Indonesia's centripetal democracy and economic growth

ABSTRACT

What has been the impact of democracy on growth in Indonesia? An early view emphasized a growth and investment slowdown, but more recent data suggest otherwise. What accounts for the reversion to New Order growth and investment rates? My answer is the adoption of more centripetal and developmental democratic institutions in an otherwise messy democracy that has pushed political parties and presidents to compete for votes by promising to clean up government and restore growth.

KEYWORDS:

• Indonesia
• centripetal democracy
• economic growth
• investment

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. As they say (2009: 355), ‘…Indonesia's political environment changed radically, from a ‘hard,’ authoritarian, corrupt but growth oriented state…to a weakened, democratic, corrupt state, with the political leadership not yet able to provide a clear and unambiguous commitment to economic growth.’

2. A test of equality of means rejects the hypothesis that mean growth rates were the same (t = 3.54, p = .00).

3. Following Hill (1994 and 1996), I date the beginning of the New Order from 1966. As is well known, the New Order collapsed in 1998 with Suharto's resignation while Polity IV dates the beginning of democracy in 1999 (See Excel spreadsheet for Polity IV at http://www.systemicpeace.org/inscr/inscr.htm)

4. The turn towards a centripetal democracy is described on pages 8–15 and in Table 3. The key elements of Indonesia's centripetal democracy include strengthening the presidency, use of vote thresholds and distribution requirements for presidential elections and for political party participation in presidential and parliamentary elections to reduce party fragmentation, adoption of a form of decentralization that protected a unitary state, reduced district magnitudes putting Indonesia's PR electoral rule in what Carey and Hix (2011) label an electoral sweet spot that enhances the accountability of office holders to the electorate, and geographic requirements for political parties and for election of presidents and vice presidents. This combination enabled Indonesia to reap the benefits of a PR electoral rule without sacrificing too much in the way of accountability usually attributable to more majoritarian systems. As Reilly (2011, 291 and 294) argues, centripetal political institutions have a majoritarian thrust without being strictly majoritarian because they pull political parties and candidates toward the moderate center forcing both to sell themselves as good median choices.

5. A test of equality of means does not reject the hypothesis that mean growth rates in real income per capita are the same under democracy (t = .53, p = .59.) or under centripetal democracy (t = .10, p = .91) as under the New Order. This result does not change if New Order growth rates are from 1970 to 1998 or if the focus is on growth rates for real GDP instead of real GDP per capita where an equality of means test for real GDP growth rates under the New Order for 1970–1998 compared to growth under centripetal democracy does not reject (t = .46, p = .64) the hypothesis that mean growth rates are the same.

6. Tests of the hypothesis that mean growth rates in seven countries in Southeast Asia (Cambodia, Laos, Malaysia, the Philippines, Singapore, Thailand and Vietnam) were no different than that in Indonesia were decisively not rejected in five (Cambodia, Malaysia, the Philippines, Singapore and Thailand) of the seven cases. Most of the economic comparisons focus on Malaysia, Thailand, and the Philippines (see, for example, Jomo 2001, Khan and Jomo 2000, Bowie and Unger 1997, MacIntyre 1994), but much of the political science literature also includes consideration of Singapore (see, for example Kuhonta, et al 2008, Case 2002 and Taylor 1996).

7. Indonesia's average growth rate in real GDP per capita between 2004 and 2012 was 4.35%, Malaysia grew at 3.19%, the Philippines grew at 3.36%, Thailand grew at 3.35% and Singapore grew at 3.32% per year (World Bank 2013)

8. A test of the hypothesis that the growth rate of real GDP per capita in East Asia and the Pacific between 2004 and 2012 was no different from that between 1970 and 1998 is decisively not rejected (t = 1.19, p = .24).

9. A test of the hypothesis that the growth rate of real GDP per capita in the world economy between 2004 and 2012 was not different from that during the New Order (1970–1998) is decisively not rejected (t = .03, p = .97) (World Bank 2013).

10. If we date the New Order from 1970, instead of 1966, a simple equality of means test decisively rejects the hypothesis that there was no difference in investment rates (t = 2.67, p = .01)

11. A simple equality of means test barely rejects the hypothesis that there was no difference in investment rates (t = −1.71, p = .09). This difference disappears if we date the New Order from 1970 (t = −1.14, p = .25)

12. Variable definitions and sources appear in Table 1, descriptive statistics appear in Table 2.

Table 2. Descriptive statistics for Indonesia data.

Table 3. Indonesia: steps in the creation of a centripetal democracy.

13. Government consumption expenditures as a share of GDP averaged 9.09% under the New Order and 8.68% during Indonesia's centripetal democracy (World Bank 2013).

14. That said, Indonesia continues to lag others in the region in attracting foreign direct investment (Lipsey and Sjoholm 2011) and only the Philippines in Southeast Asia attracts less FDI than Indonesia (World Bank 2013).

15. The mean value for control of corruption is 2.64 during centripetal democracy and 1.53 during the New Order (Political Risk Service 2013).

