Military spending and the black market premium in developing countries

Abstract

Researchers who have been concerned with the economic implications of military spending have mostly concentrated on its impact on economic growth, corruption, real exchange rate and inflation. In this paper we investigate the impact of military spending on black market premium, an area that has not been tackled so far. After adding a measure of military spending to a well established model of black market premium form the literature, we estimate the model by pooling annual data over the 1985 – 1998 period across 61 developing countries. Results from five panel specifications provide considerable evidence that higher military spending leads to higher black market premium.

Keywords::

• black market premium
• military spending
• developing countries

Acknowledgements

We would like to thank two anonymous referees for valuable comments. Any error, however, is ours.

Notes

The term “peace dividend” is defined as the benefits derived from lower defense spending and the conversion of military production into civilian production. For detailed literature on “peace dividend” see Gleditsch et al. (1996) and, especially for a definition provided by Intriligator, see p. 1.

For a detailed literature review on military spending see Dunne (1996) in Gleditsch et al. (1996), pp. 439 – 464.

Hundi (also hoondee, hoondi, hoondy) is defined as a negotiable instrument, such as a bill of exchange or promissory note, used by native bankers in India and worded in the vernacular; also money remitted by such an instrument (Oxford English Dictionary, 2nd edn, 1989, p. 490).

Note that the Dornbuschet al. (1983) model assumes that output is exogenous and the desired proportion between domestic and foreign currencies is given by a liquidity preference function. For this and more criticism of the model see Agénor (1992).

The online data sources for the black market premium, and the official real exchange rate arewww.worldbank.org/research/growth/GDNdata.htm, andwebhost.bridgew.edu/baten/ respectively.

Since data range for all countries is not the same, the analysis is one of unbalanced panel. This approach enables us to include more countries and thus increases the number of observations and degrees of freedom. It is important to include more countries because, for this particular project, country variation is more pronounced than time variation.

We greatly acknowledge the receipt of data on military spending from Dr Luiz de Mello.

We have used SAS Windows version 8 to perform most of the estimations. The details of each panel estimation technique are available in the SAS online documentation, especially in the TSCSREG procedure of theSAS User's Guide (Section SAS/ETS).

As an additional exercise, we calculated the variance inflation factor (VIF) to check for multicollinearity and found that it is approximately equal to one for all regressors except intercept. This implies that the estimation does not suffer from multicollinearity.

For example, India and Pakistan both received their independence from the British rule in the year 1947, while their current economic structures are heterogeneous and the reasons are difficult to detect in most of the cases.

The random-effect model can be estimated either through using FGLS or MLE. The result from the FGLS is only produced here. However, using MLE we have obtained the similar result.