity on the probability of acquisition is stronger among firms operating in the tradable sector. One should note however, that the sample used in Aguiar and Gopinath’s study is restricted to five East Asian countries.

International Research Journal of Finance and Economics - Issue 8 (2007)

22

form model. Nevertheless, the selection of the variables included in the model is not arbitrary but rather guided by theory and other empirical studies as explained below. The model used in this study is a dynamic panel model with a fixed-effects specification in the form of: yit = α + µ i + δyit −1 + xit' β + uit

u i ~ iid (0, σ ) 2 u

(1)

i = 1,2,...N , t = 1990,....1999

Where y it : Total M&A to GDP ratio

xit' : The matrix of explanatory variables besides the lagged dependent variable N: Total number of countries µ i : Country individual effect α , δ and β are unknown parameters u it : Error term It is assumed that the error term uit follows a one-way error component model with constant variance σ u2 . The fixed-effects representation captures the idea that countries, in most cases, have individual specific effects such as institutional settings, geographical characteristics, and cultural norms, which influence M&A but could be regarded as fixed in the short and medium terms. The endogenous variable in this model is taken to be the total M&A to GDP ratio. All the studies appearing in Table 1 with the exception of Shughart and Tollison (1984) use the number of mergers as the dependent variable. Using the value of M&A transactions is based on the perception that a great deal of information is lost if one uses only the number of successful M&A. Using values amounts to weighing each observation according to its size, while the number of transactions itself provides no information about the absolute or relative flows of capital entering a particular country in the form of M&A. I use the ratio of M&A to GDP instead of simply the aggregate value of M&A for two reasons. First, to make sure that the endogenous variable is stationary; and second, to control for the size of the target nation. The list of explanatory variables besides the lagged dependent variable in the base regression includes: the weighted average bond yield in the G7 countries, the lagged change in the S&P500 index, and the ratio of the sum of imports and exports to GDP. The existing literature has clearly established the first two variables as important determinants of M&A. Note that the interest rate measure includes G7 countries not only the US interest rate. The rational is the following. Acquiring firms are usually multinationals, so interest rates in industrialized countries not only in the US should play a role in discounting the net flows of the acquired firm or calculating the cost of capital. The bond yield, on the other hand, was chosen to reflect the fact that M&A activity is a long term commitment and the revenue associated with it usually spread over a relatively long period of time. The new variable to the M&A regressions is the ratio of the sum of imports and exports to GDP. This is a common measure of openness of the economy. Many empirical studies, such as Edwards (1992), Singh and Jun (1995), Bathattacharya, Montiel and Sharma (1997) and Kamaly (2003) have confirmed the significance of this variable in driving FDI flows. Recently, di Giovanni (2005) has found evidence for the complementarity between openness and M&A. Finally, I should note that my regressors originally included real GDP growth, which was then dropped due to its insignificance.

C. Estimation Procedure The dynamic panel model used in this study calls for a special type of estimation to deal with the problems generated by including the lagged dependent variable in the regression. In these models, if the usual panel regression techniques, such as the Within estimator, are used the results could suffer from inconsistency and bias, especially if the time dimension is small and the cross section dimension

23

International Research Journal of Finance and Economics - Issue 8 (2007)

is large, which is true in our case8. The recent development of dynamic panel models has provided a number of solutions for this problem. Among the first suggested solutions is the one introduced by Arellano and Bond (1991) which is often referred to in the literature as the GMM-IV or the standard GMM estimator. This estimator makes use of the orthogonality conditions that exist between the instruments and the model after some type of transformation (e.g. first differences or orthogonal deviations). This study, however, adopts another GMM estimation technique based on Arellano and Bover (1995) and Blundell and Bond (1998), which builds and improves upon the standard GMM estimator. In addition to the moment conditions associated with the standard GMM estimator, this estimator employs additional restrictions based on the fact that any available instrument that is not correlated with the group effects µ i can serve as an instrument for (1) in levels without any transformation. Combining all these restrictions in one system and using GMM produces a more efficient estimator the GMM-SYS estimator. Despite the fact that efficiency is gained from additional moment restrictions, having “too many” instruments per se is not desirable. Since the sample size is relatively small, a very large number of instruments could result in a small sample bias (Kiviet 1995). This is why not all of the available instruments were used in the regressions. Increasing the number of instruments did not change the coefficients estimates much, but did raise the level of significance in a handful of them9. The transformation used in the estimation is orthogonal deviations10 as opposed to first differencing. The advantage of this transformation is that autocorrelation between transformed errors will be absent if it is absent among the original errors. In fact, the orthogonal deviations procedure is equivalent to first differencing to get rid of fixed effects, and then using GLS to eliminate first degree autocorrelation resulting from first differencing (Arellano and Honoré, 2001).

