Factor Proportions, Openness and Factor Prices in Kenya 1965–2000

Abstract

This study analyses how changes in factor abundance and openness have affected relative factor prices in Kenya since 1965, using cointegration analysis and error correction models of relative factor prices. We find that factor proportions determined relative factor prices in the long run, while openness, measured by three different proxies, possibly had a short run effect on relative factor returns. The only deviation from this pattern occurred during the latter half of the 1990s when there was rapid wage growth, mainly due to labour market deregulation.

Acknowledgements

We would like to thank participants in seminars at Warwick University, New York University, Cornell University, Umeå University and Göteborg University, particularly William Easterly, Yaw Nyarko, Erik Thorbecke, David Sahn, Stephen Younger, Per Fredriksson and two anonymous referees, for helpful discussions and comments. Financial support from SAREC is gratefully acknowledged.

Notes

1. The recent review by Anderson (2005), notes that ‘there has been a large amount of research into the effect of openness on one particular factor price, the wage of skilled relative to unskilled labor’ (Anderson, 2005: 1051) and, ‘We still know little about whether, and if so how, greater openness has affected other factor-price ratios, such as the return to land (and other natural resources) relative to labor, skilled or unskilled, in low- and middle income countries. Movements in this latter ratio may have a much larger impact on overall inequality in Africa and Latin America, with their relatively abundant supplies of land and other natural resources' (Anderson, 2005: 1057).

2. Manda and Sen (2004) found that trade liberalisation had a negligible impact on the structure of the manufacturing sector, which points in the same direction as our results.

3. See Appendix 1 for details about the data used in this section.

4. The variable used for land is arable land (in hectares), which includes land defined by the FAO as land under temporary crops, temporary meadows for mowing or for pasture, land under market or kitchen gardens, and land temporarily fallow.

5. Minimum wages are still set by the Ministry of Labour but since they are increasingly falling behind average wages their contribution to the general wage increase seems to be rather modest.

6. The deviation between the price indexes lasted until 2003, when the GDP deflator caught up with the consumer price index. One reason for the deviation was a series of devaluations during the period 1989–93 that accumulated to 180 per cent for the Shilling-US dollar exchange rate. From 1993 to 2004 the Shilling was only devalued by 36 per cent (World Development Indicators, 2005).

7. See Appendix 1 for a description of the data on land prices.

8. Export taxes have not been important in Kenya.

9. Athukorala and Rajapatirana (2000) use the same measure for an analysis of the Sri Lankan experience.

10. We also calculated the index with US prices but the overall pattern was very similar. Since imports from the UK are much larger than those from the US, we prefer to use UK prices. In principle, a trade-weighted index would be the best measure but annual revisions of the trade weights generate abrupt changes in the index due to the high level of time aggregation.

11. We use the tests of Zivot and Andrews (1992) and Perron (1997) that allow for unknown breakpoint. See Patterson (2000: 277–283) for details.

12. All the unit root and cointegration tests were implemented with RATS procedures. The results from the unit roots tests can be obtained from the authors upon request.

13. Unfortunately, we do not have p-values for the Gregory and Hansen (1996) test.