Morten Jerven Morten Jerven

The Fiscal State in Africa: Evidence from a century of growth

I am very happy to announce that the database on taxation for African countries that covers the whole 20th century is finally ready. Download it here. With Thilo Albers and Marvin Suesse, we have been collecting tax and revenue data on all African polities since the 1890s until present time. Before this dataset the most comprehensive coverage was from the 1980s until today as described in this article by Wilson Prichard. There has also been some groundbreaking work to present taxation data on the colonial period such as by Ewout Frankema and Marlous van Waijenburg, who has worked mostly on British colonies. Philip Havik has worked on the Portuguese colonies, and Denis Cogneau, Yannick Dupraz, Sandrine Mesplé-Somps has done comparative work on taxation in the French Empire. Nevertheless, even with all this empirical research over the past decades, and furthermore despite the consensus view, namely that taxations is important for development, there is no simple answer to basic question such as whether Nigeria taxes more today than it did 20, 30, 50 or 100 years ago. Let alone any database that could compare the taxation burden in Nigeria with that of Niger. We can do that now.

Of course one challenge has been to physically collect all the data. We have collected data in French, Italian, Portuguese, English and German archives as well as using national statistiscal abstracts or IMF reports to get the data from the post-colonial period. But the biggest problem is of course how to express it comparable terms across time and space. The common summary statistic to express taxation is government revenue as a share of GDP. One problem is that GDP is mismeasured in post-colonal times, but the far bigger obstacle is that there is very few estimates of GDP for the period before 1960 (some exceptions). We therefore have collected urban unskilled wages for all polities for all the country years. When you deflate revenue by daily wages you get a readily interpretable expression of real revenue. And so we can present:

Figure 1: Real taxation, revenues, and their composition in Africa over the last century

Please read the whole paper, but let me stress a few remarkable things:

  1. Data availability. The availability of real revenue data on African economies just increased from partial coverage in 1980-2010 and sporadic time series for some countries 1900-1940 to a complete time series from 1900- 2010. We consider that a major contribution in itself, and since we offer a metric of fiscal capacity across the 20th century we expect this to be a major resource for other researchers.
  2. Growth. Fiscal capacity grew more than tenfold. The mainstream approach to African states and taxation is focussing on low capacity and stagnation. Our time series challenges that view.
  3. Direct Taxes. Direct taxes are consided “hard to collect”. Look at the high share of direct taxes in the colonial period. We think it makes the case for re-considering different paths to the “modern fiscal state”  in Africa.
  4. Natural Resources. The sterotypical ‘natural resource curse’ view of the African state is that it depend little on direct tax, and much on natural resource rents. This is true only for the recent growth period of ‘Africa Rising’.
  5. Timing. The perception of “state weakness” originates mostly from the crisisperiod between 1980 and 2000, neglecting periods of strong growth before and afterwards.

All in all the paper suggest that the question “why do African states tax so little” might be to reductive and misleading. Thanks to this new dataset we can now explore and test more nuanced theories and research questions.

Sharing:

Leave a Reply

Your email address will not be published. Required fields are marked *

Privacy Preference Center

Necessary

Cookies that are necessary for the site to function properly.

gdpr

Analytics

These are used to track user interaction and detect potential problems. These help us improve our services by providing analytical data on how users use this site.

_ga, _gat, _gat_, __cfduid