IV. Empirical Results and Implications This section presents and analyzes the results obtained from estimating (1) using the estimation procedure outlined above. In all of the presented estimated equations, the GMM-SYS estimation procedure was used along with the orthogonal deviations as the method of transformation11. Only the two-step results are shown12; however the one-step results are almost identical. First, a base regression was estimated with the lagged dependent variable, openness, the weighted average bond yield, and the change in S&P500 as explanatory variables (see data appendix for variable definitions). The results of this regression are shown in the second column of Table 2. As apparent from the results, all the explanatory variables are significant at least at the 5% level, and carry the expected signs. The Wald test indicates the joint significance of the explanatory variables. The Sargan test confirms the validity of the imposed moment conditions. The first and second order tests for autocorrelation of the transformed residuals point to the presence of first order negative serial correlation and the absence of second order serial correlation, both of which indicate the absence of serial correlation among the original errors u it . This last condition is a crucial one for the consistency of the GMM-SYS estimator (Arellano and Bond 1991).

8

See Nickell (1981) and Anderson and Hsiao (1981) for more details. Results of these regressions are available upon request. 10 This is a transformation introduced by Arellano (1988), and Arellano and Bover (1995) where each observation is transformed into a weighted 9

* z it =

⎛ ⎜⎜ z it ⎝

−

z i , t +1 + ... + z i ,T

⎞⎛ T − t ⎞ ⎟⎟⎜⎝ T − t + 1 ⎟⎠ ⎠

12

T − t deviation from the average of future observations for the same group: t = 1,2,3..., T − 1 Note that this transformation also standardizes the variance. 11 Estimates based on first differencing are quite similar to those based on orthogonal deviations; however, using orthogonal deviations has improved precision. 12 Note that the two-step results are more efficient if errors are heteroskedastic.

International Research Journal of Finance and Economics - Issue 8 (2007)

24

Table 2: Results of Estimated Equations Explanatory Variables

Lagged M&A/GDP Openness Bond yield in G7 countries Change in S&P500 Additional Regressor Number of observations Number of countries

Wald χ 2 Sargan’s overidentifying test (P-Value) A & B test of 1st order autocorrelation (P-Value) A & B test of 2nd order autocorrelation (P-Value)

Base Regression

348 60 117.5***

Lagged Index of Exchange Rate Variability 0.280 (2.33)** 0.01 (3.20)*** -0.083 (-2.50)** 0.026 (2.78)*** -0.001 (-0.446) 348 60 129.1***

56.95 (0.997) -2.634 (0.00)*** 1.258 (0.209)

54.42 (1.00) -2.659 (0.01)*** 1.154 (0.248)

0.318 (2.52)** 0.011 (3.62)*** -0.103 (-3.31)*** 0.027 (2.60)*** .

Lagged Market Capitalization to GDP

0.290 (2.30)** 0.009 (1.89)* -0.109 (-3.10)*** 0.026 (2.77)*** 0.005 (0.548) 337 60 105.6***

Lagged Total Value of Stocks traded to GDP 0.160 (2.15)** 0.009 (2.19)** -0.069 (-2.21)** 0.038 (3.35)*** -0.006 (-2.15)** 274 45 98.92***

0.123 (1.66)* 0.011 (2.28)** -0.065 (-2.13)** 0.036 (3.11)*** -0.007 (-1.83)* 284 47 97.25***

Lagged Change in Nominal Exchange Rate 0.284 (2.34)** 0.012 (3.39)*** -0.121 (-3.55)*** 0.028 (2.51)** 0.037 (3.42)*** 340 60 128.0***

53.76 (1.00) -2.600 (0.01)*** 1.354 (0.176)

35.50 (1.00) -2.391 (0.02)** 1.136 (0.256)

37.76 (1.00) -2.494 (0.01)** 0.596 (0.551)

57.64 (0.99) -2.592 (0.01)*** 1.105 (0.269)

Lagged M2 to GDP

Notes: t-statistics are in brackets. *** Significant at 1% level or more, ** significant at 5% level or more, * significant at the 10% level or more. The t-statistics of these estimates are based on robust standard errors, after correcting for small sample bias, as suggested by Windmeijer (2005).

