Political Connections and Accounting Quality under High Expropriation Risk

Abstract

We examine the impact of political connections and accounting quality among Venezuelan industrial firms, which face one of the highest levels of expropriation risk worldwide. Based on prior literature, we expect a negative relationship between expropriation risk and accounting quality as firms manage earnings to avoid ‘benign’ state intervention. We find that politically connected firms have higher accounting quality than non-connected firms, which is consistent with connected firms’ lower risk of expropriation due to connections with high-level government officials or ruling party members. The relationship between accounting quality and political connections appears to be strongly moderated by institutional features like expropriation risk.

Notes

¹Research in this area includes  Chaney, Faccio, and Parsley (2011), Guedhami, Pittman, and Saffar (2014), Chen, Ding, and Kim (2010), Correia (2011), Fan, Li, and Yang (2010), Leuz and Oberholzer-Gee (2006), and Ramanna and Roychowdhury (2010).

²We use ICRG's contract viability (CV) score as a measure of expropriation risk. ICRG defines CV as ‘the risk of unilateral contract modification or cancellation and, at worst, outright expropriation of foreign owned assets’.

³AON Corporation's Political Risk Insurance services in 2012 ranked Venezuela as having Very High political risk, along with Haiti, Belarus, Iran, Iraq, Syria, Yemen, Afghanistan, Pakistan, Sudan, South Sudan, the Democratic Republic of Congo, Somalia, Zimbabwe, and North Korea. The Belgian Export Credit Rating Agency (ONDD) also maintains a list of countries ranked by the risk of expropriation and government action. As of April 2013, Venezuela had the highest risk rating with the ONDD, along with Afghanistan, Argentina, Bolivia, Ecuador, Guinea, Iran, Iraq, Libya, Syria, Yemen, Palestine, Sudan, Somalia, and Zimbabwe.

⁴Zimbabwe had 60 listed, domestic non-financial firms as of April 2013, relative to Venezuela's 33, Ecuador's 26, and Bolivia's 18, according to CreditRisk Monitor's Directory of Public Companies (http://www.crmz.com/Directory/). However, Zimbabwe's nationalisation threats have primarily targeted subsidiaries of foreign companies and small landowners rather than domestic companies.

⁵In a cross-country study, Boubraki, Guedhami, Mishra, and Saffar (2012) use whether a firm is headquartered in the capital city as an instrument for the presence of political connections. While presence in, or proximity to, a capital city may generally be associated with political connection strength, we do not believe it serves as an appropriate instrument in our context, as most of the connections we find are based on long-standing school, professional, friendship, or familial ties. Additionally, Chaney et al. (2011) argue that a capital city presence instrument is unlikely to satisfy the exclusion restriction.

⁶All Venezuelan economic statistics were obtained from Global Financial Data.

⁷Venezuelan average yearly inflation was 32% between 1984 and 1989, while annual real GDP growth averaged only 1.5%.

⁸Examples include Guillermo Zuloaga, former owner of the TV station Globovisión, and the financier Nelson Mezerhame, who are both residing in the USA as of 2013.

⁹Search terms were the following: soberanía nacional, sectores estratégicos, soberanía alimentaria, especuladores, acaparamiento, remarcaje de precios, propiedad social,empresas del estado,empresas de producción social, ganancias excesivas,expropiación, expropiar, nacionalización, and, nacionalizar. These translate as: national sovereignty, strategic sectors, food sovereignty, speculators, hoarding, price gouging, social property, state enterprises, social production companies, profiteering, expropriate, expropriation, nationalisation, and nationalise. We identified articles that described both actual and threatened nationalisations towards particular companies, sectors, or the economy as a whole.

¹⁰As an example, the Venezuelan legislature in 2009 issued a law that required compensation to shareholders of nationalised energy firms equal to the accounting book value of equity.

¹¹The link between political connections and mitigated expropriation risk has been documented in prior literature. Li, Meng, Wang, and Zhou (2008) find that political connections allow Chinese firms to avoid paying ‘extralegal’ fees to government entities. Bodnurak and Massa (2011) find that family firms' ability to provide political connections positively associated with institutional investor ownership, which they attribute in part to these firms' lower government expropriation risk. We anticipate a similar relationship among Venezuelan firms. In a recent example, Guillermo Zuloaga, owner of Globovisión, a television station, recently agreed to sell Globovisión to a financial group known to be close to the Chavez regime. The sale was seen as being forced by the Chavez government, as Globovisión had been highly critical of the government, which had vowed to not renew the station's broadcasting license. The new owner, Juan Domingo Cordero, is seen as a frontman in what would amount to a government takeover of the politically unconnected Globovisión (Rueda, 2013). In an additional example, the government's takeover of several buildings in downtown Caracas was denounced by an opposition party member, who claimed ‘government intends to turn the center of the capital in a space only for its members, for members of the ruling party, into a space where only Chavistas can roam’ (El Mundo, 2010).

