Primary agricultural commodity exports and unemployment

African Economic Conference, October 28-30, 2013

JOHANNESBURG, SOUTH AFRICA.

Alassane DRABOFERDI and University of Ouagadougou

Motivations and objectives 1. Growing debate in the relationship between trade

openness and labour market

2. But the role played by different components of trade remains less investigated

3. African countries are generally characterized by the large share of raw commodities in their exports

44 Objectives: 1) Assess empirically the impact of primary agricultural commodity exports on unemployment and employment, and 2) highlight the diversification role of regional integration in Africa.

Outline

1. Relationship between Trade and unemployment

2. Primary commodities transformation and unemployment

3. Econometric specification and estimation

4. The endogeneity bias

5. Data and variables

6. Econometric results

7. Conclusion

1. Relationship between Trade and

unemployment (Theoretical)

Two categories emerge from the existing theoretical works linking trade and unemployment regarding their foundations

The models built on comparative advantage and trade specialization due to differences in technologies, factor endowment or labor market structure (Davidson et al., 1999; Dutt et al. 2009; Moore & Ranjan, 2005)

The merging of the intra-industry trade models and equilibrium unemployment models (Chao & Yu, 1997; Driffill & van der Ploeg, 1995; Felbermayr et al., 2011; Helpman & Itskhoki, 2010; Janiak, 2007; Matusz, 1996)

The results obtained are far from consensus even within categories

1. Relationship between Trade and unemployment (Empirical)

the first works on the issue use factor composition technique to assess the employment content of trade (Driver et al., 1988; Baldwin, 1994; Wood, 1991, 1995)

Recent empirical works linking trade and unemployment generally use econometric techniques

country industry level dataset, [Feliciano (2001) and Revenga (1997) for Mexico, Greenaway et al. (1999) for the United Kingdom, Davidson & Matusz (2005) for the United States, Lang (1998) for New Zealand, Jenkins (2004) for Vietnam].

country or cross-country studies [(Dutt et al., 2009; Kim, 2010)]

They generally find ambiguous results

2. Primary commodities transformation and unemployment

At least three arguments can be put forward to support the positive association between primary commodity exports and unemployment :

the transformation process of the raw products itself can be considered as additional activities in the economy

the production of raw commodity as well as other products may be expanded

the processing may increase private and public resources, and thus encourage investment in human capital, easing its integration into job market.

3. Econometric specification and estimation

•Our econometric model of interest is:

Where • Unemployi is the logarithmic form of unemployment

rate of country i. • Agrprimcom represents the logarithmic form of

agricultural primary commodity export as percentage of total commodity export whereas

• X is the matrix of control variables commonly used in the literature.

• The coefficient of interest is expected to be positive ( >0).

4. The endogeneity bias

The estimation of this equation by the ordinary Least Squares method (OLS) suffers from at least two potential problems (omission variables and reverse causality bias) as recognized by Dutt et al. (2009).

We thus instrumental variables approaches (2SLS and GMM) for which we consider two instruments:

1. The primary commodity export variable lagged

2. the logarithmic form of the percentage of arable land

Countries with more arable land are likely to produce more agricultural products and export them, and direct effect of arable land availability on unemployment (not channeled by agricultural employment) is not plausible. The quality of these instruments is tested by accurate statistics.

5. Data and variables

Variables Mean Min Max Stand. Dev. Sources

Unemployment rate 9.13 2.8 22.55 4.43 WDI

Employment rate 53.33 39.68 74.91 6.91 WDI

Agricultural PrimaryCommodity 19.43 1.05 77.01 17.0647 COMTRADE

Agricultural land 43.38 3.16 84.97 21.21 WDI

Labour Union 0.48 0.16 0.82 0.19 Botero et al. (2004)

Employment Law 0.45 0 0.71 0.2 Botero et al. (2004)

Working Age Population 20.92 18.73 25.11 1.35 WDI

Log GDP 19.07 16.62 22.71 1.39 WDI

Labour Force Participation 68.82 49.05 83.68 6.63 WDI

Black MarketPrenium 1.76 0 42 6.18 Fraser Institute

Agricultural Employment 6.96 0.29 40.28 7.37 WDI

PrimarySchoolEnrollment 93.5 63.27 99.89 6.36 WDI

6. Econometric results (1)Effect of Agricultural Primary Commodity Export on Employment and Unemployment

(1) (3) (2) (4) Log Unemployment

Rate Log Primary Commodity

Log Unemployment Rate

Log Employment Rate

(OLS) (First stage) (2SLS) (2SLS) Log Primary Commodity 0.203** 0.310*** -0.041*** (2.49) (4.41) (3.02) Labor Union 0.440 0.453* 0.136 -0.104 (1.20) (1.85) (0.36) (1.40) Employment Law 0.231 -0.259 0.416 -0.088 (0.59) (1.22) (1.17) (1.18) Log Working Population 0.310 4.169* -2.920 1.976*** (0.11) (1.70) (1.41) (4.19) Log GDP -0.116 -0.242* 0.080 -0.091*** (0.80) (1.83) (0.76) (3.84) Labor Participation -0.015* 0.006 -0.025*** 0.017*** (1.86) (1.08) (4.35) (12.69) Black Market Prenium 0.008 0.004 0.015** -0.001* (1.20) (0.86) (2.45) (1.66) Log Agricultural Employ -0.087 -0.094* -0.119** 0.020* (1.39) (1.83) (2.33) (1.82) Primary School Enrol -0.006 0.010 -0.006 0.002 (0.58) (1.33) (0.72) (1.51) Log Agricultural Land 0.099* (1.94) Log Primary Commodity Lag 0.851*** (14.49) Constant 4.196 -8.800 10.756** -1.421 (0.67) (1.62) (2.24) (1.33) Observation 58 48 48 48 R² 0.397 0.919 0.448 0.759 Shea R² 0.851 Fisher Stat P-Value 0.000 Hansen OID p-value 0.350 0.150

