Introduction
A politically connected firm is a company that maintains direct or indirect ties to individuals with past or present influence over public policy, regulation, or the allocation of state resources. These connections may arise through personal relationships, employment histories, educational networks, financial contributions, or board memberships. To be politically connected implies that a firm is strategically embedded within a network of influence that can shape, accelerate, or shield its business activities through preferential access to state power.
These connections can hold, under certain circumstances, the potential to benefit the company in several ways. Political connections may facilitate access to government contracts, licenses or favorable regulation, mitigate political or regulatory uncertainty, serve as a signal of credibility or influence, or enable various rent-seeking behaviors (such as subsidies, loans, or protection from competition). Literature shows that politically connected firms tend to be larger, more leveraged, and have greater market share in their industries than non-connected firmsFootnote 1 . Interestingly, these very features of politically connected firms make them more vulnerable to a variety of reform-oriented policies such as greater antitrust enforcement and financial and trade liberalization.Footnote 2 If political connections can be relatively harmful in reform-oriented policy environments, do they protect firms in environments characterized by potentially negative government actions? I explore this question in the context of Argentina’s renationalization of public pension funds in 2008.
Political connections can gain and lose their value with shifts in the political or economic environment. At the extreme, well-connected firms in authoritarian regimes suffer considerable losses when those regimes fall. For example, in Suharto-era Indonesia, firms connected to the Suharto family experienced much greater losses than non-connected firms when Suharto’s health declined, showing that a significant share of their worth came from ties to power, so the potential loss of those political connections reduced firm valuation.Footnote 3 More commonly, changes in political regimes can turn positive political capital into a liability when regimes that come into power expropriate the power of previous regimes.Footnote 4 The question that is less understood is how politically connected firms are affected by adverse policy changes when regimes do not change. When political or economic shocks move existing governments to whom firms are closely connected to adopt populist and/or anti-business measures, political connections may not protect firms from incurring losses.
Governments have varying reasons for enacting policies that harm the corporate sector. The push and pull of populist and market-oriented policies is common in emerging markets and has been widely documented in both Asia and Latin America.Footnote 5 Electoral politics, attempts to reestablish legitimacy, and the need for new sources of funding are key reasons why governments, including those that otherwise pursue market-oriented agendas, may adopt populist or anti-business measures. These measures range in severity from modest “get tough” policies such as the European Union’s imposition of caps on bonuses paid to bankers, to outright seizure of property or nationalizations which have occurred in countries like Russia, Venezuela, Bolivia, and Zimbabwe. The common thread across these policies is that their adverse effects tend to fall disproportionately on business interests and owners of capital.
While firms in most industries can be affected by adverse policies, private pension systems have particularly high political risk, because they are a potential financial resource for governments, especially during times of economic stress.Footnote 6 Another important source of political risk faced by private pension fund managers is government pressure to buy sovereign debt.Footnote 7 Furthermore, the inter-temporality of pension systems, in which people pay today to receive funds in the future, makes them especially prone to opportunistic behavior and requires strong supporting institutional environments.Footnote 8 When economies are struggling, and political institutions are weak, private pensions are at risk. In this setting, at the end of 2008, in the context of a global financial crisis, it was relatively easy for the government of the Argentine government of Cristina Kirchner to nationalize the Argentine private pension system.Footnote 9
Context: the renationalization of the Argentine pension system, 2008-2011
In 2008, the Argentine government nationalized the country’s private pension system, which had been created in 1993. At the time, there were ten private pension fund managing companies (called “AFJPs”—Administradoras de Fondos de Jubilaciones y Pensiones) which were forced to cease to operate. The government seized all the financial assets previously managed by these AFJPs. Those financial assets included shares in more than forty publicly traded firms in which AFJPs had invested. In simple terms, these companies became mixed enterprises, with the Argentine government as a minority shareholder.
The system created in 1993 that gave birth to AFJPs was a mixed state-private retirement system, inspired by the success of the model previously created in Chile. The system consisted of two pillars: workers contributed eleven percent of their gross salary to the AFJPs, while employers paid an additional seventeen percent to a state “pay-as-you-go” (PAYG) system with wage-related contributions and flat benefits. While the eleven percent contribution was accumulated in each worker’s account, the 17 percent contribution went to pay the pensions of current retirees. The creation of AFJPs did not terminate the old government system. All retirees at the time of the creation of the new system continued to receive their pension payments from the government, while current workers had a time window to decide to which system they wanted to belong. AFJPs became the largest institutional investors in the country, accumulating tens of billions of dollars in funds. The performance of the AFJP investments was satisfactory if compared with the state PAYG system. According to some estimates, the risk-adjusted returns were 2.4 times higher (145 percent vs. 61.8 percent) than those of the state system (if social security contributions to the state system were treated as quasi-investments).Footnote 10 However, the performance of the Argentine system was clearly below that of its Chilean counterpart, partially due to a suboptimal asset allocation and higher administrative costs (See Table 1).Footnote 11
Table 1: Comparison between Argentine and Chilean pension systems
The portfolios in Argentina were highly concentrated in public debt (58 percent of portfolio in government bonds, versus 8 percent in Chile). This was especially harmful for the value of the portfolios, since in 2001 Argentina declared a default of its sovereign debt (the largest sovereign default in history at the time). This produced steep losses to bondholders, including the millions of contributors to the AFJP system. Another problem was the intrusive investment regulation that constrained AFJPs. Foreign investments, for instance, were limited to 20 percent of the portfolios, increasing vulnerability to domestic volatility. On top of that, a law guaranteed a minimum yield determined as the average of the industry minus a permitted deviation; this encouraged ‘herd behavior’Footnote 12 , since portfolio managers did not want to risk deviating from industry averages. As a result, more than half of AFJPs foreign assets were invested in only two companies.Footnote 13 This inefficient asset allocation was also due to pressures from the Argentine government, always in need of fresh funds. As the Argentine fiscal deficit increased and the government grew more desperate for money, prior to its 2001 default, AFJPs were used as last resort lenders and forced to buy sovereign bonds that were defaulted soon thereafter.Footnote 14 Despite these setbacks, future Argentine retirees still seemed to trust AFJPs more than they trusted their government. In 2007, the Argentine government opened the option to voluntarily move back to the state system. The vast majority of affiliates (9.5 million out of 11) chose to remain in the AFJP system.Footnote 15
As the 2008 financial crisis spread across global financial markets, the value of retirement accounts dropped, which provided an opportunity for the government to criticize the entire pension system and the AFJPs’ investment strategies (conveniently omitting the fact that most of the investments were in government bonds). As the optional change to the state system had failed to produce a massive migration, in October 2008 the Argentine government took the drastic step of announcing the intent to end the private pension system altogether. The official argument for this decision was the protection of future retirees from the AFJPs’ poor management of pension funds. The real reason, however, appeared to be the need for funds for fiscal purposesFootnote 16 and avoiding debt payments by the government to the private pension funds.
