Tanzi, V., “Inflation, Lags in Collection, and the Real Value of Tax Revenue,” Staff Papers. Vol. 24, (March 1977), pp. 154–167.
Tanzi, V., “Inflation, Real Tax Revenue, and the Case for Inflationary Finance: Theory With Application to Argentina,” Staff Papers. Vol. 25, (September 1978), pp. 417–451.
The author wishes to thank Peter Wickham and Ratna Sahay for providing valuable econometric insights. He has also benefitted from suggestions by Abbas Mirakhor, Omotunde Johnson and Domenico Fanizza. Thanks are also due to Raja Hettiarachchi for providing excellent research assistance.
The insight on collection lags and its importance for developing countries was developed by Tanzi (1977). The analysis has been further elaborated by others, including Tanzi (1978), and most recently by Choudhry (1990), who analyzes and provides empirical evidence on the revenue-eroding effects of inflation for a large number of developing countries.
The expression e-βiπ is obtained by taking the limit as π tends toward zero.
If wi’s vary with inflation, then the derivative of equation (3) yields
the change in β is expected to be of second order magnitude.
This assumption is only for expositional purposes.
Ibid., pp. 3-6, for a discussion of the revenue-maximizing inflation rate in the context of collection lags.
Annual data covering country-specific intervals in the period 1970-1988 were obtained for the following countries listed according to regions: in Central America, Costa Rica, Guatemala; in Asia, Bangladesh, Myanmar, India, Pakistan, Philippines, Singapore, and Sri Lanka; in Africa, Botswana, Ethiopia, Ghana, and Zaire; in the Middle East, Egypt, Iran, Jordan, Syrian Arab Republic, and Yemen Arab Republic. All data used in this study were taken from the International Monetary Fund’s data bases, Government Finance Statistics and International Financial Statistics.
Nontax revenue, which comprises profit transfers, rent and income from various government undertakings, is included in fiscal revenue in line with the classification in Government Finance Statistics.
The constant and the aforementioned 17 dummies are omitted from this presentation and that of the estimated revenue equations in Appendix Table 1, as they indicate country-specific differences in the level of revenue in the pooled regression.
For Yemen Arab Republic, the CPI was based on 1980 = 1.0 because of nonavailability of data beyond 1984.
In Choudhry (1990), the average collection lag was estimated at four months. This is probably because the earlier study had some high inflation countries where indexation was the dominant form of changes in taxation.
Although the focus is not on the buoyancy of government revenue, the estimated value of the parameter γ = 1.113, is in line with evidence for developing countries. When the tax system is progressive or discretionary measures are taken to bolster revenue, and buoyancy is likely to be higher than unity. Like the parameter β, the country-specific values of β show wide variation, indicating differences in tax structures and the type and frequency of discretionary tax measures taken by the countries.
Excises are levied on quantity or volume while the valuation basis for import taxes is cost and freight, quoted in foreign currency at the time of shipment. As such, these tax bases may not keep pace with interim inflation and/or exchange rate depreciation. In contrast, “other” taxes from international trade and transactions, comprised of taxes on exports and exchange profits, show negative collection lags because exchange rate depreciation increases the bases for such taxes.
Since the real fiscal revenue ratio, R(π), is an average over the sample period, the estimated revenue erosion by inflation takes into account automatic and discretionary revenue effects.
A value of 2 for the coefficient of expected inflation, indicating that an increase in expected inflation reduces the stock of real money balances, may be roughly compared with one that is derived from changes in the velocity of money (v). To see this, consider a linear velocity equation:
where vo = y/m(0) can be regarded as the initial velocity when inflation is zero. Then taking the limit as π tends to zero, we obtain:
so that α = v1/v0.
The velocity equation was estimated using the pooled cross-section annual data for the 18 sample countries covering the same intervals of the 1970-88 period. The estimated equation was:
These coefficients yield a value of 1.2 for the coefficient of expected inflation. As Khan’s study employed a more elaborate model of monetary adjustments to expected inflation, a value of 2 for the coefficient of inflation is considered appropriate.
In view of the moderately high rate of inflation and the small size of real inflation revenue, a question may arise about whether inflationary financing was not inappropriate altogether in a number of these countries.