Trends in South African Agricultural Land Prices¶
Obi, Ajuruchukwu (2006). Trends in South African Agricultural Land Prices. PhD thesis, Department of Agricultural Economics, Faculty of Natural and Agricultural Sciences, University of the Free State. University of the Free State, Bloemfontein.
| Type | thesis |
| Scope | Wider apartheid literature |
| Method | econometric |
| Period studied | 1652–2003 |
| Regions | South Africa, Orange Free State/Free State, Qwaqwa, Bophuthatswana/North West |
| Themes | agriculture land, origins segregation, bantustans removals influx, macro fiscal finance debt, business and state, transition post apartheid, methods data sources |
| Citation key | obi2006 |
Argument¶
The thesis models real South African farmland prices for 1955-2003 using residual-based cointegration and an error-correction model, to understand the recent price rises and what they mean for post-1994 land reform. It finds real farmland prices tied positively to real GDP per capita, farm debt and net farm income, with policy-regime dummies and structural breaks (1970s, 1980s, c.2001-02) important. It finds no basis for concluding that rising prices are hurting land reform, and little support for the claim that foreign buyers drive prices. Chapters 2-3 supply a literature-based history of land policy from 1652 through apartheid-era farm support and deregulation to land reform, plus a review of land-price theory and methods. Policy recommendations are to release state land, promote equity-sharing and leasing, and give post-settlement support.
Evidence and method¶
Annual national time series 1955-2003: Directorate of Agricultural Economics Trends, Deeds Registry, HSRC (Aliber and Mokoena), Reserve Bank (capital-inflow proxies, exchange rates), Statistics SA and CSS, Standard Bank rates, and CPI deflators (2000=100). Van Schalkwyk's (1995) 1955-1992 dataset is extended. Chapter 2 relies on secondary literature and official documents (DLA/DoA 2005, Land Summit papers, Claasen 2000, Jordaan and Jooste 2004, Golder Associates 2004). Analysis was run in Microfit and E-Views.
Place in the debates¶
- Foreign buyers and the post-2000 rise in farmland prices (see Agricultural policy, productivity, farm size and land reform). Sceptical: the proxy results do not support government concern about foreign buyers, and the land market is argued to be segmented, though the data are weak and the conclusion tentative.
- Do rising land prices undermine market-assisted (willing-buyer willing-seller) land reform? (see Agricultural policy, productivity, farm size and land reform). The results give no basis to conclude that prices hurt reform, but price spikes create uncertainty. The thesis favours state land release, debt-relief purchases and leasing, and stresses post-settlement support.
- Sustainability and effects of apartheid-era farm support and the origins of deregulation (see Agricultural policy, productivity, farm size and land reform). Follows Van Schalkwyk, Van Zyl and Bayley: support to white farmers was unsustainable, and deregulation began in the mid-1980s, before 1994. Chow tests (1985, 1970/80/92 breaks) are read as confirming this.
- Methodological dispute over log transformation and trend- versus difference-stationarity in cointegration studies of land prices (Hallam vs Lloyd and Rayner) (see Data reliability and econometric method). Reviews both sides and uses level data with policy dummies, treating all series as non-stationary and differencing them.
Findings¶
-
The 1913 Land Act confined Africans to a small share of the land, later extended by the 1936 Act. (p. 26)
The Native Land Act No. 27 of 1913 consigned the majority black population to just about 8% of the nation’s 122 million hectares of land
-
The thesis links land to rural labour supply for mines via the Glen Grey Act, as an aim of the Act. (p. 26)
succeeded in virtually completing the proletarianization of the African farmer and forcing him to accept mine employment, a central aim of the Act
-
Apartheid-era farm policy used above-world producer prices and subsidies, which contributed to unsustainability and prompted reform from the 1980s. (p. 51)
the farmers faced state-established producer prices that far exceeded world prices for the commodities such as maize
-
Real farmland prices peaked in the late 1970s, fell until 1993, were stable to 2000, then rose sharply; by 2003 they were near their 1955 level. (p. 177)
real farmland prices rose steadily up to about 1977/78 when they attained their highest level ever. Following that peak, real farmland prices began falling steadily up to 1993.
-
Chow tests indicate multiple structural breaks and a single break in 1985, which the thesis reads as evidence that agricultural restructuring predated 1994. (p. 226)
agricultural restructuring had actually commenced long before the exit of the Apartheid regime.
-
With policy dummies, real GDP per capita and real farm debt per hectare are highly significant positive long-run determinants (Table 6.1: GDP per capita coefficient 0.104544, t=6.87). (p. 234)
RGDPC 0.104544 0.015229 6.865007 0.0000
-
The thesis treats GDP per capita as robustly positively related to land prices irrespective of policy assumptions. (p. 257)
regardless of the policy environment, the real GDP per capita maintained a positive and highly significant relationship with real farmland prices
-
Foreign-buyer proxies provide no strong support for the view that foreign buyers drive prices; the short-term inflow proxy is insignificant once policy effects are included. (p. 230)
the analysis does not provide sufficient support for the current concerns about foreign buyers.