Rumors leaked to the press on October 20th, 2008, and on the 21st the government officially announced at a press conference that a bill would be introduced in Congress to fully nationalize the system (I consider this to be Event #1 of the study). All affiliates to the AFJP systems would be forced back into the public system. With the ruling party holding a majority in both chambers of parliament, and no strong defense of the AFJP system by the opposition, the nationalization bill was passed by the congress with seventy-two percent of votes in the Senate and sixty-eight percent in the House of Representatives. This is Event #2 in the study. One striking fact that suggests the nationalization was more a short-term political gamble than a well-thought strategic decision is that during the three weeks the project was debated in the parliament, no actuarial or financial studies about the viability of the new system were presented and “apparently none existed.”Footnote 17
By taking over the control of the retirement funds previously managed by AFJPs, the government became the de facto owner of those financial assets, which included shares in forty-two publicly traded Argentine companies. Initially, government officials declared that the government was not planning massive sales of shares that could hurt the market value of the companies and had no intention of participating in the governance of the firms either.Footnote 18 Five months later, however, the government changed its position and placed a director on the board of one of those firms, Gas Natural BAN (This is Event #3 in the study). News articles and anecdotal evidence show the business community became concerned after this, anticipating that the government would use any means available to exert control over the private sector. To avoid further conflicts, most companies agreed to incorporate government representatives on their boards. The last event considered in this study was the elimination of a pre-existing voting rights cap. A law from the 1990s (24.241), established that the investments in stocks owned by AFJPs would not allow for the exercise of more than five percent of the voting rights of a company, regardless of the stake owned by the AFJP. This law had been created to facilitate the use of pension funds to finance productive activities; firms could seek financing in the stock market, knowing that AFJP-held shares would not be used to control the companies. In April 2011, however, the government passed an executive order (DNU 441/2011) to modify this, which increased its voting rights in thirty-two companies, in some cases up to nearly thirty percent from the previous five percent (This is Event #4 in the study).
To identify the relevant events of this process for my purposes, I used the same criteria used by a previous studyFootnote 19 about the effects of the negotiation for the Canada-U.S. Free Trade Agreement. The author identified six event windows around the six main events of the process. The shortest window was two days long and the longest was twelve days. The criteria used to determine the length of each window was to set the first date at the day when the information was publicly available and the last day of the window as the first trading day after the events had transpired and the news were published (For further detail about the selection of the windows and their length, see the Data section of the paper).
Research question
I examine whether political connections shield firms from some of the losses stemming from adverse policies, using Argentina’s renationalization of the private pension system as a backdrop. While I do not look at the counter-reform and renationalization of the pension funds per se in this research, I examine an important byproduct of this renationalization: the effect of having the government as a shareholder.
An interesting feature of this case is that the government’s ownership stake in partially acquired firms was predetermined. The number of shares previously held by private pension providers automatically translated into ownership stakes for the government. These stakes varied widely, as can be seen in Table 2, from nearly thirty percent to less than one percent. The fact that the government seized stakes in publicly traded companies was not a guarantee of government intervention. The government could have chosen not to intervene in the governance or management of these firms. I analyze the effect of this “accidental” partial nationalization on the value of politically connected and non-connected firms. As I discuss in the next sections, I expect that partial nationalization will cause significant decreases in the stock prices of affected firms. I examine the extent to which political connections protected firms from some of these losses. Following previous works, I estimate the value of political connections by looking at the movement of stock prices during four key events in Argentina’s pension nationalization process and subsequent appointment of the first government representative to a corporate board.Footnote 21 I found that politically connected firms incurred relatively smaller losses than non-connected firms after partial nationalization. However, the benefits of political connections tapered off significantly in firms with greater government ownership stakes.
Table 2: List of firms and government stakes.Footnote 20
In the following sections, I review literature and develop hypotheses related to the effect of partial nationalization and political connections on firm value; discuss the methodological strategy, present the results, and conclude.