16. The basic format of this regression is in the spirit of Levine and Renelt (1992, 946) who begin with a stripped down Solow like model which includes relative income, investment as a share of GDP and population growth as base regressors to which they add a series of other variables thought to effect growth.

17. The estimated equation is

In this regression, YNG is the annual growth rate of real GDP per capita, RELYN is Indonesia's income relative to the USA, IY is the investment share in GDP, POPG is the population growth rate, GCY is government consumption expenditures as a share of GDP, FDIY is the net inflow of foreign direct investment as a share of GDP, TRDY is the trade share in GDP, INFLATE is the inflation rate of the GDP deflator, and CCORR is a measure of control of corruption. Adjusted R²=.55. The t values below each regression coefficient suggest that most of the coefficients are statistically significant with the expected sign. An augmented Dickey–Fuller test rejects the assumption that the error term has a unit root (ADF= −5.16***). The Breusch–Godfrey serial correlation LM test reveals no serial correlation though 4 lags (F = .50, p =.73) and the Q statistic from a correlogram on the residual reveals that the error term is not subject to serial correlation through 16 lags. The Jarque–Bera statistic (JB = 5.66, p = .06) does not reject the hypothesis of a normal distribution of the residuals at the .05 and .01 level. Because the control of corruption variable might be endogenous, I instrumented it with a measure of ethnic tension taken from Political Risk Services (2013). Treisman (2000), among others, includes ethno-linguistic fractionalization in his regressions on corruption while Rock (2009) finds that this ethnic-tension variable increases corruption in a panel of countries. I used a Hausmann test to test for endogeneity. The regression coefficient on the residual from the auxiliary regression was nowhere near statistically (t = .41, p = .68) suggesting that CCORR can be treated as exogenous. Except for CCORR, all of the others variables are taken from World Bank (2013) and cover the period from 1981 to 2012 – the years for which the World Bank reports data on FDIY. Data definitions and sources appear in Table 1. Descriptive statistics appears in Table 2. Because the growth rate in 1998 is such an outlier, it is omitted from the regression.

18. The null hypothesis of no structural break is decisively not rejected (F = .33, p = .94).

19. Very similar results were obtained by regressing each of the variables thought to affect growth on past values of those variables and testing the hypothesis that there are structural breaks in these variables with the onset of Indonesia's centripetal democracy. Tests results are available from the author on request.

20. For rigorous empirical tests of the impact of centripetal democracy on economic growth in a panel of East Asian countries, including Indonesia, see Rock (2017, chapter 8).

21. I use the term centripetal democracy rather than majoritarian democracy because Indonesia does not possess a classic majoritarian democracy with, for example, a first past the post electoral rule (personal communication Marcus Mietzner).

22. Between 1990 and 1997, real GDP grew at an annual average rate of 7.58%, inflation averaged 8.7%, the fiscal balance as a share of GDP was 1.2%, the current account balance as a share of GDP was −2.45%, the Gini Index was 29.9 and the head count incidence of poverty was 17.6% (World Bank 2013).

23. Following the New Order's shift in 1989 to openness (keterbukaan), there was significant growth in oppositional activity prior to the collapse of the New Order, but the opposition was fragmented and ineffectual (Aspinall 2005, 242).

24. In 1998, Indonesia was relatively poor, ethnolinguistic fractionalization was high, it possessed a small middle class and an equally small working class, and its religion was thought to be an impediment to democracy (Webber 2006, 403).

25. Even though oppositional politics grew during the 1990s, few in the opposition believed they could push Suharto from power (Aspinall 2005, 252).

26. Approximately 19,000 people died in the violence that engulfed Indonesia (Aspinall 2010a, 26).

27. Tinkering with electoral rules, political party laws, and constituency size continued well past 2004 and it benefited larger parties and incumbents (Horowitz 2013, 199-206).

28. The thresholds required parties to capture at least 2% of the seats in the DPR or 3% of the seats in local legislatures to run in subsequent elections (Sulistyo 2002, 81).

29. District magnitudes for the 1999 elections varied from 4 to 82 (Sulistyo 2002, 80).

30. The presidency was initially weakened because the 1945 constitution empowered the executive more like a monarch than a president (King 2004, 76). Parliament weakened the presidency by denying the president the right of executive privilege, eliminating the pocket veto, restricting the president's residual powers, requiring an annual accountability speech, and by provisions making it easy for parliament to impeach a president (King 2004, 53–54 and 77).

31. As Mainwaring(1993) has shown, multi-party presidential systems tend to be unstable because they are prone to executive-legislative gridlock, ideological polarization and because it is difficult to build and sustain inter-party coalitions. Liddle and Mujani (2006) agree.

32. The presidency was strengthened by the Third and Fourth constitutional amendments which empowered the presidency by way of direct national elections and by provisions making it more difficult to impeach a president (King 2004, 54).