Turning to the point estimates, one notices that inertia plays an important role in driving M&A to developing countries. The coefficient of the lagged dependent variable is approximately 0.31, the largest among all explanatory variables, implying that the long-term effects of other explanatory variables are 1.5 times their short-term effects. This inertia is, however, much less than that found for aggregate FDI, where the coefficient of the lagged dependent variable in GMM-SYS specifications is found to be around 0.7 as reported in Kamaly (2003) and Amaya and Rowland (2004). It may be the case that the other component of FDI- greenfield investment- is more sluggish than M&A. This result is consistent with the natures of greenfield investment and M&A. One would normally expect that new projects are completed over a relatively long period of time. M&A transactions, on the other hand, are just a transfer of ownership of already existing operating projects from domestic control to foreign hands. The measure of openness, the ratio of sum of imports and exports to GDP, is found to be statistically significant but with a little quantitative significance. Interestingly, the magnitude of this coefficient (0.01) is almost identical to that reported for the same variable in Kamaly (2003) for FDI flows. This points to the strong association between FDI with its components and the degree of openness. The interest rate coefficient is statistically significant, and its magnitude is found to be the second largest among the included explanatory variables. According to the point estimate, a one percent point increase (decrease) in the interest rate measure decreases (increases) the M&A to GDP ratio by 0.1 percent in the short-term and by approximately 0.15 percent in the long-term. The significance of this variable is in line with existing evidence on the effect of the interest rate on M&A activity (See Table 1). The last explanatory variable in the base regression is the change in the S&P500 index. As depicted in Table 1, stock prices have a significant positive effect on M&A activity in the majority of existing empirical studies. Similarly, the change in the S&P500 index has a positive significant effect on aggregate M&A directed to developing countries. According to the point estimate, a one percent

25

International Research Journal of Finance and Economics - Issue 8 (2007)

increase in the S&P500 index raises the ratio of M&A to GDP by 0.02 percent. This is not a negligible effect, since the S&P500 index could very well register a two-digit percentage change over a couple of months. However, one has to recognize that the significance of stock prices in the base regression implies something different than its significance in the studies summarized in Table 1. In the latter case, both the M&A and stock prices refer to the U.S. In this study, however, the target companies are developing countries’ firms, while stock prices correspond to the U.S. In this case, one cannot readily apply the expectation hypothesis, since buoyant stock prices in the U.S do not necessary suggest favorable economic conditions in developing countries. Nevertheless, the significance of the change in the S&P500 index suggests that higher stock prices in the industrialized world13 have a positive spillover on M&A activity directed to developing countries. There are a couple of explanations that could be offered to account for this observed positive relation between M&A activity in developing countries and the S&P 500. One is that high US stock prices could be an indication of a present or future boom in the US (Fama, 1981). During such times, additional capacity is often needed. A possible solution is to acquire firms from developing countries, which can be obtained at a discounted price compared to the firms in industrialized countries. Another possible explanation is a spillover effect. The high price of stocks in developed countries enables firms to secure more funds for investment, part of which can be used to pursue external investment opportunities such as M&A. Next, the effects of some additional explanatory variables on M&A flows to developing countries are examined. These variable are: the lagged index of exchange rate variability, the lagged ratio of M2 to GDP, the lagged ratio of the total value of domestic stocks traded to GDP, the lagged ratio of domestic market capitalization to GDP, and the lagged change in the nominal exchange rate. All these variables could potentially have some effects on M&A activity through making a given developing country more attractive or less attractive to foreign investors. One at a time, each of these variables is added to the base regression. The third to the eighth columns of Table 2 depict the results of these regressions14. Before delving into the specifics of each of these regressions, a couple of general observations are in order. First, the inclusions of additional variables have little impact on the point estimates and statistical significance of the variables from the base regression. Only when domestic stock market indicators are included, the estimates of some of these coefficients deviate somewhat from the values observed in the base regression which could be attributed to the drop in the sample size as explained later. Second, other statistical criteria point toward the consistency of the estimation procedure. The Sargan test indicates the validity of the used instruments in all regressions. The results show a significant negative first order serial correlation and the absence of second order serial correlation among residuals, indicating that the disturbances are not serially correlated. Finally, the Wald test points to the joint significance of explanatory variables in each estimated regression. The first variable is the lagged index of exchange rate variability. This variable could be viewed as a proxy for macroeconomic risk since a high variability in the nominal exchange rate could result in increasing the level of uncertainty in the economy15. Indeed, it was found to have a slight negative effect on FDI flows (Kamaly, 2003). According to Table 2, the coefficient of the lagged index of exchange rate variability took the expected negative sign but suffered from a low level of significance. Similarly, the proxy for financial development, the M2 to GDP ratio, turned out to have a positive but insignificant effect on M&A (fourth column in Table 2). The expectation hypothesis postulates a positive relationship between M&A activity and a bustling domestic stock market. Including the change in S&P500 index does not put this conjecture to the test since this index is external to developing countries’ stock markets. To examine this link, two proxies for the level of activity of the domestic stock market are included in the regression: the total 13