Theory and hypotheses
Government ownership and firm value
A great deal of research has examined the effect of government ownership on firm value. According to property rights theory, private ownership is more efficient than government ownership because it is more concentrated.Footnote 22 Whereas government ownership is widely dispersed among motivated and non-motivated owners, private ownership is concentrated among motivated individuals with incentives to work to increase the value of the firm. Additionally, private owners whose wealth is tied to the value of the firm are motivated to monitor firm managers, but few such incentives exist regarding state-owned enterprises (SOEs). SOEs’ CEOs usually do not lose their jobs in the event of poor performance.Footnote 23 Furthermore, managers and board members in SOEs tend to lack independence to make business decisions and often are simply executing the instructions they receive from the government.Footnote 24 Consistent with the predictions of property rights theory, another study finds that firm value is positively related with concentration of ownership by institutional investors and decreases when ownership is concentrated in government hands.Footnote 25 Many empirical studies on privatization find increases in profitability, investment levels and efficiency after ownership is transferred from governments to private ownersFootnote 26 , and even innovation.Footnote 27 Another study finds similar effects following partial privatization (i.e. the sale of non-controlling stakes on SOEs) in India.Footnote 28
Another key criticism of government ownership is the idea that SOEs face “soft” budget constraints due to their abundant access to public funding. In fact, budget constraints seem to play a significant role in performance; for instance, in Italy, the productivity of Italian SOEs improved in the face of increased financial pressure (i.e., “hard” budget constraints).Footnote 29 Another study finds that minority state ownership through the National Development Bank of Brazil increased firms’ return on assets.Footnote 30 The authors attribute their results to a reduction in capital constraints without interference of the state in firms’ governance and management. Thus, it may be the case that a minority stake in firms can increase firm value if it promotes long-term investment but shields firms from governmental interference and potential minority shareholder expropriation.
Other research (particularly from China) finds more nuanced effects of government ownership, suggesting that governance and institutional factors might matter more than ownership per se. For example, the market value of publicly traded SOEs in China increased when ownership was transferred from government agencies to other SOEs.Footnote 31 Because managers of Chinese SOEs are partially rewarded based on the financial performance of their companies, they are better incented to maximize firm profits than managers in government agencies.Footnote 32 The Chinese case, however, is not necessarily comparable to that of Argentina. While both Argentina (after 2002) and China have pursued interventionist economic policies, the nature, institutional context, and consequences of state involvement differ significantly between both countries. While Argentina has often been infamous for the arbitrary and unpredictable nature of its economic policies, China promoted market mechanisms, investments, and encouraged private entrepreneurship (within a tightly controlled political framework). China used long-term planning (e.g., Five-Year Plans) and its interventionism has been strategic, technocratic, and consistent, aimed at maintaining political control while fostering rapid economic development. Argentina’s interventions, on the other hand have been reactive, discretionary, and destabilizing. Based on this, having government minority ownership in private companies carried different implications in Argentina and China, due to divergent political institutions and state-market relationships. In Argentina, minority government stakes often imply disproportionate political influence rather than strategic economic coordination. Even small state ownership can be used to influence board decisions, especially in strategic sectors, or advance short-term political objectives, such as controlling prices or employment. Given Argentina’s history of institutional volatility and politicized policymaking, minority state stakes are seen less as a means of coordination and more as a vehicle for state interference.
I therefore expect that government’s reputation for partnering successfully with business will be an important determinant of how the acquisition of a minority stake will affect firm value. In the case of Argentina, a long history of irresponsible money printing resulted in dismal indicators of economic development and a bad reputation for economic governance.Footnote 33 Additionally, press reports regarding the behavior of Argentine government officials in various corporate settings suggest the government was not an easy partner. I therefore expect to see a negative association between partial nationalization and firm value in Argentina.
Hypothesis 1: The share price of partially nationalized firms will decline more during the four event windows in the pension nationalization process than firms in the control group.
Political connections, government ownership, and firm value
Despite the predicted negative effect of partial nationalization on firm value, I do not expect this effect to be equal for all Argentine firms. I focus particularly on the effect that political connections might have in protecting some firms from the adverse effects of government ownership.
As discussed earlier, previous research finds that political connections provide firms with significant advantages in the marketplace.Footnote 34 However, the value of political connections may be contingent upon institutional factors and can change in cases of economic and regime change. For example, based on interviews with Chinese managers, in urban industrial China where legal and economic institutions are improving, the use of social and political connections to receive special treatment (i.e., guanxi practice) is becoming less important.Footnote 35 However, the degree to which Chinese managers view these connections as less important depends upon the firms’ position in the administrative hierarchy of the pre-liberalization economy. For example, in the early onset of the Asian economic crisis, politically connected Malaysian firms experienced much greater losses than non-connected firms due to the large decline in the expected value of subsidies received by politically connected firms.Footnote 36 However, these same firms enjoyed large increases in value capital as a result of capital controls subsequently adopted by the Mahatir government in 1998. The authors of that study argue that, in contrast to an open system of global capital flows, the imposition of capital controls creates an environment that enables politicians to better channel financing to favored firms. In a similar vein, firms’ network ties to the Korean government retained their value, despite significant political and economic liberalization.Footnote 37 However, the benefit for some Korean firms of having a CEO or officer with ties to a particular regime became a liability in cases where regime change resulted in the expropriation of the power base of the previous regime.
