52 SAJEMS NS Vol 1 (1998) No 1 

Causes of Bankruptcy amongst Maize and 
Extensive Beef Farmers in South Africa: 1970-
1994 

D S Swaneplel, G F Ortmann and MAG Darroch 

Department of Agricultural Economics, University of NaJal. Pietennaritzburg 

ABSTRACT 

The number of maize fanns and extensive beef fanns annually declared bankrupt 
in South Africa rose sharply over the period 1970 to 1994. Principal components 
regression confirmed a priori expectations that maize farm and extensive beef farm 
bankruptcies were negatively related to annual rainfall (business risk factor), but 
positively related to the lagged aggregate farm debt/asset ratio and lagged real 
interest rates (financial risk factors). Maize farm bankruptcies also increased as 
lagged real maize and beef producer prices fell (business risk factors). Beef fann 
bankruptcies rose with lower Jagged real beef producer prices and higher lagged 
real stockfeed subsidies and transport rebates (business risk factors). Part of the 
rise in maize and extensive beef fann failures between 1970 and 1994 can 
therefore be ascribed to changed agricultural price and macroeconomic policies 

INTRODUCTION 

The number of maize fanns and extensive beef fanns annually declared bankrupt 
in South Africa rose sharply over the period 1970 to 1994. Bankrupt maize fanns 
increased from 16 to around 150 farms per year, while bankrupt extensive beef 
fanns increased from 12 to about 50 per year (Van Niekerk, 1995). In 1988, there 
were about 7 500 maize fanns and 2 500 extensive livestock farms in the summer 
rainfall and cattle grazing regions of South Africa respectively (Central Statistical 
Service. 1988a and 1988b; Directorate Agricultural Statistics. 1996, p.IIO). 
While the rate of bankruptcy for both fann types is relatively low. the marked rise 
in the number of bankrupt fanns is of concern. Maize is the major field crop in 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATEB NR Voll (1998) Nr 1 53 

South Africa comprising some 40 per cent by value of all field crops, and about 
nine per cent of the gross value of all agricultural products in 1994/95. Beef 
constituted approximately 37 per cent of the gross value of animal products, and 
approximately 11 per cent of the total gross value of agricultural production in 
1994195 (Directorate Agricultural Statistics, 1996). Therefore, maize and 
extensive beef farm bankruptcies impose major adjustment costs on the agricultural 
economy and give rise to demands for government assistance to alleviate financial 
stress for farmers. Substantial government involvement generally arises when a 
large number offarming operations are threatened with bankruptcy. For example, 
nominal subsidies on farm carry-over debt in South Africa in 1992/93 totalled 
R2,7 billion (Directorate Financial Assistance, 1996). Given that farm debt is 
concentrated in the maize and extensive beef sectors (Human, 1989; Volkskas 
Bank, 1988), research on the causes of farm failure can help to identify 
appropriate future policy and management measures to avoid having to reorganise 
an insolvent business or liquidate the business and pay creditors (Barry et al., 
1995, p. 557). 

Shepard and Collins (1982) studied aggregate United States (US) farm 
sector bankruptcy data over the period 1910 to 1978. Prior to World War II, the 
farm bankruptcy rate appeared to be linked with financial risk (leverage), while 
postwar bankruptcy was associated with business risk factors (variable real net 
farm income). Agricultural suppon payments since World War II did not induce. 
defer or reduce farm failures. Chan and Rotenberg (1988) identified financial 
leverage and energy-related expenses as key causes of farm loan arrears and 
ultimate bankruptcy in Canada during 1979 to 1986. Davies (1996) found that the 
annual rate of insolvency in agriculture in England and Wales from 1%9 to 1986 
was negatively related to the current price of land, but positively related to the land 
price two years previously. Past Common Agricultural Policy price supports, 
which were capitalised into higher land values that encouraged farmers to use 
more debt. were thus partly responsible for higher insolvency. 

In South Africa, Van ZyJ et aI. (1987) found that the initial farm solvency 
position. nominal interest rates and inflation together affected survival of 'typical' 
Western Transvaal and Nonh-Western Transvaal Bushveld farms. Leslie and 
Darroch (1993) reponed that successful farms (positive long-run real return on 
equity) in Natal, the Eastern Orange Free State and Western Transvaal in 1993 
had higher rates of return to assets and equity and lower costs of debt than 
unsuccessful farms. Rates of return to assets on successful farms exceeded costs of 
debt, implying positive use of leverage. De Jager and Swanepoel (1994) identified 
insolvent farmers in the Nonhern Springbok Hats during 1990 as having higher 
directly allocatable costs, relatively more carry-over debt, liquidity problems, less 
land as collateral and lower gross farm incomes relative to long-term debt. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



54 SAJEMS NS Vol 1 (1998) No 1 

Given that no published study, to the authors' knowledge, has yet analysed causes 
of fann bankruptcy at a product sector level, this paper considers sources of 
business and financial risk which may have caused maize and extensive beef fann 
bankruptcies in South Africa to rise since 1970. Bankruptcy trends and possible 
causes are outlined in the next two sections, after which research methodology and 
results are reported. A concluding section considers the management and policy 
implications of the results. 

TRENDS IN BANKRUPTCIES OF MAIZE AND EXTENSIVE BEEF 
FARMERS IN SOUTH AFRICA: 1970-1994 

Figure 1 shows trends in annual maize and extensive beef fann bankruptcies in 
South Africa during 1970-1994 for the areas defined by the Directorate 
Agricultural Statistics (1996, p.1IO) as summer rainfall and cattle grazing areas 
respectively. 