-
Error-correction adjustment is fast: the lagged residual coefficient is -1.45, implying deviations are corrected within about a year. (p. 254)
The results in Table 6.4 show that the coefficient of the lagged residual is -1.45.
-
Land redistribution under the post-1994 reform has been slow against its target. (p. 47)
by February, 2005, the programme had managed to re-distribute no more than 3.5 million hectares to some 168,000 households.
Evidence ledger¶
Sources used¶
| Source | Kind | Period | How used |
|---|---|---|---|
| Directorate of Agricultural Economic Trends data (land values, farm area, number of farms, real estate debt, net returns to farming), 1955-1992 series used by Van Schalkwyk (1995) | official statistical series | 1955-1992 | Supplies the original series from which real farmland prices per hectare, net farm income and farm debt are calculated. |
| Deeds Registry and HSRC farmland price data, as reported in Aliber and Mokoena (2003) | secondary literature compilation | 1993-2003 extension | Extends the farmland price series from 1992 to 2003. |
| Van Schalkwyk (1995) dataset, 'Modeling South African Agricultural Land Prices' (PhD, Pretoria) | secondary literature compilation | 1955-1992 | Baseline time series of land prices and macro aggregates that the study updates and extends to 2003; also its structural model guides variable choice. |
| Statistical Abstracts / Central Statistical Service 'South African Statistics' (1993, 1994) and Statistics South Africa series | official statistical series | 1955-2003 | Source for real interest rate on debt, farm debt, CPI/inflation and other macro series. |
| South African Reserve Bank time series KBP5548J and related: foreign liabilities, direct investment by private non-banking sector, other short-term and long-term capital | international trade or finance data | 1956-2003 (1955 imputed) | Proxy for foreign buyers' purchases of farmland (rfbsph, rfblph) in the regression. |
| Reserve Bank / Standard Bank historical interest rates (Standard Bank Intranet, 2005) | stock market financial data | 1955-2003 | Fresh interest-rate data for the full period, checked against Van Schalkwyk's series and deflated to give real interest on debt. |
| Reserve Bank exchange-rate data (Rand/US$ real exchange rate; effective exchange rate from 1971) and Van der Merwe (1996) historical account | international trade or finance data | 1955-2003 | Builds the continuous real exchange rate series (rsaus/russa) used as an explanatory variable. |
| Consumer Price Index series (2000=100, 1985=100, 1995=100), Appendix 1 | official statistical series | 1955-2003 | Deflator for converting nominal values to constant 2000 prices and to compute inflation. |
| Real GDP per capita series (Reserve Bank/SSA) | national accounts macro | 1955-2003 | Used as rgdpc, the proxy for ability to pay and non-agricultural land demand. |
| Reserve Bank personal communications (Swart 2006; Mandeya 2006) on 1955 missing observation | interviews oral history | 2006 | Advice to set the 1955 foreign capital inflow values equal to 1956 values. |
| Department of Land Affairs / Department of Agriculture (2005) 'Land and Agrarian Reform in South Africa: An Overview' (Land Summit document) | commission or official report | 1994-2005 | Source for land transfer figures (3.5 million ha to 168,000 households), restitution outcomes (Table 2.2) and land reform design. |
| Katz Commission (1998) Eighth Interim Report, Land Tax Sub-Committee | commission or official report | 1995-1998 | Cited on the land tax and capital gains tax debate and the tax's potential effect on land prices. |
| AgriBEE Framework (Department of Agriculture 2004) and Swanepoel et al. (2004) evaluation of empowerment policies | commission or official report | 2004 | Establishes policy context requiring land values not be negatively affected and the proposals on state land, joint ventures and leasing. |
| Department of Agriculture (2006) Economic Review of South African Agriculture 2005/06 | official statistical series | to 2006 | Used to note that net farm income continued declining after the sample ended. |
| Legislation: Native Land Act 27 of 1913, Native Trust and Land Act 18 of 1936, Marketing Act 1968, Marketing of Agricultural Products Act 1996, Restitution of Land Rights Act 22 of 1994, ESTA 1997, Labour Tenants Act 1996, CLARA 2004, Capital Gains Tax Act 2001 | legislation gazette | 1913-2004 | Frames the policy regimes and break dates (1968, 1985, 1996, 2001) used for the structural break and dummy-variable analysis. |
| Claasen (2000) survey of 114 emerging farmers in Qwaqwa; Jordaan and Jooste (2004); Golder Associates (2004) Taung irrigation scheme assessment | secondary literature compilation | 1979-2004 | Descriptive case evidence on apartheid-era emerging farmer schemes and their post-1994 difficulties. |
