Survival of the fittest: productivity versus corruption effects
Indonesian manufacturing plants for the purpose of this study is interesting in several respects. The history of Indonesian industrialisation has been largely studied and documented at the macroeconomic and industry levels (Booth, 1992; Hill, 1996a). Microeconomic studies have shown that in the case of Indonesian manufacturing, micro-dynamic approaches can inform our understanding of Indonesian economic growth rates, and feed the debate over the intensity versus extent of economic growth. Vial (2006) estimates total factor productivity (TFP) growth rates at the establishment level, and shows that allowing for plant heterogeneity leads to higher TFP growth rates for the period 1975-1995. Ter Wengel and Rodrigez (2006) study plant entry and exit during the period 1994-2000, describing ‘vibrant firm dynamics’ (abstract), and attribute a large part of labour productivity gains to the replacement of lower-productivity exiters by higherproductivity entrants. Vial (2008) conducts a parallel study over the period 1975-1994, and finds that without the contribution of small and medium plants’ turnover, TFP growth rates would turn out to be negative for the Indonesian manufacturing sector. Behrman and Deolalikar (1989) propose an analysis of survival duration factors for the period 1975-1986. The authors find that age, size, industry concentration, labour productivity and bribe payments have a positive effect on survival duration. They find that state ownership has no significant effect, while foreign ownership increases survival duration. A higher number of family workers lower survival duration, while the average wage of workers has no impact. Bernard and Sjöholm (2003) study the survival of Indonesian manufacturing plants over the period 1975-1989, and find that a larger size, along with greater labour productivity, increases their survival probabilities. These two studies describe a fairly competitive picture of the Indonesian sector, with labour productivities influencing positively plant survival, while also recognising the potential positive effect of bribe payments that, however, only seem to grease fairly competitive markets. This is in line with Vial and Hanoteau (2010), who demonstrate that corruption, measured as bribe payments and indirect taxes, boosts both output and labour productivity growth. In a perfectly competitive environment we would expect a relatively lower TFP to be one of the main determinants for exit. However, many other studies have underlined that the process of exit is not a clean process. Indeed, since there are several strategic games and agency problems in the decision to stay or exit, there is no obvious reason why TFP would be the main determinant for exit and survival. For example, within an industry with a small number of players, each player has an incentive to force competitors to exit in order to capture their market share. But if exit costs are too high, low productivity plants may delay the decision to exit. Another example is the case where plant closure would result in high unemployment costs for the state, so the state could prevent or delay exit. In the same vein, managers and workers can delay or prevent closure decided by the owners in order to protect their jobs. Last but not least, exit may occur only when quasi-rents have been exhausted: a rational plant will exit when the profits expected in the next period are less than the interest that could be
gained from reallocating the assets elsewhere. Exit may be delayed or prevented if assets have depreciated very quickly, or if the possibilities of reallocating assets are limited. Last but not least, and particularly relevant in the case of Indonesian manufacturing: exit can be prevented by soft budget constraints on some plants, allowed by political protection; for example, via cheap and easy access to credit, or by the non-existence or non-implementation of bankruptcy laws. Corruption can also help low-productivity firms secure a market to the detriment of more productive competitors. Two questions then arise. First, in the widespread corrupt settings that seem to boost individual plants’ output and productivity growth, how do productivity and corruption influence hazard rates? Second, does this effect change after a shift in competitive conditions via changes of the regulatory environment? We test the determinants of hazard rates on the entire population of firms. First, we discuss the potential determinants of hazard rates in the case of Indonesian manufacturing by means of economic history framing. Second, we present the data used. Third, we present our methodology and results using the non-parametric Kaplan-Meier survivor function along with the Cox proportional hazard function to empirically test potential determinants of plants’ hazard rate. The last section concludes.