The S&P500 index includes mostly multinational companies. In addition, the S&P500 index is highly correlated with other stock market indices in the rest of the industrialized world. 14 Two political variables, democracy index and durability of the political regime, were tested but they turned out to be totally insignificant. 15 Another proxy for macroeconomic risk, the index of interest rate variability, also turned out to have a negative but insignificant effect on M&A.

International Research Journal of Finance and Economics - Issue 8 (2007)

26

value of stocks traded as a percentage of GDP and market capitalization of listed companies as a percentage of GDP16. The inclusion of these variables led to a decrease in the sample size of almost 20% due to the unavailability of stock market data for a number of countries in the sample. The results of these regressions are summarized in the fifth and sixth column of Table 2. Surprisingly, the coefficients of these two variables are negative and statistically significant, especially the total value of stocks traded. This is an unexpected result since it implies that stock market booms have a negative effect on M&A activity. It is true that a substantial amount of M&A transactions are conducted outside the stock market (Melicher et al., 1983); however, this observation would imply that these indicators should have an insignificant effect on M&A. One could think of two possible justifications for this result. First, it is often the case that a slowdown in the level of activity of the stock market is associated with a similar drop in the stock market return or stock prices. This depreciation in the stock prices may create some valuable investment opportunities, especially when prices reach “fire-price” levels. Another possible explanation appeals to the observed inverse relationship between different types of capital inflows, especially during the second half of the 1990s (Kamaly, 2003). Specifically, one could think of a situation where a foreign investor wishes to invest in a certain sector in a given developing country. If the stock market is underdeveloped, then most probably our investor may decide to acquire some shares in a company within this sector, which is in most cases classified as an M&A transaction. The converse is also possible. If the stock market improves in terms of depth and development, then our investor would be inclined to invest in the stock market rather than directly acquire equity shares in some companies (Fernández-Arias and Hausmann 2000). This significant negative effect of domestic stock market indicators and the insignificance of real GDP growth suggest that aggregate M&A activity does not follow the expectation hypothesis. On the other hand, the negative effect of the interest rates on M&A activity to developing countries should not be regarded as an evidence for the cost of capital hypothesis, since the hypothesis links aggregate M&A activity to the prevailing domestic interest rate not the international interest rate. The significance of both the international interest rate and the change in the S&P500 index supports the push story where changes in industrialized countries’ macroeconomic variables could bring about changes in the flow of capital directed to developing countries. Similar to aggregate FDI, M&A flows seem to be procyclical in nature as lower international interest rates and bulls market in the US encourage M&A activity directed to developing countries. A number of empirical studies have tried to verify the link between FDI and the exchange rate, see for example Caves (1989), Froot and Stein (1991), Swenson (1994), and Blonigen (1997). Most of these studies have confirmed the existence of such a relationship. To verify the existence of such a link and to determine its direction, the lagged change in the nominal exchange rate was added to the base regression. The results of this regression are depicted in the seventh column of Table 2. The coefficient of the lagged change in the exchange rate is positive and highly significant. This result is in line Froot and Stein (1991), and Gastanaga, Nugent and Pashamova (1998), who argue that a depreciation of the host’s currency makes its firms look cheaper in the eyes of foreign investors. In addition, this result is consistent with the course of M&A activity which soared during the time of the Asian crisis in 1997 and 1998. This positive association between M&A and depreciation is likely to be driven by the effect of financial and currency crises on the nontradable sector, as argued in the previous section and supported by changes in the composition of M&A as depicted in Figure 2. This association between M&A activity and depreciation could be used to elucidate one of the puzzles surrounding FDI flows in the 1990s; namely, the stability of FDI, as opposed to other types of capital flows, in the face of financial turmoil. Despite this buoyant track of M&A during and after the Asian crisis, FDI flows were stagnant. In fact, the same variable, the lagged change in nominal exchange rate was found, to have a negative effect on FDI flows in the 1990s (Kamaly, 2003). Juxtaposing these results and observations, one could deduce that a depreciation in the host’s currency 16