Maize fann bankruptcies rose from 16 in 1970 to 206 fanns by 1986 and 
then fluctuated around the 150 farm level. Bankrupt extensive beef farms 
followed a similar pattern, rising from 12 fanns in 1970 to 62 farms in 1987, 
fluctuating around the 50 farm level during 1988 to 1992, and then falling to 35 
farms in 1994. TIle absolute level of farm failures and bankruptcy rate are higher 
for maize farmers compared to extensive beef farmers. The fall in both maize and 
beef farm bankruptcies in 1993 and 1994 was probably due to a drought relief 
package (carry-over debt subsidy and loan guarantee scheme instalment) in 
1992193 totalling some R3,O billion (Directorate Financial Assistance, 1996). TIle 
following section explains possible causes of the sharp rises in the level of farm 
bankruptcies. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATEB NR Vol 1 (1998) Nr 1 55 

Fagure 1: Maize and Extensive Beef Farm Bankruptcy Trends in South 
Africa, 1970-1994 

250 70 
en 

./\ 
!!:! 

60 t.1 I-
~ 200 a.. 

f\~ 
::l 

U 
50 I I-!5 <II: 

; 150 m 

f \/ \ 
40 ::IE 

<II: c:: 
CD <II: 

~ 100 
.... 

30 .... 

j ~ 
w 
W 1.1.. m 

w .,......1 20 ~ N ~ 50 !2 ::IE 
Wf~ 10 ~ x w 

0 0 
1970 1975 1980 1985 1990 1994 

YEAR 

--- MAIZE -- BEEr 

Source: Directorate Agriculrural Economic Trends (1995); Central Statistical Service (1994 and 
1995); Van Niekerk (1995). 

POssmLE CAUSES OF BANKRUPTCY OF MAIZE AND EXTENSIVE 
BEEF FARMERS IN SOUTH AFRICA 

Commercial maize and extensive beef farmers in South Africa experience business 
and financial risk:. BusiIX'!sS risk: refers to risk: inherent in a busiIX'!ss independent 
of the way the busiIX'!ss is financed and is reflected in variability of IX'!t operating 
income. It arises from factors such as price variability in both output and input 
markets. Financial risk reflects added variability of IX'!t cash flows due to fixed 
financial obligations associated with debt financing (Gabriel & Baker. 1980). 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



56 SAJEMS NS Voll (1998) No 1 

Business Risk Factors 

Product and /npw Prices 

Success or failure in fanning is closely linked to prevailing trends in output and 
input prices. Farmers are usually debtors and this makes them vulnerable to a 
decline in farm product prices (fomek & Robinson. 1991. p.l79; Tweeten, 1985, 
p.78). Variable product and input prices can impact on farm failure rates by 
producing wide fluctuations in farm income (liquidity effects). Lower real net farm 
income is likely to increase bankruptcy rates (Shepard & Collins, 1982). Reliable 
net farm income data for the maize and extensive beef sectors were not available, 
hence product prices were used as a proxy for net farm income. Real input prices 
since 1970 have remained relatively stable (Directorate Agricultural Statistics. 
1996). A negative relationship between real producer prices and farm bankruptcy 
is thus expected. In this analysis, producer prices are adjusted to real terms using 
the Consumer Price Index (CPl) (1990= 100). 

In South Africa, an added risk dimension affecting maize farm incomes 
would be the fall in real maize producer prices since the 1987/88 marketing year 
when Maize Board pricing policy changed and losses on export sales were 
reflected in lower fixed real net maize producer prices (Farninow & Laubscher, 
1991). The Board administered a single-channel pool price maize marketing 
system until 1995, whereby farmers had to market maize grain via the Board or its 
agents. The real producer price of maize set by the Maize Board in South Africa 
declined from R350/ton to R290/ton between 1987/88 and 1993194. 

Maize farmers in the sununer rainfall area derive some 30 per cent of gross 
farm income from beef cattle (Central Statistical Service, 1992) which can be sold 
to provide liquidity in times of financial stress. A negative relationship between 
real beef producer price and maize farm bankruptcy is thus expected. 

Drought 

Drought is expected to increase bankruptcies by reducing net cash flows and 
creating financial stress. Particularly severe drought conditions occurred in the 
summer rainfall and extensive beef areas in 1982, 1990, 1991 and 1992 (CCWR, 
1996). Annual rainfall in the summer rainfall and cattle grazing areas of South 
Africa was used as a proxy for drought conditions. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATED NR Vol 1 (1998) Nr 1 57 

Agricultural Policy 

Government support, such as drought relief schemes, can improve farm liquidity 
by reducing current liabilities and hence improving the prospects of short-tenn 
survival (Standard Bank, 1994). Rucker and Alston (1987) found that govenunent 
programmes successfuUy aUeviated farm financial stress in the 1930s in the US. 
Decreased government support has possible costs; if, e.g., future farm policy is 
more market orientated with reduced credit programmes, then private lenders and 
farmers are likely to feel the effects of future farm recessions more severely 
(Drabenstotl, 1983). 

Stockfeed purchase subsidies paid to extensive beef farmers in South Africa 
since 1 %5 may have promoted more intensive production in higher risk 
production areas. Also, stockfeed transport rebates to these farmers could have 
caused production over time to relocate away from areas where beef had a 
comparative advantage in production (Nieuwoudt, 1985). Therefore, a positive 
relationship is anticipated between bankruptcy and lagged real stock feed subsidies 
and transport rebates. Annual subsidies were adjusted to real terms using the CPI 
(1990=100) (Directorate Agricultural Statistics. 1996). It is paradoxical that 
policies intended to make farming less risky. may lead to more risk for farmers in 
the Iong-tenn. 

Fmancial Risk Factors 

Real/nterest Rates 

Expected real interest rates are a critical explanatory variable for invesnnent 
decisions as they represent the real cost of borrowing (Mishkin, 1988). Murdock 
and Leistritz (1988, p.4S) identify both direct and indirect effects of high nominal 
and real interest rates on agriculture. The direct effects include an increase in 
interest payments due from indebted farmers, and a negative impact on land 
values. Indirect effects include a higher value of the local currency, as relatively 
high interest rates in local markets attract capital from abroad. 