A common theme that emerges from previous research on the value of political connections in changing environments is that one of the most significant potential threats to the value of these connections seems to stem from the prospect of regime change.Footnote 38 Regardless of whether a country is autocratic or democratic, and whether the regime change is due to the fall of a dictator or a turnover in the party control of congress, the value of firms with “the wrong friends,” i.e., ties to a past regime that loses favor, tends to decline.Footnote 39 In the absence of regime change, political connections tend to retain their value—even during periods of economic liberalization.
In the case of partial nationalization in Argentina, I hypothesize that the value of firms in which the government gained an ownership stake will decline. As I discussed above, the nationalization of private pensions can be viewed as anti-business in the sense that the government’s aim was to gain access to a large pool of cash resources.Footnote 40 In so doing, the government eliminated an important source of private funding for Argentine firms in a manner analogous to the imposition of capital controls in Malaysia. In both cases, governments retained the ability to support favored companies. In Argentina, the government did not choose the size of the stake it owned in partially nationalized firms, and it was free to decide whether or not to appoint a voting representative to the boards of acquired firms. Because there was no regime change, I have no reason to expect that firms who enjoyed the benefits of political connections prior to the pension fund nationalization would suddenly lose their relative position after nationalization. I therefore expect firms with strong connections to the government to incur much smaller losses than other partially nationalized firms.
Hypothesis 2: The share price of partially nationalized firms with strong political connections will decline less than the share price of partially nationalized firms with weak or no political connections during the critical event windows in the pension nationalization process.
Ownership stake and the value of political connections
Finally, I expect that political connections might not protect firms in which the government has acquired very large ownership stakes from incurring substantial losses. As I discussed, the “accidental” nature of Argentina’s partial nationalization of publicly traded firms gave the government complete discretion as to whether or not to interfere in the management of the firms in which it gained an ownership stake. Political connections might protect firms in that they affect investor predictions of the likelihood of any government intervention. However, the potential for governments with very large ownership stakes to exert near full control over the operations of firms, regardless of the firms’ political connections, might cause investors to discount the value of political connections as a safeguard against government interference.
In the context of privatization, researchers have found evidence that the size of the stake that government retains in partially privatized firms has important implications for firm performance. In Chinese privatizations, studies have found a U-shaped relationship between government ownership stake and firm performance.Footnote 41 At high levels of ownership (above a critical value of 36 percent), investors may not be convinced that the government is fully committed to privatization. In such cases, in the absence of strong market institutions, partial privatization reduces firm value. At levels of ownership below the critical value, market discipline and monitoring reduce agency costs and improve firm performance. Some studies find a similar U-shaped relation between firm value and government ownership stake, arguing that the government has both “grabbing hands” and “helping hands” in these partially privatized firms.Footnote 42
Applying these findings to the case of nationalization rather than privatization, I therefore expect that regardless of how politically connected a firm may be, its value will decline as the government’s ownership increases.
Hypothesis 3: Government ownership stake will moderate the relationship in H2; specifically, greater levels of government ownership will weaken the positive effect of political connections on the share price of partially nationalized firms.
Data and methods
In this section, I describe the sample of firms and the stock market and accounting data I used to construct variables; I define the series of events I used in the partial nationalization process and discuss the econometric approach.
Sample
The sample consists of seventy publicly traded firms on the Buenos Aires stock exchange (BCBA). Accounting and stock price data were obtained from the website Bolsar.com, which contains the official stock data from the BCBA. These seventy firms were used to construct a market sample, m, made of thirty firms in which the government did not own shares and a partially nationalized sample, g, made of forty firms in which the government acquired an ownership stake as a result of the pension fund nationalization. The list of companies was highly heterogeneous in terms of industry sectors. Energy and utilities (eleven firms) was the most represented industry, followed by banking, oil, real estate, retail, food production, agriculture, textile, cement, transportation, steel, and chemicals.
The sample was small for two primary reasons. First, there was only a small number of publicly traded firms in Argentina, of which, a significant percentage had little or no regular trading activity. Reinforcing this point is the fact that the private pension funds owned shares in forty-two of the more than one hundred companies that trade. Thus, the potential universe of firms to include in the market sample, m, was barely north of fifty. Second, I had to drop firms with very intermittent trading activity. Thus, of the forty-two total firms in which the government acquired an ownership stake, I dropped two firms that had fewer than thirty trading days in the first nine months of 2008Footnote 43 .
Econometric approach
The empirical analysis consisted of two distinct stages.Footnote 44 First, I did a conventional event study, identifying four event windows. I used stock price data to determine whether the affected firms (those where the government became a minority shareholder) suffered abnormal returns during those windows. The second stage was a more granular analysis aimed at identifying the firm-specific factors that affected the magnitude and sign of said abnormal returns. For this, I used cumulative standardized abnormal returns (CSAR) (for each firm) as dependent variable and ran linear regressions using firm-specific dependent variables and controls.
The goal of the event study method is to measure the effect of an unanticipated event on stock prices. The standard approach is based on estimating a market model for each firm and then calculating abnormal returns, which should reflect the market reaction to the new information contained in each event.Footnote 45 To estimate the relative changes in market value of partially nationalized firms, I constructed a portfolio g (firms with future government participation) and regressed the returns of each stock i from portfolio g on the average daily returns of portfolio m (firms without future government participation) during the L days of the estimation period (January–September 2008). This allowed me to observe whether there were abnormal returns during the events studied. In simple terms, I predicted the expected return of each stock in portfolio g during each of the four events, based on the behavior of that stock relative to the control group m during the preceding months and then determined if the actual return (during each event) was abnormal or not.