The number of listed companies was also considered and yielded similar results but with a lower level of significance.

27

International Research Journal of Finance and Economics - Issue 8 (2007)

props up M&A, but reduces greenfield investment17. Since greenfield investment constitutes more than 60% of FDI flows, the combined effect of a change in the nominal exchange rate is slightly negative.

V. Conclusions Using a comprehensive dataset on M&A to estimate a dynamic panel data model, this study sheds some light on the direction and determinants of aggregate M&A activity directed to developing countries in the 1990s. Results indicate that aggregate M&A activity embodies a moderate level of inertia, though much less than the one bolstering total FDI to developing countries. Interest rates do affect M&A in the anticipated negative direction. Openness has a significant effect on M&A but, quantitatively, its effect is minimal. Increases in S&P 500 index seem to abet the demand for developing countries’ firms. The trend of aggregate M&A activity throughout the 1990s and the positive effect of depreciation on M&A explain the exuberant M&A transactions- “fire-sale FDI”- that coincided with the Asian crisis. The mélange of depressed domestic demand, financial crunch, and sinking local currency caused a dramatic fall in the value of local firms-especially in terms of dollars-, which led foreign investors to stampede to acquire these bargain deals. This positive association between aggregate M&A activity and the nominal exchange rate, together with the reported weak negative relationship between the depreciation and FDI flows, lead us to suspect a strong negative connection between greenfield investment and the change in nominal exchange rate; however, further research is needed to verify this conjecture. The level of activity in the domestic stock market appears to deter M&A to developing countries. This is a somewhat unexpected result since empirical evidence using US data has often confirmed the existence of a positive link between stock returns and M&A. Two explanations were offered to reconcile this result: First, if the level of activity of the stock market moves in tandem with its return, then it could be the case that foreign investors decide to acquire developing countries’ companies when domestic stock markets plunge. Another explanation is related to the investor’s choice of the type of investment, where investors faced with an underdeveloped stock market are more likely to opt for acquiring equity directly through M&A; and vice versa.

Acknowledgement I’m grateful to Carmen Reinhart, Michael Binder, Fernando Broner, Harry Kelejian, and Guillermo Calvo for refining my ideas, and useful suggestions through the different stages of my research. I also benefited from discussions and comments from Sergio Schmukler, Daniel Ortega, Federico Guerrero, John Shea, John Haltiwanger, and other participants of the international breakfast seminar and Macro/International seminar in the University of Maryland at College Park.

17

Aguiar and Gopinath (2005) have verified that M&A activity increases during crisis; however, the mechanism is totally different. They argue that domestic firms face a liquidity constraint that intensifies during financial crisis which adversely affects firm’s investment opportunity and hence its value. Foreign firm on the other hand, does not face this liquidity constraint; moreover, it has access over a superior technology. Consequently, foreign firm can extract rent by acquiring domestic firm especially during crisis time when the value of the domestic firm drops even further.