6. Econometric results (2)Effect of Agricultural Primary Commodity Export on different type of Unemployment

(1) (2) (3) (4) (5) Log Unemp

IFS Log Long Term

Unemp Log Youth

Unemp Log Male Unemp

Log Female Unemp

Log Primary Commodity 0.342*** 0.499** 0.367*** 0.264*** 0.395*** (4.67) (2.01) (3.45) (3.24) (4.39) Labor Union 0.692* -0.312 0.067 0.020 0.403 (1.79) (0.33) (0.15) (0.06) (0.88) Employment Law 0.187 1.663 0.357 0.432 0.471 (0.61) (1.54) (0.88) (1.19) (1.09) Log Working Population -4.799** -22.267 -3.060 -3.473 -4.189 (2.19) (1.27) (0.75) (1.25) (1.25) Log GDP 0.227** 1.032 0.124 0.127 0.123 (2.10) (1.25) (0.58) (0.97) (0.73) Labor Participation -0.025*** -0.032 -0.031*** -0.020*** -0.041*** (3.43) (1.10) (3.73) (2.65) (4.74) Black Market Prenium 0.013** 0.206* 0.021*** 0.013* 0.017*** (2.43) (1.73) (2.65) (1.89) (2.74) Log Agricultural Employ -0.081* -0.133 -0.009 -0.116** -0.127** (1.65) (0.34) (0.12) (2.13) (1.97) Primary School Enrol -0.024*** -0.013 0.004 -0.008 -0.009 (2.83) (0.30) (0.39) (1.00) (0.88) Constant 15.015*** 52.539 10.245 11.475* 14.923* (2.96) (1.37) (1.14) (1.74) (1.92) Observation 43 30 46 47 47 R² 0.517 0.232 0.282 0.305 0.459 Shea 0.846 0.923 0.843 0.850 0.850 Fisher Stat P-Value 0.000 0.000 0.000 0.000 0.000 Hansen OID p-value 0.775 0.113 0.867 0.587 0.640

6. Econometric results (3)Effect of Agricultural Primary Commodity Export on Employment and Unemployment (GMM panel data estimation).

(1) (2) (3) Uuemployment

Rate Youth

Unemployment Employment

Rate Log Uuemployment lag 0.578*** (4.45) Log Youth Uuemployment lag 0.514* (1.73) Log Employment lag 0.047 (0.20) Log Primary Commodity 0.279** 0.349* -0.078* (2.02) (1.74) (1.80) Log Working Population 1.517 1.476 0.082 (1.63) (1.28) (0.23) Log GDP -0.035 0.045 0.017 (0.48) (1.09) (1.27) Labor Participation -0.011 -0.014 -0.003 (1.41) (0.69) (0.81) Log Agricultural Employ -0.044 -0.054 0.018 (1.54) (0.73) (1.44) Primary School Enrol -0.009 -0.008 0.004** (1.41) (0.93) (2.29) Constant -3.916 -5.299 3.550** (0.97) (1.00) (2.03) Time dummies yes yes yes Observations 269 226 210 Countries 102 90 78 AR1 0.051 0.089 0.664 AR2 0.127 0.127 0.167 Hansen OID P-value 0.676 0.770 0.140 Group/Instrument 102/14 90/15 78/21

7. Conclusion (1)This paper extends the empirical side of the relation between trade openness on labour market outcomes by arguing that the effect depends on the composition of trade, and focusing on the role played by agricultural primary commodity exports..

It is found that high share of primary commodity in the exports is associated with high unemployment rates and low employment rates. The effect is particularly large for long term unemployment

The commodity-based industrialization should be promoted to reduce the high and challenging young unemployment rate. As recognized by the Istanbul Programme of Actions, poor countries should “adopt and strengthen, as appropriate, sector and commodity-specific policies, measures and strategies to enhance productivity and vertical diversification, ensure value-addition and increase value-retention” (United Nations, 2011, paragr. 66b).

This can be possible through the transformation of raw products before exporting them. In addition to the creation of value addition, this will result in low unemployment rate.

7. Conclusion (2)

• One way to solve for this is to abandon the short term views, and target long term policies in the education and financing systems. Education should target long term development needs through appropriate technology acquisition, research and development, and improvement and implementation of traditional existing knowledge.

• Another important issue concerns the difficulties faced by Africans to finance their initiatives. Banking systems should be reformed in order to ease the access to credit, and the development of the private sector.

• By providing economies of scale and larger market, regional integration is undoubtedly a solution to boost diversification in African countries.

THANK YOU FOR YOUR ATTENTION!