¹²Debt capital may be still remain expensive, as connected firms may enjoy preferential access to debt financing (Khwaja & Mian, 2005), which may raise debt financing costs for connected firms if the pool of loanable funds is scarce. However, our expectation is that external equity capital will be relatively much more expensive, and this may prompt firms to shift leverage ratios upward to increase firm value.

¹³Venezuelan accounting rules require firms to revalue non-monetary assets and liabilities, as well as associated expenses like cost of sales and depreciation, based on movements in local price indices.

¹⁴Specifically, all t−1 numbers are multiplied by the ratio of the Venezuelan CPI as of the year t fiscal month end to the Venezuelan CPI as of the t−1 fiscal month. For stand-alone size measures such as total assets, the number was deflated using the Venezuelan CPI as of December 1997 as a base.

¹⁵Because of the highly sensitive nature of the data, we were asked to withhold the names and identifying details of the businessmen we interviewed. Interviewees feared government harassment or, in the extreme, exile from Venezuela if they were associated with political connection assessments.

¹⁶Relatives included spouses, children, siblings, or cousins.

¹⁷The measures of political connection used in Fisman's (2001) seminal study also relied on subjective assessments made by consultants from the Castle Group, a political consultancy operating in Indonesia. We were not able to identify the number of consultants who contributed to the assessments used by Fisman, but at present, CastleAsia, the successor firm to the Castle Group, lists six advisory board members.

¹⁸Venezuelan news media sources we utilised include Globovision, Union Radio, El Universal, Agencia Bolivariana de Noticias, Radio Nacional de Venezuela, Venezolana de Televisión, El Nacional, El Mundo, Noticiero Digital, Noticias 24, and Aporrea. In addition to news stories, we also searched for interviews with government officials among these sources.

¹⁹To evaluate the existence of connections based on these news sources, we relied on the narrative-defined categories listed in Table 1, and we also searched for stories on Venezuelan government officials either (a) criticising the company, its executives, its board members, or its large blockholders or (b) threatening the company with nationalisation.

²⁰Faccio (2010) suggests that political connection measures derived from public sources may represent more durable ties, as opposed to those related to more ephemeral campaign contributions. While the greater durability of connection measures derived from publicly available sources may hold true in general, we believe it is less likely so in the Venezuelan context, where these connections we identify through non-public sources represent long-standing friendship, familial, and professional ties.

²¹The sales growth variable was subject to severe outliers, so we winsorised instances of over 200% sales growth to the 95th percentile of sales growth. Results were unchanged when we only winsorised at the 1% level. Some firm-years had zero revenue, owing to income primarily derived from equity method investments. When prior year revenue was zero, we recorded sales growth as equalling zero.

²²We adopt a 0.005 threshold for small positive earnings, versus 0.01 as in Lang et al. (2003). Overall sample mean return on assets (net income divided by beginning-of-period assets, as per Section 4) is quite low, at 0.007, with a median of 0.011. Around 45% of the sample has return on assets plus or minus 0.03 of the mean of 0.01. Given that the distribution is fairly tight and has a mean at exactly at the 0.01 used by Lang et al. (2003), adopting the 0.01 threshold may be a poor indicator of actual earnings management behaviour. When we use this less stringent 0.01 threshold, our results on small positive earnings are similar.

²³For consistency with our small positive earnings measure, we adopt a 0.005 threshold for earnings increases. Results are similar using the less stringent 0.01 threshold, with 14% of firm-years having small earnings increases, versus 9% using the 0.005 threshold.

²⁴Barth et al. (2008), following  Lang, Raedy, and Wilson (2006), use an OLS framework because logit models are prone to severe heteroskedasticity problems in small samples. Lang et al. (2006) use SMALLPOS as a right-hand side variable, as their controls are meant to control for determinants of the cross-listing decision.

²⁵A higher ratio of total assets to net PPE may also simply represent cash-rich firms (where most of assets are composed of cash and short-term investments, rather than PPE) that are targeted by the government. This does not appear to be among the government's espoused expropriation motivations; however, results are similar if we subtract cash and short-term investments from total assets in calculating the ratio.