Higher than expected real interest rates transfer wealth from debtors to 
creditors, placing farmers who are net debtors at a disadvantage (fweeten. 1985, 
p.l(0). When real interest rates are expected to be relatively high, farmers should 
shift from debt to retained earnings to finance expansion (Drabenstotl. 1983). In 
these periods of tight money and high interest rates, lenders are less willing to 
extend loan terms and farmers are less able to afford additional credit. Low 
interest rates assist potential bankrupts who acquire credit to see them through 
difficult times and prevent foreclosure. This would suggest a positive relationship 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



58 SAJEMS NS Vol 1 (1998) No 1 

between real interest rates and farm bankruptcy. High real interest rates add to 
fann expenses and reduce real wealth by increasing the rate of discount on 
expected future earnings of durable farm resources resulting in declining collateral 
for loans. High nominal interest rates create cash flow problems but not 
necessarily low returns because land values appreciate with inflation that led to the 
high norninal interest rates for farmers (fweeten, 1985, p.lOO). 

Interest rates affect agriculture directly through cost and stock effects 
(Devadoss, 1985; Rausser, 1988, p.l50). A higher interest rate will increase the 
cost of production through higher financing costs which in rum will decrease farm 
supply (cost effect). An increase in the interest rate will raise the cost of holding 
stocks causing farmers to run down inventories (stock effect). This reflects the 
increased opportunity cost of non-farm invesonent in interest-bearing assets 
(Hughes et aI., 1985; Rausser, 1985, p.220). 

Financial sector reform in South Africa during the rnid-1980s led to the 
banking sector's reserve requirements being changed, and subsidised interest rates 
to farmers from the Land Bank were discontinued (Vink, 1993). In response to 
the De Kock Commission's recommendation in 1983, monetary policy became 
more market-orientated and market-related interest rates were increasingly applied 
to agljculture. This subjected the farming sector to a 'double' increase in interest 
rates; firstly from a decline in subsidised interest rates, and secondly, due to the 
imposition of positive real interest rates on the economy as a whole. Commercial 
bank overdraft interest rates adjusted to real terms using the change in CPI 
(1990 = 100) are used in this study as a proxy for market interest rates. Annual 
real overdraft interest rates fell from two per cent to around -1,5 per cent from 
1970 to 1975, rose to one per cent for the period 1976 to 1978 and fell to -4,5 per 
cent by 1980. De Kock Commission recommendations for more market-
orientated commercial and Land Bank interest rates led to historically high real 
overdraft interest rates of between five and 10 per cent during 1983 to 1985, while 
positive real rates between 2,5 and 6,5 per cent have continued since 1988 (South 
African Reserve Bank, various years; Standard Bank, 1994). More market-related 
rates imply greater expected future interest rate volatility and higher financial risk. 

Debt/Asset (Leverage) Ratio 

The aggregate fann debt/asset Oeverage) ratio (total farm debt as a per cent of total 
farm assets) shows the solvency and risk-bearing ability of farmers. Farmers with 
substantial net worth or equity have the potential to borrow additional funds to 
meet shon-term needs (Murdock & Leistritz, 1988, p.78). Higher debt burdens 
imply higher fixed debt service charges and greater financial risk, as debts must be 
repaid in high and low income years. This increases the probability that highly 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATEB NR Vol 1 (1998) Nr 1 59 

leveraged fanners will face difficulties in servicing debt (Chan & Rotenberg, 
1988). Increasing financial leverage increases the variation of expected returns on 
equity and the potential for loss of equity capital, and reduces liquid credit 
reserves. Furthennore, variations in interest rates magnify these financial risks as 
leverage increases (Barry et aI., 1995, p.I69). Leverage and bankruptcy of maize 
and extensive beef fanners are expected to be positively related. Farm sector 
leverage in South Africa rose from 0,06 in 1970 to a peak of 0.17 in 1985 and 
remained around 0,15 to 1994. These relatively 'safe' aggregate leverage levels 
mask the distribution of farm debt, which was concentrated in the summer rainfall 
and cattle grazing areas over this period (Central Statistical Service, 1972 and 
1994; Human 1989; Volkskas Bank, 1988). Favourable accelerated depreciation 
allowances on machinery investment, negative real interest rates in the early 1980s 
and drought in the early 1980s and 19908 probably encouraged use of more debt. 
Investing in equipment and land was an attractive alternative to paying income tax. 
This probably encouraged borrowing from farm co-operatives and commercial 
banks and brought about a level of mechanisation that was economically 
unsustainable. 

Higher real interest rates would reduce asset values and increase the 
leverage ratio, thereby reducing solvency and increasing the potential for 
bankruptcy. This interaction is incorporated in the models of maize and extensive 
beef farm bankruptcy specified later. A real return problem occurs when interest 
rates remain high and disinflation removes capital gains as a compensating return 
(Tweeten. 1985. p.87). A high ratio of debt to assets becomes a low return 
problem when interest rates exceed total rates of return on assets for an extended 
period. Highly leveraged farms are likely to experience cash flow problems 
because the rate of return to assets is less than the interest rate that must be paid 
(Murdock & Leistritz. 1988, p.78). Realising this problem. local lenders now 
place more emphasis on fanner repayment capacity than on fanner collateral. 
Farmers must now have the ability to cover expected production COSts, fixed costs, 
existing commitments (capital and interest) and personal expenditures from farm 
and non-farm income (Louw, 1995). 