| Press and commentary on foreign buyers: Property24.com (2005), Aida National Franchises (2002), Real Fact (2005), National Association of Realtors (2005), ABSA house price indices, The Economist (18 June 2005) | newspapers periodicals | 2002-2005 | Evidence on perceived foreign buyer role, their share (3-10% of buyers) and housing price comparison (Table 3.2). |
| Media and government statements on land reform and foreign ownership: SABC-2 (2004, 2006), Business Day (2006), Reuters (2006), Budget Speech 2006, State of the Nation 2006, Land Summit speeches (Didiza 2005) | newspapers periodicals | 2004-2006 | Context for perceptions of rising land prices, alleged collusion and foreign ownership policy. |
| Previous empirical studies' data on South African land prices (Behrmann and Collett 1970; Hattingh and Herzberg 1980; Nieuwoudt 1980; Davies 1983; Van Schalkwyk and Groenewald 1993) | secondary literature compilation | 1939-1993 | Literature-based evidence on land price trends and links to farm income. |
New data¶
Annual 1955-2003 time series (49 observations) of real farmland prices per hectare and nine other macro/farm variables, assembled by extending Van Schalkwyk's 1955-1992 series to 2003 with Deeds Registry, HSRC, Reserve Bank and SSA data. Includes a constructed land-transfer dummy (0 before 1995, 1 for 1995-1999, 2 from 2000), an imputed 1955 foreign inflow value, and a spliced CPI and exchange-rate series. Listed in Appendix 2.
Inference¶
Quantification: Econometric (causal design).
Single-equation residual-based (Engle-Granger) cointegration on a 1955-2003 national annual time series. It uses OLS levels regression with annual policy dummies, recursive coefficient plots, Chow breakpoint tests, unit-root tests, an error correction model and pairwise Granger tests. Chapters 2-3 are literature review and descriptive policy history.
Key claims¶
| Claim | Basis | Strength |
|---|---|---|
| Real farmland prices have a strong positive long-run relationship with real GDP per capita and real farm debt per hectare, and a positive relationship with real net farm income. (p. 216) | OLS long-run regression of rlph on nine variables plus annual dummies (Table 6.1), 1955-2003. | Partial data |
| Real farmland prices in 2003 were about the same as in 1955, and the sharp rise began around 2000 after a long decline from the 1977/78 peak. (p. 159-160) | Constructed real land price series (Figure 4.2), deflated by the spliced CPI. | Systematic data |
| The data show multiple structural breaks (1970, 1980, 1992 and a single break in 1985) linked to policy regimes. (p. 208) | Chow breakpoint tests with dates chosen from recursive coefficient plots (Tables 5.3-5.4). | Partial data |
| Foreign capital inflows (proxy for foreign buyers) show no significant positive effect on farmland prices, so the evidence does not support concerns about foreign buyers. (p. 230) | Reserve Bank short- and long-term private non-bank capital inflow proxies in the long-run and ECM regressions. | Partial data |
| Deviations from long-run equilibrium are corrected rapidly, within about a year. (p. 254) | Error correction model with lagged residual coefficient of -1.45 (Table 6.4). | Partial data |
| Land transfers to the black population show a negative and weakly significant relationship with real farmland prices, and the land reform dummy is not what drives price rises. (p. 244) | Crude 0/1/2 dummy (lndtr) from 1995 in the regressions because no transfer time series existed. | Illustrative or anecdotal |
| Apartheid-era law and subsidy policy dispossessed black farmers and favoured white commercial farmers, shaping the land market that later reform confronted. (p. 8) | Secondary literature (Davenport and Hunt, Van Schalkwyk 1995, DLA/DoA 2005, Bayley 2000); no primary archival work. | Asserted |
The author's caveats¶
The author concedes that the foreign buyer variable is only a proxy including other investments, that data on actual land transfers to black owners are unavailable (hence the dummy), and that land quality and improvements cannot be netted out. Data on collusion and non-agricultural land use were unavailable, the series is short, and the 1955 foreign inflow value is imputed. Results are called tentative and exploratory.
Gaps in the evidence¶
The argument needs transaction-level Deeds Registry data disaggregated by buyer nationality, race, land use and province to test the foreign buyer, land reform and non-agricultural demand claims. Actual redistribution volumes and grant amounts, and land quality-adjusted prices, would better test the policy effects than dummies. Much of this (Deeds records, DLA transfer data) existed then or has since become obtainable; the historical apartheid-era claims also lack any primary archival or Commission evidence.
Replicability: Data described but not public.