²⁶McKay and Phillips (2005) use employees to net PPE as a measure of labour- versus capital-intensity.

²⁷In untabulated regressions, we evaluate, as a benchmark, whether the assets to net PPE ratio for industrial firms in Compustat North America from 2000 to 2008 is associated with the likelihood of small positive earnings; the likelihood of small earnings increases; and the absolute value of abnormal accruals. We adopt the same empirical specification and controls for our tests above, though we exclude XLIST and LOCALAUD, and adopt a logit specification for binary dependent variables. Among US firms, we find that the logarithm of the ratio of assets to net PPE is actually negatively and significantly associated with both small positive earnings and small earnings increases. Estimating abnormal accruals at the GICS group and fiscal year level with at least 10 observations, we find no significant association with the absolute value of performance-matched abnormal accruals. Results are identical if we restrict tests to only those industries present among Venezuelan sample firms, or substitute in the logarithm of the ratio of employee headcount to net PPE.

²⁸For all measure validation tests, we employ White (1980) robust standard errors, rather than clustered standard errors, as F-statistics cannot otherwise be produced for models with all control variables included. Although we rely on potentially manipulated financial reports to measure interest rates, we note that methods to reduce reported interest expense, such as utilising hybrid or off-balance-sheet debt, will reduce both reported interest expense and reported debt; imputed interest rates should therefore still accurately measure financing costs for at least on-balance-sheet debt

²⁹In Venezuelan GAAP, benefits from the use of deferred tax assets are typically included as part of extraordinary items, though over a quarter of sample firm-years recorded an extraordinary tax benefit on their income statements. Tests with all control variables included are robust to excluding extraordinary tax benefits in measuring effective tax rates.

³⁰If pretax income is negative while tax benefits are positive (i.e. tax expense is reported as negative), then firms are recording deferred tax assets in anticipation of future tax benefits. If pretax income is negative while the provision for taxes is positive or zero, it suggests firms are not confident about recording the gross value of deferred tax assets.

³¹It is possible that, given our respondents are generally sympathetic to the anti-Chavez opposition, firms that are truly connected will have been classified as unconnected, to spare themselves or associates' firms from embarrassment with having been associated with the regime. If our hypothesis is true on the negative relationship between connections and earnings management, we will have a more difficult time finding significant results due to this bias. If our hypothesis is false, and connectedness has no relationship or even a positive relationship with earnings management, then we would have to believe that a non-negligible number of the firms associated with our respondents who are then misclassified as unconnected have poorer accounting quality, presumably for capital market or contracting purposes. While we cannot rule out this possibility, there is no reason to believe the firms associated with the businessmen that were interviewed happen to have poorer accounting quality. Moreover, through our construct validation tests, we have indeed found that connected firms in our sample derive economic benefits (lower debt cost of capital, lower effective tax rates) that unconnected firms do not. There are certainly alternative plausible stories for how connected firms would have greater incidence of earnings management. However, it is more difficult to derive a story for connected firms enjoying a lower debt cost of capital, and also contrary to evidence in Khwaja and Mian (2005) and Faccio (2010). It is even more difficult to explain why connected firms would suffer a higher tax rate than unconnected firms.

³²Chaney et al. (2011) use industry and year-matched firms for their benchmark accruals model. However, Dopuch, Mashruwala, Seethamraju, and Zach (2011) note that the assumption of a homogeneous accrual-generating process within industries may be suspect. Ecker, Francis, Olsson, and Schipper (2013) propose instead using within-country year and lagged asset-matched firms for the accruals benchmark model, especially in contexts like ours where there is an insufficient number of within-country industry peers to compute abnormal accruals. They also note that differences in institutional features (such as accounting standards, legal codes, or culture) may affect the usefulness of cross-country, industry-matched peer models. Results are similar when we use the 10 largest lagged asset peer firms to measure ‘normal’ accruals, though only the interaction term of EXPROP and CONN × EXPROP remain significant and in the predicted direction.

³³Specifically, for asymmetric loss timeliness and abnormal accruals tests, per Cameron et al.’s (2008) optimal specification, we use wild bootstrap with +1 and −1 weights on negative versus positive residuals; we ‘impose the null’ in generating predicted values and residuals; and we use 999 bootstrap replications. For small positive earnings, small earnings increase, and large negative earnings tests, we use non-parametric bootstrap, as wild bootstrap produces simulated dependent variables no longer equal to dichotomous outcomes necessary for these tests.

³⁴We identified family ownership through news searches and from our interviews evaluating political connections (Section 4.1).