International Research Journal of Finance and Economics - Issue 8 (2007)

28

References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21]

[22]

Aguiar, M and G. Gopinath (2005) “Fire-Sale Foreign Direct Investment and Liquidity Crises” Review of Economics and Statistics 87(3), pp. 439-52. Amya, C and P. Rowland (2004) “Determinants of Investment Flows into Emerging markets” Borrodores de Economia, no. 313, Banco de la Republica de Columbia. Anderson, T.ZW and C. Hsiao (1981) “Estimation of Dynamic Models with Error Components” Journal of the American Statistical Association 76, pp. 598-606. Arellano, M. (1988) “An Alternative Transformation for Fixed Effects Models with Predetermined Variables” Applied Economics Discussion Paper, Institute of Economics and Statistics. Arellano, M. and S. Bond (1991) “Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations” Review of Economics Studies 58(2) pp. 277-97. Arellano, M. and O. Bover (1995) “Another Look at the Instrumental Variables Estimation of Error-Components Models” Journal of Econometrics 68, pp. 29-51. Arellano, M. and B. Honoré (2001) “Panel Data Models: Some Recent Developments” Handbook of Econometrics, volume 5, Elsevier Science, North Holland. Bathattachaary, A. Montiel, P and S. Sharma (1997) “Private Capital Flows to Sub-Saharan Africa: An Overview of Trends and Determinants” in Z. Iqbal and R. Kanbur eds., External Finance for Low-Income Countries, IMF. Benzing, C (1991) “The Determinants of Aggregate Merger Activity- Before and After CellerKefauver” Review of Industrial Organization 6, pp. 61-72 Benzing, C (1992) “The Macroeconomic Determinants of Merger Activity” New York Economic Review, Spring, pp. 3-14. Benzing, C (1993) “Mergers- What the’Grimm’ Data Tell Us” Review of Industrial Organization 8, pp. 747-53. Blonigen, B. (1997) “Firm-Specific Assets and the Link Between Exchange Rates and Foreign Direct Investment” American Economic Review 87(3), pp. 447-65. Blundell, R and S. Bond (1998) “Initial Conditions and Moment Restrictions in Dynamic Panel Data Models” Journal of Econometrics 87, pp. 115-43. Calvo, G., and C. Reinhart. (1998) “The Consequences and Management of Capital inflows: Lessons from Sub-Saharan Africa” Expert Group on Development Issues, 2. Caves, R.(1989) “Mergers, Takeovers, and Economic Efficiency: Foresight vs. Hindsight” International Journal of industrial Organization 7(1), pp. 151-74. Clark, J., A. Chakrabarti, and T. Chiang (1988a) “Trend and Stochastic Movements in the U.S Merger Activity” Quarterly Review of Economics and Business, 28, pp. 5-18. Clark, J., A. Chakrabarti, and T. Chiang (1988b) “Stock Prices and Mergers Movements: Interactive Relations” Weltwirtschaftliches Archiv, 124(2), pp. 287-300. DPD Package Ver. 1.2 by Doornik, J. M. Arellano and S. Bond (2001), www.nuff.ox.ac.uk/users/doornik/ Edwards, S. (1992) “Capital Flows, Foreign Direct Investment, and Debt-Equity Swaps in Developing Countries” In H. Seibert (ed). “Capital Flows in the World Economy” Institut Für Weltwirtschaft an Der Univerität Kiel. Fama, E. (1981) “Stock Returns, Real Activity, Inflation and Money” American Economic Review 7, pp. 545-65. Fernández-Arias, E. and R. Hausmann (2000) “Is FDI a Safer Form of Financing” Paper Prepared for the Seminar: The New Wave of Capital Inflows: Sea Change or Just Another Tide? Annual Meetings of the Board of Governors, Inter-American Development Bank and Inter-American Investment Corporation, Working Paper 416. Froot, K and J. Stein (1991) “Exchange Rates and Foreign Direct Investment: An Imperfect Capital Markets Approach” Quarterly Journal Economics 106(4), pp. 1191-1217.