The expected rate of return of security i on day t was estimated as:
$${R_{it}} = {\alpha _i} + {\beta _i}{R_{mt}} + {\rm{ }}{e_{it}}$$
where α is a constant, β is an estimate of the systematic risk, R mt is the rate of return of market portfolio m on day t, and e it is the stochastic error term, with E(ϵ it ) = 0.
The daily abnormal returns (AR) for each individual stock were calculated using the following equation:
$${\rm{A}}{{\rm{R}}_{{\rm{it}}}} = {{\rm{R}}_{{\rm{it}}}} - \left( {{{\rm{a}}_{\rm{i}}} + {{\rm{b}}_{\rm{i}}}{{\rm{R}}_{{\rm{mt}}}}} \right)$$
The abnormal returns (AR it ) represent any difference between the actual returns earned by stock i and the returns estimated by the ordinary least squares market model for that stock. I standardized abnormal returns by its standard deviation: SARit = ARit/Sit,
where Sit = {S2
i [1 + (1/Li) + [(Rmt – Avg(Rm))2/
$\mathop \sum \nolimits_{{\rm{t}} = 1}^{{\rm{Lj}}} {\rm{\;}}\left( {{\rm{Rmt\;}} - {\rm{\;Avg}}\left( {{\rm{Rm}}} \right)} \right)$
2]}1/2
S2 i is the estimated residual variance from the market model regression for security i, Avg(Rm) is the average return of portfolio m over the Li days used for the regression, and Rmt is the return of portfolio m at day t.
For each security i, the standardized abnormal returns for each of the days of the event window under study were summed to form a standardized cumulative abnormal return CSARi, where
$${\rm{CSAR_{i}}}=\mathop \sum \nolimits_{{\rm{d}} = 0}^{{\rm{d}} = {\rm{n}}} {\rm{\;}}{\rm{SAR_{it}}}*{1/n}^{1/2},$$
where d = 0 is the first day of the event window and n is the number of days of the event window. If the daily abnormal returns RAit are normal and independent across t, then the standardized cumulative abnormal return CSARi is distributed Student-t with (Li – 2) degrees of freedom; since Li is large, CSARi is assumed to be distributed normally.
To test the significance of the average standardized cumulative abnormal return in a sample of N securities (in this case the 40 stocks of portfolio g), I computed:
$\begin{align}Z = Avg (CSARi)& Sqrt;N\end{align}$
where Avg (CSARi) = 1/N
$\mathop \sum \nolimits_{{\rm{i}} = 1}^{\rm{N}} {\rm{\;}}$
CSARi,
Assuming that the standardized cumulative abnormal returns are independent across different stocks, then the expected value of the standardized abnormal return is 0 and the test statistic Z will be distributed unit normal for the assumed unit normal CSARi’s.
In hypothesis 1, I predict negative and significant abnormal returns for partially nationalized firms in portfolio, g, relative to firms in the market portfolio, m, for each event, j.
$\begin{align}H1: & Amacr;Rgj \lt 0 ; H0: & Amacr;Rgj = 0\end{align}$
Following the established methodologyFootnote 46 , I pooled together each firm’s abnormal returns for each event, j, and used these as the dependent variable in the second part of the analysis. To do so, I constructed each firm’s CSAR for each event, j. This involved first, standardizing the abnormal returns by their standard deviation. Then, for each security i, the standardized abnormal returns for all of the days of the event windows (for each event, j) were summed to form the standardized cumulative abnormal return CSARij. This was the dependent variable for the second stage and can be interpreted as each firm i’s abnormal returns for each event j. In hypothesis 2, I predict that political connectedness will be positively related to CSARij. In hypothesis 3, I predict that government ownership stake will moderate the positive relation between political connectedness and CSARij.
Variables
I have already described the two dependent variables in the analysis. The independent variable of interest is the measure of political connectedness. Following other studies, I wished to include broad-based measures of political connectedness.Footnote 47 In this research, I measured political connectedness using an index that comprises five different types of connections, and I allowed the measure to take on negative values if firms had bad relations with the [Kirchner] government (See Table 3). The potential range of values for political connectedness was −1 to 5. A value of −1 meant no political connections plus a declared conflict with the government (see Item #2). A value of 5 meant the firm had all possible sources of political connections. The following types of political connections were included in the measure:
1) A background in politics: Set equal to 1 if the CEO or a member of the board of directors of the firm had occupied political positions in the past.Footnote 48
Table 3: List of event windows
2) Current relationships: if the CEO or a board member were identified as having close personal relationships with members of the government. For example, for one of the firms, I found articles in the press that considered the chairman of the company to be “the Kirchners’ banker.”Footnote 49 In contrast, the Clarín Group, a media conglomerate, had an openly hostile relationship with the Kirchner government and was assigned a score of −1, i.e., a negative relationship.Footnote 50
3) Financial contributions: if the firm, a top executive or a board member contributed funds to the political campaign of the ruling party. I investigated campaign contributions in official reports and press articles.Footnote 51
4) Legal or administrative interaction: firms that operated in “regulated” sectors and therefore had to negotiate with the government on a regular basis, with regards to prices, tariffs, and investment decisions. Those sectors included energy, utilities, oil, and toll-road operation.
5) Existing minority government ownership at the time of the events.
To measure items #1 and #2 I searched for CEOs and board members of all the firms as of 2008 (time of the first event) and analyzed their bios and press information about them. A total of 402 people were searched online in company websites, social networks like LinkedIn, shareholder meeting notes, and other sources.