There is likely to be a time lag between the incidence of business and 
financial risk factors and ultimate farm bankruptcy. For example. drought and 
higher interest rates in one year will affect borrowers' future ability to meet debt 
repayments. as they reduce present income (and possibly savings) and raise the 
commitments against future income (Rucker & Alston. 1987). Lagged proxy 
variables for business and financial risk in the maize and extensive beef farm 
bankruptcy models estimated below are used to indicate that the bankruptcy 
process is dynamic. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



60 SAJEMS NS Vol 1 (1998) No I 

RESEARCH METHODOLOGY AND RESULTS 

Factors affecting bankruptcy of maize and extensive beef fanners during 1970 to 
1994 were estimated from time series data using ordinary least squares (OLS) 
regression and principal component analysis. Regional sequestration data were 
obtained for the swnmer rainfall and cattle grazing areas over this period for all 
farms exceeding 50 hectares, excluding farm companies and close corporations 
(Van Niekerk, 1995). The number of farm sequestrations was taken as a proxy 
for the number of farm bankruptcies. 

Maize Farm Bankruptcy Model 

The preliminary OLS model is given by equation (1): 

BANKRM = a. + al RMP + lU WEA + 3.l LEV +a. RINT + 3.lIL + at. RBP + e, (1) 

where BANKRM = number of annual maize farm bankruptcies; RMP = lagged 
real maize producer price; WEA = lagged annual rainfall in summer grain areas 
(annual rainfall is used because of the importance of soil moisture levels for crop 
growth) (CCWR, 1996); LEV == lagged annual farm sector leverage ratio; RINT 
= lagged real annual commercial bank overdraft interest rate; IL interaction 
term (RINT x LEV) showing how higher real interest rates reduce asset values and 
heoce increase leverage and potential bankruptcy; RBP = lagged real farm beef 
price, and e. == disturbance term. Correlation coefficients reported below were 
used to identify the appropriate length of lag for the explanatory variables. The 
LEV variable is a reasonable proxy for maize farm leverage as farm debt is 
coocentrated in the swnmer rainfall and cattle grazing areas (Human, 1989; 
VoJkskas Bank, 1988). 

CorrelaJion Coejficiems 

A correlation matrix of the variables used in the analysis is presented in Table I. 
All coefficient signs agree with a priori expectations, with BANKRM negatively 
related to RMPI (real maize producer price in previous year), WEAl (annual 
rainfall in swnmer grain areas in previous year) and RBPt (real beef producer 
price in previous year). However, BANKRM is positively related to RINTI (real 
annual commercial bank interest rate two years prior), LEVI (farm sector leverage 
lagged one year) and IUt (RINTI x LEVI). Multicollinearity is likely to be a 
problem due to statistically significant pairwise correlations between the 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATEB NR Vol I (1998) Nr I 61 

explanatory variables. For example, RINTI and LEVI were significantly 
correlated with lUI and RBPI, and lUI respectively, at the I per cent level. 

Table 1: Correlation Coefficients Between Variables of the Maize 
Bankruptcy Model, 1970-1994 

BANK- RMPI RINT2 LEVI WEAl 1L21 RBPI 

RM 

BANK 1,0000 
-RM 

RMPI -0,4531"' 1.0000 

RINT2 0.6663'" -0.0906 1.0000 

LEVI 0.8418'" -0,3688' 0,4998" 1,0000 

WEAl -0,4011' 0,0122 -0,4704" -0.4393" 1.0000 

1L21 0,6959'" -0.0896 0.9790'" 0.5474'" -0.4767" 1.0000 

RBPI -0,4470" 0,4312" -0,6606'" -0.2705 0.3730' -0.5939" 1,0000 

Note: ***, ** and * mdlcate slgruficance at the I per cent, 5 per cent and 10 per 
cent levels respectively. 

Regression model 

The initial maize farm bankruptcy model estimated by OLS (GENSTAT, 1993) 
was: 

BANKRM = 30,427 - 0.202 RMPI+ 0.012 WEAl + 945.783LEVI + 2.935 RINT2 + 

(-1,615) (0,154) (3.515)'" (0.277) 

27,568IL21+0,023 RBPI 

(0,407) (0,193) (2) 

where adjusted R2 = 75,87 per cent, d= 1,077, degrees of freedom = 16, t-
values are in parentheses, and *** indicates significance at the I per cent level. 

The Durbin-Watson d statistic for detecting autocorrelation falls in the iocooclusive 
range, but the Geary test passed at the 5 per cent significance level, so the 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



------ .... -~-

62 SAJEMS NS Voll (1998) No I 

hypothesis of randomness is accepted (Gujarati, 1995, p.419). However, the high 
adjusted R2, non-significant coefficients for RMPI, WEAl, RlNTl, lUI and 
RBPI and wrong signs for the WEAl and RBPI coefficients indicate expected 
multicollinearity. Principal compo~nts extracted from the standardised 
explanatory variables (ZRMP1, etc.) to cope with this problem are shown in Table 
2. 

Table 2: Principal Components Extracted for the Maize Farm Bankruptcy 
Model,I97O-l994 

Principal Component 

Variable 1 PCI PC2 PC3 PC4 PCs PC6 

ZRMPI 0,18784 -0,84989 0,01351 0,04826 0,48923 0,02333 

ZWEAl 0,35692 0,26765 0,57445 0,63961 0,24870 0,00519 

ZLEVI -0,39584 0,21561 1-0,58040 0,45820 0,49893 -0.03269 

ZRINTl -0,50236 -0,23602 0,28239 0,21360 -0,21157 -0,72236 

ZIUl -0,50015 -0,24167 0,20173 0,30876 -0,29644 0,68377 

ZRBPI 0,42205 -0.21319 -0.46100 0,48750 -0,56323 -0,09495 

Eigen 3,292 1.156 0,789 0.550 0,197 0,015 

value 

% vari- 54,86 19,27 13.16 9,17 3.28 0,26 

ation 

The principal compo~nts (PCs) are used to restate equation (2) in terms of the 
original variables purged of multicollinearity (Chatterjee & Price, 1977). 
Standardised annual maize farm bankruptcy, ZBANKRM, is first regressed on 
PC I, PC2 and PC3. These three PCs explain 87 .29 per cent of the variation in the 
explanatory variables (PC4, PCs and PC6 were omitted as they showed the linear 
relationships between the explanatory variables which led to multicollinearity and 
instability in the model): 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATEB NR Vol 1 (1998) Nr 1 63 

ZBANKRM =- 0,444 PCI + 0,217PC2 - 0,231PC3 (3) 

(-6,940)'" (2,010)' (-1,770)* 

where adjusted R2 = 70,40 per cent, d = 0,991, degrees of freedom = 19, t-
values are in parentheses, and *** and * indicate significance at the I per cent and 
10 per cent levels respectively. 