29 [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41]

International Research Journal of Finance and Economics - Issue 8 (2007) Gastanaga, VC., J. Nugent and B. Pashamova (1998) “Host Country Reforms and FDI Inflows: How Much Difference Do They Make?” World Development 26(7), pp. 1299-314. de Giovanni, J. (2005) “What Drives Capital Flows? The Case of Cross-border M&A Activity and Financial Deepening” Journal of International Economics 65, pp. 127-49. Gort, M. (1969) “An Economic Disturbance Theory of Mergers” Quarterly Journal of Economics 83, pp. 624-42. Kamaly, A (2003) “What Derives FDI to Developing Countries in the 1990s? An Empirical Investigation” Working Paper, Department of Economics, The American University in Cairo. Kiviet, J. (1995) “On Bias, Inconsistency, and Efficiency of Various Estimators in Dynamic Panel Data Models” Journal of Econometrics 68, pp. 53-78. Krugman, Paul (1998). “Fire Sale FDI” mimeo, MIT. Lehmann, A (2002) “Foreign Direct Investment in Emerging Markets: Income, Repatriations and Financial Vulnerabilities” International Monetary Fund, IMF Working Papers, no 02/47, Washington DC. Melicher et al (1983) “A Time Series Analysis of Aggregate Merger Activity” Review of Economic and Statistics 65, pp. 423-29. Mercereau, B. (2005) “FDI Flows to Asia: Did the Dragon Crowd Out the Tiger?” International Monetary Fund, IMF Working Papers, no 05/189, Washington DC. Nelson, R. (1959) Merger Movements in American Industry: 1895-1965, Princeton. Nelson, R. (1966) “Business Cycle Factors in the Choice between Internal and External Growth” in W. Alterts and J. Segall eds. “The Corporate Merger” Chicago: University of Chicago Press. Nickell, S. J. (1981) “Biases in Dynamic Models with Fixed Effects” Econometrica, 49(6), pp. 1417-26. Polonchek J. and M. Sushka (1987) “The Impact of Financial and Economic Conditions on Aggregate Merger Activity” Managerial and Decision Economics 8, pp. 113-9. Shughart, W. and R. Tollison (1984) “The Random Character of Merger Activity” Rand Journal of Economics 4, pp. 500-9. Singh, H. and K. Jun (1995) “Some New Evidence on Determinants of Foreign Direct Investment in Developing Countries” Policy Research Working Paper # 1531, World Bank, Washington DC. Swenson, D (1994) “The Impact of U.S Tax Reform on Foreign Direct Investment in the United States” Journal of Public Economics 54(2), pp. 243-66. Weston, J. (1953) The Role of Mergers in the Growth of Large Firms, University of California Press. Windmeijer, F. (2005) “A Finite Sample Correction for the Variance of Linear Two-Step GMM Estimators” Journal of Econometrics 126(1), pp. 25-51. World Bank Global Development Finance, several issues.

International Research Journal of Finance and Economics - Issue 8 (2007)

30

Data Appendix M&A data is obtained from SDC Platinum Worldwide Mergers and Acquisitions Databases. However, the main source of economic data is the World Economic Outlook (WEO) database (IMF). The data for the exchange rate and interest rate is obtained from the International Financial Statistics (IFS) database (IMF). Domestic stock market indicators are obtained from World Bank Development Indicators database. As for the political data, it is obtained from the Center for International Development and Conflict Management (CIDCM) at the University of Maryland at College Park. Below is a description and the derivation of variables used in this study. 1. Weighted average of Bond yield in the G7 countries: This variable is calculated as a weighted average of bond yield in the Group of Seven countries (G7) taking their GDP as weights. 2. Openness: This variable is defined as the ratio of the sum of export and import to GDP. 3. Index of exchange rate variability: This variable is calculated from the series of nominal monthly exchange rates. For each country, the rate of change of monthly exchange rate was used to compute its variance. For each year, these figures for variance are used to construct an index of exchange rate variability. This index varies from 1 to 4 (1 indicating the least variability and 4 indicating the highest variability). 4. Change in nominal exchange rate: This variable is calculated by taking the average of the change in the nominal monthly exchange rate. 5. Index of interest rate variability: This variable is calculated from the series of nominal monthly (deposit) interest rates. For each country, the rate of change of monthly interest rate was used to compute its variance. For each year, these figures for variance are used to construct an index of interest rate variability. This index varies from 1 to 4 (1 indicating the least variability and 4 indicating the highest variability). 6. Inflation rate: This variable is calculated as the rate of change of Consumer Price Index (CPI). 7. Democracy index: This index varies from –10 (strongly autocratic political system) to 10 (strongly democratic political system). 8. Durability of the political regime: This represents the number of years since the most recent regime change.