• No se han encontrado resultados

4. CAPITULO IV ESTUDIO DE MERCADO Y LA COMPETENCIA

4.4. ANALISIS DE LA DEMANDA

4.4.4. DEMANDA DE ALUMINIO EN AREQUIPA

The first instrument to be used is ethnic affinity, measured as the proportion of a district’s population that are co-ethnic with the sitting President. Malawi people are of Bantu origin and comprise many different ethnic groups; Malawi Human Rights Commission (2005) approximates that there are about 15 ethnic groups in Malawi. Figure (4.4.1) is a map of Malawi showing locations of the major ethnic groups.As the map reveals, the Chewa people are the largest ethnic group making up to 38.4% of Malawi’s population and are mainly found in the Central Region of Malawi, the Lomwe make up about 17.6% and are found in the Southern Region, the Yao are about 13.5% and are mainly in the South Eastern part of Malawi, the Ngoni make up 11.5% found in the Central Region and the Tumbuka are about 8.8% and cover much of the Northern Region.

42Particularly infrastructure such as rural roads, boreholes, education infrastructure which are easily used as rents.

Figure 4.4.1: Spatial distribution of ethnic groups in Malawi

Source: Figure from Robinson (2016)

Ascension to presidency in Malawi is through public elections held once every 5 years in a first-past-the-post system. Political representatives in the presidential election can come from any district and indeed from any ethnic group. Despite the Chewa being a majority, over the study period the sitting President has been either Lomwe or Yao, as shown in figure (4.2.4).

The relevance condition requires that the instrument be a strong enough predictor of the endogenous regressor, in our case aid. There is already overwhelming evidence that indicates that disproportionate amounts of aid are allocated to incumbent president’s district of birth or regions of their ancestral origins especially in Sub Saharan Africa. Franck and Rainer (2012) use data from 18 African countries for over 50 years and find significant evidence of large and widespread ethnic favoritism in allocation of aid resources. As an example of this in Malawi figure (4.4.2) shows aid allocation in Malawi between 2002 and 2004 and between 2005 and 2007, thus under two President’s of different ethnic

origins.

Figure 4.4.2: Allocation of aid in the periods 2002-2004 and 2005-2005

Dr. Bakili Muluzi of Yao origin was reelected as President for Malawi in 1999 and was in power up to 2004. As the figure shows, the Yao districts of Machinga (his birth district), Mangochi and Balaka are allocated disproportionately high amounts of aid than any of the other districts. Between 2005 and 2007, when Dr Bingu wa Muntharika of Lomwe origin was President, figure 4.4.2 shows that the Lomwe districts of Thyolo, Mulanje and Phalombe received more aid than other districts.

A concern, recognised in much of the literature43 on African political studies, is that

African political behavior (particularly for Sub Saharan Africa) is subsumed in ethnicity. This implies that often voters will vote for presidential candidates that are co-ethnics with the expectation of favourable outcomes in terms of outputs from the public decision- making process, which in the case of Malawi imply disproportionate allocation of state resources. If it is the case that ethnic clientelism is indeed prevalent in Malawi’s voter

43There is a significantly large literature indicating that ethnic clientelism is prevalent in African politics including Chabal (2009), Mozzafar et al. (2003), van de Welle (2003), Francois et al. (2015), Posner (2005a), Lindberg (2003).

behavior (i.e. districts supported Presidential candidates primarily along ethnic lines), then a president’s ethnicity ceases to be random – a district’s vote is for the candidate that will send aid their way. If the poorest districts are the ones that most vote on ethnic basis, the instrument may not be exogenous.

Despite this strongly held view of ethnicity in African electoral behavior, there are few independent surveys of voter behavior that can provide evidence to support the rationale that ethnicity is a primary factor in voter’s decision making. Recent surveys on electoral results in African democracies that analyze voter behavior conclude that ethnicity itself is not a primary factor that influences voters decision. Anyangwe (2012) surveys national and provincial elections held in South Africa from 1999 to 2009 and Municipal elections held in 2011 and conclude that ethnic identity only has a marginal effect on South African voters. Carlson (2015) conduct an experiment designed to determine how voters in Uganda make their electoral choices and find that co-ethnicity on its own does not increase support

for a candidate.44

There is growing evidence that suggest that voters are less influenced by ethnicity of individual candidates, rather they are driven by a number of different factors. Centre for Democratic Development (1999) found a high level of partisan identification rather than individual candidate ethnicity (they vote for representatives of political parties that have a stronghold in that district or region regardless of whether the candidate has ethnic origins from the same district or region). Ratsimbaharison and Marcus (2005) presents evidence from Madagascar and show that evaluative decision making which includes rational choice

and prospective evaluation of political parties as the key vote determinant45.