The Durbin-Watson d statistic falls in the inconclusive range (1 per cent 
significance level), and the hypothesis of randomness is still accepted as the Geary 
test passed at the I per cent significance level. Standardised annual maize farm 
bankruptcy could also be estimated by OLS regression of ZBANKRM on the 
standardised explanatory variables as per equation (4): 

ZBANKRM = bl ZRMPI + b2 WEAl + b3 LEVI + b4 RINT2 + bs IL21 + 
~~I ~ 

The b coefficients of equation (4) can be estimated from equation (3) coefficients 
and the PCI, PC2 and PC3 loadings in Table 2 as: 

k 

hi = l:mijIlij 
. j=l 

(5) 

where ffiij = estimated loading for variable i in PCj, Ilij = estimated coefficient for 
PCj from equation (3), and k = number of PCs retained. For example, bl = 
(0,18784 x -0,444) + (-0,84989 x 0,217) + (0,01351 x -0,231) = -0,271. 
Substituting these expressions into equation (4) gives the estimated standardised 
maize farm bankruptcy regression model as: 

ZBANKRM= -0,271 ZRMPI- 0,233 ZWEAI +0,356 ZLEVI + 

0,107 ZRINT2+0,123 ZIL21-O,127 ZRBPI (6) 

The standardised variables are independent of the original units of measurement, 
and their coefficients show the relative importance of the variables. Lagged 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



64 SAJEMS NS Vol 1 (1998) No 1 

leverage. ZLEVI. is the most important explanatory variable. followed by the 
lagged real maize producer price. ZRMPl, lagged annual rainfall, ZWEAl, 
lagged real beef price, ZRBPl, the interaction term, ZIL21, and lagged real 
interest rate, ZRINn. Standard errors and t-values of the b coefficients were 
estimated following Gujarati (1995, p.70). The t-values are equivalent to those in 
original scale since scaling does not affect the correlation of the variables. Finally, 
the regression coefficients in equation (6) were multiplied by SBANKRMlSx; (standard 
deviation of BANKRM divided by standard deviation of the relevant explanatory 
variable) to express the amended OLS annual maize farm bankruptcy model in 
original scale (Chatterjee & Price, 1977, p.178) as per equation (7): 

BANKRM = 244,800 - 0,235 RMPI - 0,140 WEAl + 612,818 LEVI + 
(-2,921) ••• (-2.779)" (4,271)'" 

1.840 RINT2 + 14,265 IL21 - 0.082 RBPI (7) 
(1,933)* (2,512)" (-1,816)* 

where adjusted R2= 70,40 per cent. t-values are shown in parentheses, and "'n . 
• '" and-· indicate significance at the I per cent, 5 per cent and 10 per cent levels 
respectively. 

Compared to equation (2), the adjusted R2 falls slightly but the t-values 
increase markedly. All coefficients are now statistically significant and the WEA I 
and RBPI coefficient signs are correct. Estimation of factors affecting bankruptcy 
of extensive beef farms is described in the next section. 

Extensive Beef Farm Bankruptcy Model 

The preliminary OLS model is given by equation (8): 

BANKRB = co+ Cl RBP+ C2 RINT+ c; LEV+ CA WEA + C5 RBS + coiL + e, (8) 

where BANKRB = annual number of bankrupt extensive beef farms; RBP = 
lagged real beef producer price; RINT = lagged real annual commercial bank 
overdraft interest rate; LEV = lagged farm sector leverage ratio; WEA = Jagged 
anrmaI rainfall in cattle grazing area (CCWR, 1996); RBS = lagged real stockfeed 
subsidies and transport rebates; IL = lagged interaction term (product of lagged 
RINT and lagged LEV, showing that higher real interest rates reduce asset values 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATEB NR Vol 1 (1998) Nr 1 65 

and hence raise leverage and potential bankruptcy), and et = disturbance tenn. 
The LEV variable is again a reasonable proxy for extensive beef farm leverage as 
farm debt is concentrated in the sununer rainfall and cattle grazing areas (Human. 
1989; Volkskas Bank, 1988). Correlation coefficients were used to identify the 
most appropriate length of lag for the factors affecting BANKRB. 

Correlation coefficients 

A correlation matrix of the model variables is shown in Table 3, where BANKRB 
is negatively related to RBPI (at the 15 per cent significance level) and WEA4. 
but positively related to RINT3, LEV3, RBS3 and IL43 (the number(s) in each 
variable again show(s) the lag period). These coefficient signs agree with a priori 
expectations. Multicollinearity is again likely to be a problem due to statistically 
significant pairwise correlations between most of the explanatory variables. 

Table 3: 

BAN· 
KRB 

RBPI 

RlNT 
3 

LEV3 

WEA 
4 

RBS3 

IU3 

Correlation Coefficients between Variables of tbe Extensive Beef 
Bankruptcy Model, 1970-1994 

BANKRB I ! RINT3 LEV3 RBS3 

1,0000 

-0,3085 

-0,3221 1,0000 

0,9158'" -0,3626' 0,5182'" 1,0000 

..(1,6337"" 0.4038' -0.5716'" ..(1,6083'" 1.0000 

0,6319'" 0,1929 0,4173" 0,5885'" -0,4473" 1,0000 

0,4121' -0,0090 0,5876'" 0,5689'" -0,3649' 0,4600" 1.0000 

Note: ***, ** and * indICate slgruflCance at the 1 per cent, 5 per cent and 10 per 
cent levels respectively. 

