Estimating the Value of Political Connections

Literature › Works

Raymond Fisman (2001). Estimating the Value of Political Connections. The American Economic Review, Vol. 91, No. 4. American Economic Association.

Type	journal article
Scope	Comparative and methods
Method	econometric
Period studied	1995–1997
Regions	Indonesia
Themes	business and state, comparative theory, methods data sources
Citation key	fisman2001

Argument

Fisman estimates how much Indonesian firms' value depended on political connections to President Suharto. He uses an event-study design built on rumours about Suharto's health in 1995-97. He compares share-price returns of firms with different scores on the Castle Group's Suharto Dependency Index. Firms more dependent on connections lost more when the rumours hit, and the effect grows with the severity of the rumour, proxied by the Jakarta Composite Index return net of regional effects. Extrapolating, he argues that connections were a very large share of value for well-connected firms. He also suggests the finding may extend to other high-corruption economies.

Evidence and method

Extel Financials and Extel Securities databases (1995 accounting data and share prices), Investamatic, Kompass Indonesia and JSX directories for group affiliation, the subjective Suharto Dependency Index (Castle Group, 1995) covering the 25 largest groups and 79 listed firms, six rumour episodes identified through a Lexis-Nexis search, and informal interviews with Jakarta investment bankers and business people.

Findings

• The author says earlier work on firm reliance on connections had not gone beyond anecdote and had not tried to estimate it. (p. 2)

  there has been no attempt to estimate the degree to which firms rely on connections for their profitability.

• The index gives the highest dependency score to firms tied to Suharto's children and close allies. (p. 4)

  All of the companies affiliated with President Suharto's children (Bimantara and Citra Lamtoro Groups) received a score of five

• Politically dependent firms had lower returns than less-dependent firms in every rumour episode. (p. 3)

  show that in every case the returns of shares of politically dependent firms were considerably lower than the returns of less-dependent firms.

• The size of the dependency effect across events tracks the market's perceived severity of the rumour. (p. 6)

  the correlation between p and NRe( JCI) is 0.98.

• In the pooled interaction regression, a one-point higher dependency score means a 0.28 percent larger fall per 1 percent market fall. (p. 6)

  Thus, if the overall market declined by 1 percent in reaction to news about Suharto's health, we might expect a firm with POL = x to drop 0.28 percent more than a firm with POL = x - 1.

• Using bankers' expectations of a 20 percent market drop if Suharto died suddenly, the model implies a large implicit value of connections. (p. 7)

  then (2) implies that the coefficient on POL would have been about 5.8 in the event of a sudden regime shift.

• The 20 percent figure is an informal modal answer from investment bankers, and the extrapolation is far out of sample. (p. 7)

  20 percent was the modal response to this question.

• The actual 1998 fall of Suharto cannot be used as an event because of confounding events and other problems. (p. 7)

  there is no systematic relationship between share price returns and political dependence

• A placebo test on non-Suharto shocks (rupiah float) shows no relation between dependence and returns, ruling out a general high-beta explanation. (p. 8)

  On these days, there was no relationship between share price returns and dependence.

• The flexible-functional-form test supports the monotonic, linear use of the index. (p. 8)

  The coefficients on the indicator variables were uniformly negative, ranging from -0.35 to -0.87.

• The 25 groups covered account for a very large share of the Indonesian economy. (p. 8)

  with revenues of more than U.S. $60 billion in 1995