For Malawi, the study analyses election results from four general elections held in Malawi in 1999, 2004 and 2009 by conducting a simple regression analysis of district vote shares received by the winning Presidential candidate. We regress the winning candidate’s dis- trict vote share on, among other determinants, the proportion of the winning candidates co-ethnics in a district and party identification, a dummy variable that takes the value 1

44Lindberg and Morrison (2008) survey voters in 2 elections in Ghana and conclude that ‘an over- whelming majority of voters do not vote based on clientelism or ethnic ties but cast their votes after careful evaluation of candidates.’

45Other studies with evidence of partisan voting and evaluative decision making include Piombo (2005) with evidence from South Africa while Posner (2003) and Lindberg (2003) both give evidence from Zambia.

if the winner’s political party is a dominant party46 in that district, or 0 otherwise. As the results reported in table 4.4.1 reveal, ethnicity does not seem to affect the vote share that a candidate gets in the district, being found to be statistically insignificant, in the equation for district vote shares.

Table 4.4.1: Results for OLS regression of district vote share

1 2 3 4 5 6 7

0LS 0LS 0LS 0LS 0LS 0LS 0LS Winner’s birth district 0.1230 -0.0014 -0.0061 0.0306 0.0288 -0.0234 -0.0153

(0.1579) (0.1860) (0.1803) (0.1004) (0.1078) (0.1061) (0.1116) Population -0.9644 -1.3619 -1.2799 -0.9059 -0.5950 -1.0814 -0.7134 (2.1541) (1.9110) (2.2686) (0.8304) (0.8840) (0.7984) (0.8382) Poverty rate -0.0003 0.0392 0.0156 0.0038 0.0126 0.0212 0.0210 (0.1292) (0.1197) (0.1427) (0.0574) (0.0595) (0.0602) (0.0603) Northern region 0.0921 0.1035* 0.1008 0.0436** 0.0357 0.0490** 0.0395* (0.0636) (0.0558) (0.0659) (0.0208) (0.0247) (0.0190) (0.0227) Central region 0.0403 0.0526 0.0509 0.0255 0.0184 0.0311 0.0226 (0.0692) (0.0610) (0.0675) (0.0246) (0.0249) (0.0233) (0.0234) Southern region 0.0402 0.0526 0.0508 0.0254 0.0183 0.0310 0.0226 (0.0691) (0.0610) (0.0674) (0.0246) (0.0249) (0.0233) (0.0234) Urban districts -0.0032 0.0003 0.0003 0.0134* 0.0133** 0.0148** 0.0142** (0.0143) (0.0137) (0.0125) (0.0068) (0.0063) (0.0070) (0.0065) City -0.0001 -0.0000 -0.0000 -0.0001** -0.0001* -0.0001* -0.0001* (0.0001) (0.0001) (0.0001) (0.0000) (0.0000) (0.0000) (0.0000) Chewa 0.0003 -0.0074 -0.0076 (0.0173) (0.0101) (0.0100) Yao -0.0001* 0.0000 0.0000 (0.0001) (0.0000) (0.0000) Lomwe -0.0091 0.0047 0.0021 (0.0184) (0.0068) (0.0076) Winner’s ethnic population (% district population) 0.2180 0.2346 0.0959 0.0818

(0.1887) (0.1958) (0.0731) (0.0824) Party identification 0.4771*** 0.4824*** 0.4729*** 0.4781*** (0.0418) (0.0439) (0.0404) (0.0428) Observations 360 360 360 360 360 360 360 R-squared 0.2606 0.2800 0.2845 0.8075 0.8124 0.8112 0.8150 District - Year FE Y Y Y Y Y Y Y Number of id 24 24 24 24 24 24 24

Notes: The table presents results of fixed effects panel regression on the share of votes that a winning candidate received during a general election (held in 1999, 2004 and 2009) from each district on ethnicity, measured as the proportion of population that is co-ethnic with the winning candidate; and party identification, a dummy variable that takes the value 1 if the winner’s party is dominant in the district, and 0 otherwise. Robust standard errors in parentheses:

∗p <0.10;∗ ∗p <0.05;∗ ∗ ∗p <0.01.