The three and four year lags in the explanatory variables are expected as beef 
herds take from three to four years to build up (Directorate Agricultural Statistics, 
1996), A real interest rate increase. for instance, would probably take a number 
of years to impact on farm bankruptcy due to the liquidity effect of fanners' 
destocking decisions. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



66 SAJEMS NS Vol 1 (1998) No 1 

Regression model 

The initial extensive beef farm bankruptcy model estimated by OLS (GENST AT, 
1993) was: 

BANKRB= 13,463 - 0,042 RBP1+ 0,513 RINT3 + 318,178 LEV3 - 0.006 WEA4 

(-1,855)- (0,931) (4,497)'" (-0,372) 

+ 0,095 RBS3 - 2,472 1L43 (9) 

(2,006)" (-0,678) 

where adjusted R2 = 86,40 per cent, d = 1,544, degrees of freedom = 14, t-
values are in parentheses, and *** and * indicate significance at the I per cent and 
10 per cent levels respectively. 

Expected multicollinearity occurs in equation (9) as the IL43 coefficient is 
not statistically significant and has the wrong sign. The model has a high adjusted 
R2 but the coefficients of variables RINTJ and WEA4 are not statistically 
significant. The Vurbin-Watson d statistic falls in the inconclusive range, but the 
hypothesis of randomness is accepted (the Geary test passed at the 5 per cent level 
of significance). Extracted principal components to remedy multiCOllinearity are 
shown in Table 4. 

Table 4: Principal Components Extracted for the Extensive Beef 
Bankruptcy Model, 1970-1994 

Principal Component 

Variable I PCI PC2 PC) PC4 PC~ POI 
ZRBPI 0,20901 0,77605 0,08287 0,26251 0,16217 0,50198 

ZRINTJ ~'~~ ~,47238 0,55335 -0,57255 0,23684 
ZLEV3 -0,48 , 0,26170 -0,57255 ~,18101 0,57431 

ZWEA4 0,44198 0,24607 -0,37438 -0,43112 -0,64285 -0,06950 

ZRBS3 ~,40651 0,47060 0,43355 0,06505 ~,33541 ~,55551 

ZIL43 -0,39492 0,33010 ~,61105 ~,32709 0,45390 -0,22062 

Eigenvalue 2,968 1,292 0,694 0,528 0,353 0,165 

% variation 49,46 21,54 11,57 8,80 5,88 2,76 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATEB NR Vol I (l998) Nr I 67 

Standardised annual beef farm bankruptcy, ZBANKRB. is regressed on PCl and 
PC2. 1bese two PCs explain 71 per cent of the variation in the explanatory 
variables (the other components, as sources of the multicollinearity and model 
instability, were omitted), as per equation (10): 

ZBANKRB = - O,S04 PC1 - 0,115 pe2 (10) 
(-7,665)'" (-1,152) 

where adjusted R2 = 74,40 per cent, d = 1,082, degrees of freedom = 18, t-
values are in parentheses, and *** indicates significance at the 1 per cent level. 

The Durbin-Watson d statistic falls in the inconclusive range at the 5 per cent 
significance level, and the hypothesis of randomness is still accepted as the Geary 
test again passed at the 5 per cent significance level. Following the same 
procedure outlined in equation (5), the Standardised beef farm bankruptcy model 
was estimated as: 

ZBANKRB= - 0,195 ZRBPI + 0,234 ZRINT3 + O,2S1 ZLEV3 - 0.251 
ZWEA4 + 0,151 ZRBS3 + 0,161 ZIL43 (11) 

Lagged leverage, ZLEV3, and lagged annual rainfall, ZWEA4, are the most 
important explanatory variables, followed by the lagged real interest rate ZRINT3, 
lagged real beef producer price ZRBPl, the interaction term ZIL43 and lagged 
real subsidies ZRBS3. The amended OLS beef farm bankruptcy model in original 
scale was: 

BANKRB= 42,269 - 0,036 RBPI + 1,130 RINT3+ 126,473 LEV3-

(-2,476)" (7,717)" (7,744)'" 

0,035 WEA4+0,OS RBS3+S,367 IL43 (12) 

(-6,603)'" (2,798)*" (3,848)*" 

where adjusted R2= 74,40 per cent, t-values are shown in parentheses, and *** 
and ** indicate significance at the 1 per cent and 5 per cent levels respectively. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



68 SAJEMS NS Vol 1 (1998) No I 

Compared to equation (9), the adjusted R2 falls but remains relatively high, the t-
values increase markedly and all estimated coefficients are highly statistically 
significaru.. 

The regression coefficient estimates in equations (7) and (12) are biased as 
SOtre information was lost by dropping respective PCs, but the new estimates have 
more precision than the OLS estimators in equations (2) and (9) (Chatterjee & 
Price, 1977, p.l7S; Doran, 1989, p.106). 

CONCLUSIONS 

Higher lagged aggregate farm leverage and lagged real interest rates (financial risk 
factors), and lower annual rainfall (business risk factors) increased bankruptcies of 
South African maize and extensive beef farmers over the period 1970 to 1994. 
Farm bankruptcy in both sectors was also negatively related to lagged real 
producer prices (business risk factors). Bankruptcy of extensive beef fanners was 
positively related to lagged real stockfeed purchase subsidies and transpon rebates 
(business risk factors). Bankruptcies of both fann types were positively related to 
an interaction term between real interest rates and the aggregate leverage ratio. 
Bankruptcy is therefore a dynamic process, with a time lag between the incidence 
of causal factors and ultimate farm failure. Lags for the beef bankruptcy model 
are longer than for the maize bankruptcy model, probably due to the longer 
production cycle for beef and beef farmers being able to increase liquidity via 
stock sales. 