In addition to the measure of political connectedness, I included several additional independent variables. First, I employed a variable stake, which measures the percentage of shares acquired by the government after pension fund nationalization divided by the total outstanding shares of the firm. Second, following other studiesFootnote 52 , I used an indicator set equal to 1 if a firm had been previously privatized. Although past privatization can be a type of political connectedness, it is not straightforward whether it implies a positive, negative, or neutral relationship with a current government.
I used the control variable foreign for foreign subsidiaries listed on the Buenos Aires Stock Exchange. I have no prediction regarding the sign of foreign. On the one hand, foreign firms can be a target for regimes that want to establish populist bona fides. If this were true, I would expect foreign firms to be hurt relatively more by partial nationalization than domestic firms. Conversely, foreign-owned firms listed in Buenos Aires are likely to have significant operations outside Argentina, which may be a small part of overall global operations. Under this scenario, I might expect foreign firms to be relatively less hurt by partial nationalization than domestic firms.
I controlled for size using the log of the firm’s stock market capitalization which I measured prior to partial nationalization.Footnote 53 I also included size squared to examine whether the largest firms were affected differently. In Mexico, as a result of financial liberalization, large firms lost privileged access to credit.Footnote 54 I expect firm size to be important, but it is again not clear whether large firms will be hurt relatively more or less by partial nationalization. Again, large firms are highly visible targets for leaders wanting the appearance of populism. Large firms also have more assets available for “grabbing hands.” These features of large firms would lead to the expectation of greater negative returns to firm size. However, large firms are likely to be more politically connected and have more employees who could stand to be hurt if government intervention is detrimental to the operations of the firm.Footnote 55 These features of large firms would lead to the expectation of positive returns to firm size. Thus, I included size and size squared in the regression estimates with the expectation that they will be significantly associated with firms’ abnormal returns in one direction or the other.
Considering that the 2008–2009 crisis hit banks especially hard, I used the binary variable bank to control whether financial institutions were affected differently. I also used the variable trades abroad for companies that trade in other financial markets. Finally, I used indicator variables in the pooled regression to control for the different events and, to adjust for correlation across repeated observations on the same firm, the models contain standard errors clustered by firm.
Construction of events
I constructed a chain of four events (See Table 4) in the pension nationalization process. Event 1, dated 10/20/2008, was the first news of the government’s plan to nationalize pension funds or AFJPs. Event 2, dated exactly one month later, was the debate and subsequent passage of the bill by the Argentine National Congress. Event 3, dated 3/19/2009, was the date the firm Gas Natural BAN (ticker: GBAN) communicated to the Buenos Aires Stock Exchange that the Argentine Government had informed the firm of its intention to participate and vote in an upcoming shareholder meeting. On March 24, 2009, during the shareholder meeting, a director nominated by the government was designated to sit on the board of BAN. Event 4 was the signing of the executive order to eliminate the government’s five-percent voting rights cap in public companies. The cap was a “market-friendly” law designed to limit the AFJP’s voting rights in public companies to five percent, regardless of the stake in the company. The elimination of this cap gave the government larger voting rights in companies where government ownership stakes were above five percent.
Table 4: Descriptive statistics and correlations
The events were identified by searching keywords on the website of La Nacion (www.lanacion.com.ar), one of the two main newspapers in Buenos Aires, with 19 million unique visitors per month according to Google News. The website allows searching for multiple keywords and sorting by date the articles where the keyword appears. The first keyword searched was “AFJP” and by sorting the articles by date, it was determined that the news about the nationalization first reached the press on 10/20/2008. The search of the word AFJP yielded nearly six thousand articles. I analyzed the content of all of them to identify the most relevant events in the process, and to collect background information.
There are three critical assumptions underlying the use of event studies for identification of abnormal returns: market efficiency, unanticipated events, and absence of confounding effects.Footnote 56 Market efficiency implies that the markets incorporate all relevant information into the stock prices as soon as it is available. The assumption of unanticipated events implies that traders learn about the new information through the press and the media. The assumption of lack of confounding effects implies that the researcher can isolate the effect of an event from the effect of other events. The longer the window, the higher the likelihood of the occurrence of confounding events.
Regarding market efficiency, I have no particular indication that this assumption was violated; the Merval Index (index of “blue chips” in BCBA) showed quick reactions immediate to these and other events relevant to the economy, which is consistent with the notion that the financial market was incorporating the available information to the stock prices as soon as this information was available. Nonetheless, the use of extended windows of up to five days allowed me to control if the market reacted slowly to the event.
With regards to confounding effects, I chose for each event the shortest possible window between the release of information and the first full trading session. The longest window (Event #1) was only three trading days long and the rest (Events #2, 3, and 4) were two days long. I added an extra day to the time window for Event 1 for two reasons. First, rumors of the pending nationalization announcement began to appear the day before the official announcement was made. Second, the official announcement was made after the closing of the trading day, so I wanted to include a full trading day after the official announcement. I also tested for abnormal returns using “extended” event windows, which extend from five days before to five days after the original “short” event windows. Although the use of very narrow windows allowed me to curb the problem of confounding effects, extending the windows allowed me to control whether the abnormal returns started before the event itself (which could indicate information leakages) and whether they also extended for several days after the event (which may be an indication that the market is not perfectly efficient and is realizing the effects of the events over a longer period). The main results were largely unchanged in estimates with longer event windows.