Furthermore, the R-squared for specification with only ethnicity in the regression are between 0.26 and 0.28; when we include party identification the R-squared increases significantly and to between 0.81 and 0.82, implying that the model that includes party identification is a better fit to explain distribution of share of votes. These findings suggest that ethnicity is not the primary factor in voter’s behavior in Malawi, rather party identification seems to play a more influential role.

As an example, table (4.4.2) shows vote shares received by Dr. Bingu wa Muntharika (from Lomwe ethnic group) who contested in all the elections that took place during

46Dominant party is determined by the proportion of parliamentary seats in the district won by the winning Presidential candidate’s party. A party that is dominant in a district wins more parliamentary seats than one that is not.

the study period (in 1999, 2004 and 2009), and represented a different political party in each election. In 1999, he represented the United Party (UP) and had less than 1% vote

share even in his birth district where United Democratic Front (UDF47) candidates won

both elections getting 79% in the Presidential election. In 2004, he represented the UDF and managed higher vote share in Yao districts (80%) than the average in his co-ethnic districts including his birth district. In 2009 he represented the Democratic Progressive

Party (DPP48) and got significantly high vote shares from Lomwe (87%) and Tumbuka

districts (95%) which identified with the DPP but could only manage 26% from Yao districts since he no longer represented the UDF which the Yao identify themselves with.

Table 4.4.2: District vote shares received by Dr Bingu wa Muntharika Election Year Political Party Vote share (birth district)

Avg vote share (co-ethnic districts)

Avg vote share (Chewa districts)

Avg vote share (Yao districts)

Avg vote share (Tumbuka districts) 1999 UP 0.7% 0.9/% 0.3% 0.4% 0.3%

2004 UDF 61% 51.2% 17% 80% 20%

2009 DPP 91% 87% 42% 26% 95%

Notes: UP stands for United Party, UDF is United Democratic Party and DPP is Democratic Progressive Party

The explanation for these electoral outcomes is that between 1999 and 2005, the UDF was the dominant party so that despite being predominantly a Yao (and Islamic religion) party, many other ethnic groups identified themselves with the UDF and hence would vote for UDF representatives at the expense of co-ethnics representing other political parties. The formation of DPP in 2005 shifted political powers as it became the dominant political party, and hence its candidates received the majority vote shares in 2009.

Consistent with the evidence that ethnicity is not the primary factor in voting behavior is the observation that since Malawi become a democracy, no candidate from the largest ethnic group (the Chewa) has ever won in an election. The Chewa (38.4% of population) comprise over twice as much as the population of the next biggest ethnic group (the Lomwe with 17.6%) and are also the ethnic group who are found in most other parts of Malawi other than their region of stronghold. If indeed ethnicity was a key factor in who gets elected, as believed, then Chewa candidates would have a head start just by the sheer advantage they hold population-wise over other races.

47The UDF party has its political foundations built along Yao ethnic group and Islamic religion. Thyolo district population and Lomwe people in general are Christians by religion, so religion could not have influenced vote outcomes in Thyolo.

48The DPP was formed by Dr Muntharika with its political foundations in a coalition between the Lomwe ethnic group (average 91% vote share) and Tumbuka people from Northern Malawi (averege 95% vote share)

This evidence is significant; if ethnicity itself is not a primary factor in the voting behav- ior in Malawi, then an ethnic group cannot form expectations as regards to the future development of their district based solely on the ethnicity of the presidential candidate they voted for since the winner may not necessarily be their co-ethnic. Since the evidence suggests that a winning candidate does not ascend to power on the back of ethnicity, then if an incumbent president directs disproportionate amounts of aid to his district of birth or region where his co-ethnics are located geographically, it is unlikely that it is in return for any political favors that he received, rather it is merely because of his desire to favor co-ethnics in aid allocation.

Finally, it is important to consider if there are any other channels through which certain aspects of ethnicity would affect the level of economic activity at the district level but are not captured within the framework of the instrument. For instance, it may be possible that certain cultural practices for a particular ethnic group are consistent with higher economic activities than other races. An example can be ethnics who require payment of dowry/bride price such as the Tumbuka and for the Ngoni tribes, as such for one to marry they (or their relatives) must be engaged in some form of economic activity that enables them to raise the dowry requirement. To eliminate any bias that may arise due to the possibility of any other channel from ethnicity to changes in the level of economic activity other than through the instrument, in the robustness checks the study also includes dummy variables for the major ethnic groups in Malawi that account for more than 85% of the population.