33. Decentralization by-passed provinces and focused on local governments to minimize secessionist threats and enable the central government to maintain central control (Fitrani et al. 2005, 61).

34. Parties that ran for election in 1999 could only run in 2004 if they held 2% of the seats in the DPR or 3% of the seats in the DPRD in one-half of the provinces or 3% of the DPRD seats in one-half of Indonesia's regencies (Choi 2009, 679). Parties also had to have executive committees in two-thirds of the provinces, two-thirds of the regencies and at least 1000 party members in both or 1/1000 of the population as party members in regencies (Choi 2009, 680). These rules increased disproportionality in the assignment of votes to seats from 1.54 in 1955 to 4.45 in 2004 and 6.84 in 2009 and they reduced the effective number of parties in parliament from 7.07 in 2004 to 6.13 in 2009 (http://www.tcd.ie/Political_Science/staff/michael_gallagher/ElSystems/)

35. For the 2009 presidential election, parties had to have captured at least 25% of the popular vote and 20% of the seats in the DPR (Choi 2004, 682). In addition, parties or aggregations of parties had to run presidents and vice presidents as a team. Finally, unless first round winners had an absolute majority and at least 20% of the vote in one-half of the provinces, elections were to be conducted over two rounds.

36. The failure of the Konstituante was attributed to deadlock on the issue of the role of Islam in the state (Horowitz 2013, 26).

37. There is one important exception, even though Tim Tujuh proposed adoption of plurality electoral rules in single member districts (Horowitz 2013, 60–63), the political elites driving reform favored a PR list rule because the former was thought to give too big an edge to Golkar (Horowitz 2013, 64).

38. In 2004, Yudhoyono campaigned on a promise to achieve a 6% GDP growth rate (Liddle and Mujani 2006, 137) and he delivered on it as real GDP grew by 5.9% during his first term (World Bank 2013). He also promised to clean up corruption, which he did as Indonesia's score on the Control of Corruption Index of the Political Risk Service rose from a low of 1 throughout Megawati's presidency to 3.58 by the end of Yudhoyono's first term (Political Risk Service 2013).

39. Public opinion polls show parties are not trusted by voters (Tan 2006, 104) and the poor economic (real GDP grew at 4.5% World Bank 2013) and anti-corruption performance of Megawati's presidency cost her votes (Liddle and Mujani 2006, 133).

40. During Yudhoyono's first term, his most popular initiative was the launching of an aggressive anti-corruption campaign that netted a number of high profile arrests of legislators, bureaucrats and investigators (Meitzner 2009, 150–151). As a result, Indonesia jumped 17 ranks in Transparency International's Corruption Perceptions Index (Mietzner 2009, 147). Moreover, Mujani and Liddle (2010, 37–38) argue voters have become more rational by setting standards, identifying goals, and choosing leaders based on each.

41. This has not come without setbacks (Tomsa 2010, 311–313 and 315).

42. But there is some evidence of efforts to undermine Indonesia's Anti-Corruption Court (Butt 2011, 381–382).

43. This is quite important because as Shleifer and Vishny (1993) argue, this kind of corruption is likely to be more damaging to growth than the centralized corruption extant during the New Order.

44. Indonesia's score on the Bureaucratic Quality Index declined during Megawati's presidency, but has been steady since (Political Risk Services).

45. What follows draws on Bird et al. (2008, 952–954).

46. That said several recent developments are troubling and cause for some worry. First, parties and individual candidates tend to adopt protectionist policy stances during elections, even if they back away from these protectionist stances following elections (Personal comment from Marcus Mietzner). Second, non-tariff barriers have been rising for several food and horticultural products (Nehru 2013, 155–157) as well as for several natural resource products and minerals (Marks and Rahardja 2012, 66–68). Third, the new draft trade and industry laws grants Indonesians preferential access to natural resources and they provide for protection from the threat of global competition (Nehru 2013, 160). While individually these developments may not add up to much, taken together they suggest that protectionist instincts may not be too far from the political surface in democratic Indonesia.

47. The Islamic parties have accepted defeat in the constitutional debate and have suspended demands for an Islamic state (Mietzner 2008, 452). They have reached out to non-Muslims (Aspinall 2010a, 29) and they have been tarred with corruption and patronage (Aspinall 2010a, 31).

48. The Islamic parties have also been forced to the center by declining vote shares (Aspinall 2010a, 29).

49. It also freed political elites to re-engage in the kind of patronage politics that afflicted the New Order (Webber 2006, Slater 2004 and Hadiz 2003).

Additional information

Notes on contributors

Michael T. Rock

Michael T. Rock is the Samuel and Etta Wexler Professor of Economic History at Bryn Mawr College, Bryn Mawr, PA USA. His published research focuses on industry and the environment in Northeast and Southeast Asia, industrial policy in Southeast Asia, and the politics of development policy in Southeast Asia. His most recent book – Dictators, Democrats and Development in Southeast Asia – was published by Oxford University Press in 2017. He is currently working on a book titled Democracy, Development and Islam.