Finally, the assumption that events were unanticipated is likely to be valid for Events #1 and 3. I conducted a lengthy interview with a journalist from the newspaper La Nación, identified as an expert on this topic based on the number of articles she had written about it. This journalist confirmed that event 1 was unexpected for her, for the market, and even for most government officials (except for the president and a very small circle). Event #2 was probably expected by the marketplace, since the governing party had a majority in both chambers of parliament. Regarding Event #4, the degree to which it was anticipated was not clear; there had been hints that the government might seek to eliminate the 5-percent voting rights cap, but it was not confirmed.
Results
Table 5 presents summary statistics for the forty firms in the partially nationalized portfolio. I did not include firms in the control sample in the pooled regressions and therefore did not report descriptive statistics for these firms. In column 1, I listed the variables in the regression model. Note that I only included indicator variables for the first two events. I excluded the fourth event from the pooled regressions because there were no significant abnormal returns for this event. Event 3 is the referent event in the regressions, so I did not construct an indicator for it.
Table 5: Stock market reactions to pension system renationalization and subsequent events
Several correlations are noteworthy. First, firms’ cumulative abnormal returns were negatively correlated with the indicator for event 1, which is not surprising, because this event was the one that took the market by total surprise. Political connectedness and previously privatized were positively correlated: previously privatized firms were more likely to have legal and administrative connections to the government (primarily due to their over-representation in regulated industries). Foreign and previously privatized were also highly correlated, since many foreign multinational firms bought Argentine firms during the privatization process. Another noteworthy correlation was the negative relationship between size and stake; not surprisingly, since it is easier to obtain larger stakes in smaller firms. As expected, political connectedness was positively correlated with cumulative abnormal returns. Stake was negatively correlated with cumulative abnormal returns and foreign ownership. Domestic AFJP’s held relatively smaller stakes in foreign-owned firms, which translated into smaller government stakes for these firms. Trades abroad was positively correlated with political connectedness.
In Table 6, I reported average cumulative abnormal returns for each event and compared the size of the daily returns for the two portfolios of firms—partially nationalized firms, g, and the control group of firms, m. For each event, the first row of Table 6 indicates the Z score and significance level for the abnormal returns in portfolio g. As predicted in Hypothesis 1, the Z scores were negative and significant for Events 1-3. However, there was no significance in the magnitude of abnormal returns in the partially nationalized and control sample of firms for Event 4. In Rows 3 and 4 of Table 6, I report the size of the returns for the two portfolios of firms. The first event was clearly the most important, in terms of market reaction. Share prices in the portfolio of partially nationalized firms declined by almost twenty four percent upon the announcement of the government’s intention to nationalize pension funds. Subsequent events saw much more modest declines in share prices in both portfolios of firms. Row 5 of Table 6 indicates the percentage of firms in the partially nationalized portfolio that experienced negative abnormal returns during each of the four event windows.
Table 6: Linear regression analysis for cumulative standard abnormal returns (events 1, 2, and 3)
*** p < 0.001, ** p < 0.01, * p < 0.05, † p < 0.10.
The results supported Hypothesis 1. In three of the four events, the share prices of partially nationalized firms declined significantly more than share prices of firms in the control group. It was not terribly surprising that the share prices of partially nationalized firms did not react strongly to the executive order that ended the 5-percent voting rights cap (Event 4). Event 4 occurred two full calendar years after Event 3. Additionally, as discussed earlier, rumors suggested that investors may have anticipated the elimination of the voting rights cap prior to the announcement. It is also likely that two years after the first appointment of a government official to the board of a public company (Event 3), investors had already figured out the consequences of government intervention in partially nationalized firms and the impact was not significant.
Figures 1-3 break down the Cumulative Abnormal Returns of partially nationalized firms by political connectedness, firm size, and foreign vs. domestic ownership for Events 1-3. I excluded Event 4 for reasons discussed above. Some interesting differences are evident. For example, in Figure 1, I see that the share prices of politically connected firms declined less than non-connected firms in all three events. In Figure 2, I note much smaller and less regular differences in share price movements of large and small firms. Event 1 was worse for large firms, whereas Events 2 and 3 were modestly worse for small firms. In Figure 3, the differences in share price movements between domestic and foreign firms were very small in the first two events, although domestic firms were more affected by event 3. This makes sense, because having a government representative on the board is more influential for a domestic firm than for a foreign multinational.
Figure 1. Cumulative abnormal returns, politically vs. nonpolitically connected firms*. * Politically connected firms: those with scores above the median of Political Connectedness.
Figure 2. Cumulative abnormal returns by firm size*. * Larger and Smaller firms are above and below the median for size, respectively.
Figure 3. Cumulative abnormal returns by firm nationality.
Finally, in Table 7 I report regression results pooling across events 1–3. In column 1 I report the main effects model, and in column 2 I report the full model that includes the interaction of political connectedness and stake. The sample used in the regression estimates in Table 7 consisted of the partially nationalized firms in portfolio g. The total N for the regression estimates was 105, rather than 120 because not all firms in portfolio g traded during all three events.
Table 7: List of political connections by firm
† p <0.10; * p < 0.05; ** p < 0.01; *** p < 0.001.
In Hypothesis 2 I predict that the share prices of politically connected firms would decline less than the share prices of non-connected firms during the event windows in the pension nationalization process. In models 1 and 2, political connectedness was positive and significant, indicating that political connections reduced negative abnormal returns. In Hypothesis 3 I predicted that the positive association between political connectedness and cumulative abnormal returns would taper off for firms with a large stake. In model 2, the interaction of political connectedness*stake was negative and significant, indicating support for Hypothesis 3. I discuss this result in greater detail below.