When estimated standardised coefficients for both models are compared, the 
aggregate farm debt/asset ratio is the most important determinant of farm 
bankruptcies. Higher leverage probably reflects a combination of poor borrowing 
decisions by eventual bankrupts, past tax policy measures (e.g., accelerated 
depreciation allowances on machinery investments) which may have contributed to 
increased debt use aId fann bankruptcies, and past monetary policy (negative real 
interest rates) which made borrowing attractive. 

Changes in Maize Board maize producer price policy from 1987/88 created 
another source of risk for maize farmers to manage. Recent further deregulation 
of dotrestic maize pricing means that maize farmers must give more attention to 
managing price risk, possibly by forward contracting, electronic marketing or 
hedging a ponion of their maize crop via recently introduced futures contracts on 
the South African Futures Exchange (SAFEX) or enterprise diversification. 
Producer price and rainfall effects on extensive beef farmers emphasise the need to 
build up fodder banks to counter drought, and possibly use forward contracting 
and hedging a portion of their intended beef sales via new SAFEX beef futures 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATES NR Vol 1 (l998) Nr 1 69 

contracts 00 manage price risk:. Stockfeed purchase subsidies and transpon rebates 
intended 00 help beef farmers led 00 more risk and potential bankruptcy for some 
of these farmers in the long-term as they probably encouraged beef production in 
unsuitable areas. 

Macroeconomic policy changes towards more market-related real interest 
rates directly affected maize and extensive beef farmers by raising fmancing costs 
and indirectly raised leverage and potential bankruptcy. Stable monetary policy 
over time can thus contribute to stability in the maize and extensive beef sectors. 
Highly leveraged farmers are particularly vulnerable to higher interest rates 
associated with deflationary policy. Farmers must closely monitor changes in 
agricultural price, trade and macroeconomic policies to form accurate expectations 
of potential bankruptcy causes and improve management of debt and business and 
financial risk: at farm level. Specialist extension personnel, consultants and lenders 
need to advise clients on the relationship between net farm income, interest costs 
and leverage levels for successful debt management. Fanners could use improved 
information now available from local researchers on forecasting shon-term 
regional weather patterns to better manage recurring droughts (e.g., reduce 
fenilizer input at planting if below average rainfall is expected). 

A vailable data limited the study to analysis of farm bankruptcies at maize 
and extensive beef regional level, but more research is needed on the individual 
characteristics of bankrupt farmers. For example, are farmers operating relatively 
larger farms going bankrupt? and are YOWlger, more leveraged farmers, or those 
less able to manage business and fmancial risk, failing? Highly leveraged maize 
and extensive beef farmers may have incurred more debt than they can realistically 
service in a changing agricultural and macroeconomic policy environment. 
Therefore, the rise in maize and extensive beef farm bankruptcies in South Africa 
during 1970 to 1994 could have been a necessary fmancial adjuStment. 

NOTE 

This research was conducted in the Agricultural Policy Research Unit, University 
of Natal, South Africa, which is finaoced by the Centre for Science Development 
(CSD). Financial suppon from the CSD is gratefully acknowledged. Opinions 
expressed are, however, those of the authors. The authors thank colleagues at the 
University of Natal for constructive criticism on earlier drafts of the paper. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



70 SAJEMS NS Vol I (1998) No 1 

REFERENCES 

1. BARRY, P.J., ELLINGER, P.N., HOPKIN,l.A. & BAKER, C.B. 
(1995). Financial Management in Agriculture. 5th ed., Danville, Illinois, 
Interstate Publishers. 

2. CCWR (1996). Computing Centre for Water Research, University of Natal. 
Pietermaritzburg. Personal communication. 

3. CENTRAL STATISTICAL SERVICE (1995). CSS Bulletin of Statistics. 
29(30). Central Statistical Service, Pretoria. 

4. CENTRAL STATISTICAL SERVICE (1992). Agriculture Sun'ey, Central 
Statistical Service, Pretoria. 

5. CENTRAL STATISTICAL SERVICE (1972; 1994). South African 
Statistics, Central Statistical Service, Pretoria. 

6. CENTRAL STATISTICAL SERVICE (1988a). Statistics on a Regional 
Basis, Development Regions, C (Orange Free State), Census of Agriculture 
1988, Central Statistical Service. Pretoria. 

7. . CENTRAL STATISTICAL SERVICE (1988b). Statistics on a Regional 
Basis, Development Regions, F, G, H and J (Transvaal), Census of 
Agriculture 1988, Central Statistical Service, Pretoria. 

8. CHAN, M.W.L. & ROTENBERG, W.D. (1988). Financial Distress in the 
Canadian Agricultural Sector: A Macro Analysis, Canadian Journal of 
Agricultural Economics, 36,531-538. 

9. CHATTERJEE, S. & PRICE, B. (1977). Regression Analysis by Example. 
New York, lohn Wiley and Sons. 

10. DA VIES, A.S. (1996). Insolvency in AgricuJrure: Bad Managers or the 
Common Agricultural Policy? Applied Economics, 28,185-193. 

11. DE lAGER, F. & SWANEPOEL, V. (1994). Factors Associated with 
Farm Financial Failure in the Northern Springbok Flats, Agrekon, 
33(4),242-247. 

12. DEVAOOSS, S. (1985). The Impacts of Monetary Policies on US 
Agriculture. Unpublished PhD thesis, Ames, Iowa. Iowa State University. 