Another interesting result on Table 7 is the U-shaped relationship between size (negative), size 2 (positive) and cumulative abnormal returns. This indicates that medium-size firms were more affected than both small firms and large firms. Previously privatized was significant and negative in model 2, which may indicate that the market anticipated that privatized firms may be under threat of being renationalized. Contrary to the expectations, the estimates for foreign ownership and bank were not significant.
Finally, to further examine the joint effect of political connectedness and stake, I constructed the interaction plot shown in Figure 4. On the horizontal axis, I held political connectedness constant at the 1st and 99th percentiles (−1, 4) and on the vertical axis I measured the size of firms’ abnormal returns. Holding other regression variables at their sample medians, I plotted firms’ abnormal returns for three different levels of stake (25th, 50th, and 75th percentile) at the two different levels of political connectedness. At the 25th and 50th percentile of Government Ownership Stake, greater Political Connectedness was associated with higher abnormal returns, although the relationship is much stronger at the 25th percentile of stake than at the 50th percentile. At the 75th percentile of stake, the relationship between political connectedness and firms’ cumulative abnormal returns disappeared, as indicated by the perfectly flat line that remains fixed at a level of abnormal returns approximately equal to negative one.
Figure 4. Joint effect of political connectedness and stake on cumulative abnormal returns.
As I predict in Hypothesis 3, the market predicted that higher levels of stake would completely cancel out any expected benefits of political connectedness. Since the government could exercise nearly full control over firms in which it owned very large stakes, this could eliminate the benefits of having good relations with government. Once the government owned the firm to which managers are connected, quid pro quos may become meaningless.
Discussion and conclusion
This research investigates whether political connections protect the value of firms from harmful actions by governments to which they are connected. In instances where governments adopt anti-business measures, either to establish their populist credentials or to tap into new sources of funding, political connections could protect firms from losses. I examined this question in the context of Argentina’s renationalization of the private pension system in 2008. As a byproduct of the pension fund nationalization, forty-two publicly traded Argentine firms were partially nationalized as the government seized the shares of these firms previously held by private pension funds. Although the government initially promised not to intervene in the acquired firms, it soon began placing government-appointed directors to corporate boards and subsequently repealed a policy limiting the government’s vote share on publicly traded firms to five percent.
I investigate specifically whether the value of politically connected firms that were partially nationalized was hurt less than non-connected firms. In theory, political connections could presumably protect firms from unwanted intervention. I found that the value of politically connected firms was less affected by partial nationalization than that of non-connected firms. However, the protective effect of political connections depended on the size of the stake acquired by the government. When the government acquired a very large stake in a firm, the firm’s political connections no longer offered protection from losses. Regarding this point, I must acknowledge an important limitation of this work: this event study analyzes the market reactions to the events that led to the government becoming a minority shareholder in these companies. A follow-up investigation to determine whether years later the government effectively made decisions that affected these firms negatively is outside of the scope of this paper. Furthermore, such an investigation would be difficult, because most interactions between government and firms happen behind closed doors and firms may be understandably reluctant to share such information.
It is also important to acknowledge that 2008–2009 were not random years for stock markets. The global financial crisis created an environment of uncertainty and volatility that affected all major financial markets and hit banks and financial institutions particularly hard. While these were important considerations during this study, I believe that the behavior of global financial markets does not explain what I observed in Argentina. For instance, during the first event window (10/20–10/22/2008), the S&P 500 (SPX) had cumulative returns of −7.84 percent. During the same three days, the portfolio g (firms partially nationalized by the Argentine government) had a cumulative return of −23.28 percent and the portfolio m (control firms in which the Argentine government did not obtain a stake) had a cumulative return of −15.70 percent. In other words, regardless of the “bearish” behavior of the New York Stock Exchange, the stock market in Argentina performed much worse and the impact was clearly more severe for partially nationalized firms, which lost nearly a quarter of their market value in three days. The difference is also remarkable during the second event window (11/20–11/21/2008): while the S&P 500 showed a positive trend with a cumulative return of 14.63 percent over those two days, the stock market in Argentina went the opposite way, with a cumulative return of −8.58 percent for partially nationalized firms (g) and −8.24 percent for control firms (m). And even though, as discussed, banks were particularly affected during this period, the variable bank (which identifies firms in the banking industry) was not significant in the empirical analysis, showing that this was not a meaningful factor determining the abnormal returns observed in the Argentine stock market during the events studied.
This research complements several studies that have looked at whether politically connected firms are better or worse-off in environments characterized by economic liberalization or regime change.Footnote 57 In general, previous research has found that economic liberalization does not erode the value of political connections, but regime change does. In the context of this research, there was no regime change. I examined the limits of political connections when the same government to whom firms were connected pursued policies with adverse effects. Consistent with the liberalization studies, the evidence suggested that political connections did retain some of their value. However, I also found that increasing levels of policy adversity eventually erode all the value of political connections.
The results suggest that there are limits to political connections that should be considered when firms invest in building them. Fisman finds that firms with strong connections to the Suharto family in Indonesia were hurt relatively more by the prospect of the end of his regime than non-connected firms. It is interesting but not surprising that firms with strong connections to an authoritarian regime would be hurt more than non-connected firms when the regime falls. However, the results suggest that when governments to which firms are connected pursue anti-business policies, a firm’s political connections might not protect it from harm. Consistent with the results, the reputation of a country for economic and political volatility should enter a firm’s assessment of the value of its political connections.
Open access