13. DIRECTORATE AGRICULTURAL ECONOMIC TRENDS (1995). 
Abstract of Agricultural Statistics. Government Printer, Pretoria. 

14. DIRECTORATE AGRICULTURAL STATISTICS (1996). Abstract of 
AgriCultural Statistics. Government Printer, Pretoria. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



SATES NR Vol 1 (1998) Nr 1 71 

15. DIRECTORATE FINANCIAL ASSISTANCE (1996). Financial Assis-
tance Schemes of the Depanment of Agriculture. Directorate Financial 
AssistaJ:x.:e, Pretoria. 

16. DORAN, H.E. (1989). Applied Regression Analysis in Econometrics. New 
York, Marcel Dekker, Inc. 

17. DRABENSTOTT, M. (1983). Financial Crisis in Agriculture: Discussion. 
American Joumal of Agricultural Economics, 65(5),955-957. 

18. FAMINOW, M.D. & LAUBSCHER,l.M. (1991). Empirical Testing of 
Alternative Price Spread Models in the South African Maize Market, 
Agricultural Economics, 6,49-66. 

19. GABRIEL, S.C. & BAKER, c.B. (1980). Concepts of Business and 
Financial Risk, American Journal of Agricultural Economics, 62(3),560-
564. 

20. GENSTAT (1993). Genstat 5. Release 3.1. Lawes Agricultural Trust, 
Rotharnsted Experimental Station, United Kingdom. 

21. GU1ARATI, D.N. (1995). Basic Econometrics, 3rd ed. New York, 
McGraw-Hili. 

22. HUGHES, D.W., RICHARDSON, 1.W. & RISTER, M.E. (1985). Effects 
of Sustained Financial Stress on the Financial Structure and Perfonnance of 
the Farm Sector, American Joumal of Agricultural Economics. 67(5),1116-
1122. 

23. HUMAN, C.J.F .. (1989). The Implications of Fiscal and Monetary Policy 
on Management in South Africa, Agrekon, 28(1),9-13. 

24. LESLIE, 1.R. & DARROCH. M.A.G. (1993). Financial Characteristics of 
Successful and Unsuccessful Farms in Selected Areas of South Africa: 
Management Implications, South African Journal of Agricultural Extension, 
22,6-12. 

25. LOUW, A. (1995). Expanding the Role of Private Financial Institutions in 
the Creation of a More Financially Viable Agricultural Sector. Paper 
presented at AGRI-TECH Conference, August 1995. CSIR Conference 
Centre, Pretoria. 

26. MISHKIN, F.S. (1988). Understanding Real lnterest Rates, American 
Joumal of Agricultural Economics, 70(2),1064-1072. 

27. MURDOCK, S.H & LEISTRITZ, F.L. (1988). The Farm Financial Crisis: 
Socioeconomic Dimensions and Implications for Producers and Rural 
Areas. Boulder, Colorado, Westview Special Studies in Agricultural 
Science and Policy. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.



72 SAJEMS NS Vol I (1998) No 1 

28. NIEUWOUDT, W.L. (1985). Agricultural Policy, the Economic 
Perspective. Inaugural lecture delivered at the University of Natal, 
Pietennaritzburg. 

29. RAUSSER, G.C. (1985). Macroeconomics and US Agricultural Policy, in: 
Gardner B.L. (ed), US AgricuLtural Policy: The 1985 Fann LegisLation. 
Washington, DC, American Enterprise Institute for Public Policy Research, 
207-252. 

30. RAUSSER, G.C. (1988). Stability Issues and Policy Analysis, in: 
Sunmer D.A. (ed) , AgriculturaL Stability and Fann Programs: Concepts. 
Evidence and Implications. Boulder, Colorado, Westview Special Studies 
in Agricultural Science and Policy, 143-170. 

31. RUCKER, R.R. & ALSTON, LJ. (1987). Fann Failures and Govenunent 
Intervention: A Case Study of the 1930s, American Economic Review, 
77(4),724-730. 

32. SHEPARD, L.E. & COLLINS, R.A. (1982). Why Do Farmers Fail? 
Fann Bankruptcies 1910-1978, American JournaL of AgricuLturaL 
Economics, 64(4),609-615. 

33. SOUTH AFRICAN RESERVE BANK, (various years). SOUlh African 
Reserve Bank QuanerLy BulLetin of Statistics, Pretoria. 

34. STANDARD BANK (1994). AgriReview, Agricultural Department, 
Commercial Services Division, Standard Bank of South Africa, January 
1994. 

35. TOMEK, W.G. & ROBINSON, K.L. (1991). AgricuLtural Product Prices. 
3rd ed, Ithaca, Cornell University Press. 

36. TWEEfEN, L.G. (1985). Farm Finaocial Stress, Structure of Agriculture 
and Public Policy, in: Gardner B.L. (ed), U.S. Agricultural Policy: The 
1985 Farm LegisLation. Washington, DC, American Enterprise Institute for 
Public Policy Research, 83-112. 

37. VAN NIEKERK, T. (1995). NAMPO, BothavilIe, South Africa. PersonaL 
communication. 

38. VAN ZYL, J., VAN DER VYVER, A. & MOSTERT, C.W. (1987). The 
Effect of Debt Burc:!n, Interest Rates and Inflation on the Survival of 
Fanning Enterprises: A Case Study in the Western Transvaal and the 
North-Western Bushvdd, Agrekon, 26(1),1-7. 

39. VINK N. (1993). Enr-epreneurs and the Political Economy of Reform in 
South African AgriculLure, Agrekon, 32(4),153-166. 

40. VOLKSKAS BANK (1988). Economic Spotlight, March 1988. 

R
ep

ro
du

ce
d 

by
 S

ab
in

et
 G

at
ew

ay
 u

nd
er

 li
ce

nc
e 

gr
an

te
d 

by
 th

e 
P

ub
lis

he
r 

(d
at

ed
 2

00
9)

.