HOUSING FINANCE and INCLUSIVE GROWTH in Africa Benchmarking, Determinants and Effects 1

Transcription

1 HOUSING FINANCE and INCLUSIVE GROWTH in Africa Benchmarking, Determinants and Effects 1 Christian-Lambert Nguena 2 The University of Yaounde II and the World Bank Fulbert Tchana Tchana 3 The World Bank Albert G. Zeufack 4 The World Bank ARTICLE SUBMITTED FOR African Economic Conference (AEC) 2015 on Addressing Poverty and Inequality in the Post 2015 Development Agenda Kinshasa, Democratic Republic of Congo, November 2-4, 2015 Abstract Using a panel database of 48 Sub-Saharan African countries from 2000 to 2013, this paper analyses the structure of housing finance in Africa, its determinants, and its impact on inclusive growth. Our empirical investigation finds that market capitalization and urbanization are key positive determinants of housing finance while a post-conflict environment is conductive for greater housing finance development. This result suggests that housing finance is driven by demand and supply as any standard market. Besides, we find that housing finance development in Africa is not yet an effective tool for inequality reduction, given that it remains at a very early stage. However, we show that above a given threshold, housing finance could be efficient for inequality reduction. Finally, housing finance is loosely positively related to greater economic development in Africa. All these findings suggest that policies to boost housing finance development in Africa should be viewed as measures that would provide benefits in the medium to long terms. KEYWORDS: Housing Finance, Sub-Saharan Africa, inclusive growth, Shared prosperity. JEL: G21, R1, O4. 1 The views expressed in this paper are not necessary those of the World Bank Group. The authors thank Simon Walley for sharing its database. All errors and omissions are the authors sole responsibilities. 2 Research Fellow, Research in Applied Micro and Macroeconomics (REMA), clanguena@yahoo.fr. & President, African Association of Young Economists (AAYE), nguena@aaye.org. 3 Senior economist, The World Bank, ftchanatchana@worldbank.org 4 Practice manager, The World Bank, azeufack@worldbank.org

2 I. Introduction Sub-Saharan African countries (Africa) have experienced stable economic growth averaging 5 percent over the past decade. However, growth has not been inclusive and poverty as well as unemployment have remained very high. The challenge of inclusive growth is made even more daunting by unmanaged urbanization that has led to millions leaving in sub-standard housing. Africa has the highest growth rate of urbanization but the least developed housing finance in the world (see World Bank, 2014). UN Habitat (2008) reports that 46 African cities are now larger than one million people and that every day for the coming fifteen years, Africa s cities will have to accommodate an extra 40,000 people. Over the next 25 years, more and more people will be added to the number of urban dwellers in Africa. This implies a growth in the demand for housing that African countries will face and the need to address this by developing a housing finance system. This new challenge is emerging in a context of already widespread poverty and inequality in cities, with a lot of people living in slums without adequate basic services. However, the low capacity of the construction industry and the absence of a strong housing finance sector are likely to exacerbate tensions on the housing market, leading to high rental rate and further widening income inequality. Despite a positive trend, the share of mortgage to GDP remains extremely low in Africa (3 percent) compared to 70 percent in developed countries. Africa s mortgage markets are nascent and small by international comparisons. However, according to data from Badev et al. (2013), the average proportion of population who can afford the minimum income required for a prudent mortgage is 16 percent in Africa. So far, efforts of African governments, international partners, and financial institutions have barely reduced the gap between housing demand and housing supply in Africa urban areas (with housing demand higher that supply). As in other parts of the world, this gap is even more important for low and moderate income households. In fact, currently most low and moderate income household in Africa only have access to the unregulated informal housing sector. With the exception of South Africa, formal sector housing programs (public and/or private) have mostly targeted middle and upper income households in Africa. Because of its importance, housing finance has generated at least three main strands in economic literature. The first strand focusing on its determinants (see Badev et al., 2014; Egert and Mihaljek, 2007; Haibin Zhu, 2006; and Buckley and Madhusudhan, 1984); the second on the channels with which it affects inclusive growth (see Buckley, 1996 and Dubel, 2007); and the third on its impact on growth and shared prosperity (see Hongyu et al., 2002 and Gutierrez et al., 2007). This literature has answered numerous questions of interest on housing finance determinants but so far has not focused on specific characteristics of SSA housing finance, such as its very low level of development. The literature still has to provide a comprehensive analysis of the linkage between housing finance and shared prosperity especially in Africa. Building on existing literature and to fill the gap, this paper answers the following questions: What is the structure and typology of housing finance in Africa? What are the determinants of housing finance development in Africa? What is the relationship between housing finance, growth and inequality in Africa? Data investigation suggests that the housing finance sector is very weak in Africa. The ratio of mortgage to GDP is less than 10 percent for almost all countries but South Africa and Namibia. Housing penetration is also very low with no country posting more than 9 percent penetration rate and three out of four countries with a rate lower than 5 percent. Also, the mortgage rate, the legal system and country s GDP per capita are correlated with housing finance in Africa. Countries with higher GDP per capita and low 2

3 mortgage interest rates enjoy greater housing finance. In turn, greater access to Africa housing finance is positively related to GDP growth, inequality reduction, and human development indicators Our econometric investigation provides important findings on each of our questions of interest. Regarding the determinants, stock market capitalization and urban population growth rates are the strongest positive determinants, while recovery from conflict is another determinant albeit less strong. In addition, some variables seem to be hampering housing finance depending on the economic environment. These variables are: the credit to the economy, which is a proxy of the development of the banking sector and trade openness which may masque the low housing finance in oil exporting countries. In addition, housing finance impacts positively on inequality reduction and inclusive growth. Firstly, there is a threshold effect in the impact of housing finance depth on inequality; indeed a higher value of housing finance depth is negatively and significantly correlated to inequality while a very low and very high value presents a non-significant result. Secondly, our results suggest that an increase of 1 point of percentage (pp) of housing finance depth could lead to an increase of 0.37 percent of GDP per capita growth. This result confirms Hongyu et al. (2002), Erbas and Nofthaft (2002), Uy (2006) and Freire et al. (2006) findings in other regions. The remainder of this paper is organized as follows: Section 2 provides a literature review; section 3 presents the characteristics of housing finance policy in Africa and investigates the link between housing finance and shared prosperity; section 4 provides an econometric analysis using a panel dataset we ve built and draw some policy implications; and section 6 concludes the paper. II. Housing finance description, determinants, and inclusive growth effects: an empirical literature review The literature related to housing finance determinants and impact on inclusive growth is growing. The literature has been mostly empirical and has focused less on specific issues facing the housing market in Africa. In this section, we present the literature in three strands. The first rand strand reviews papers on housing finance determinants, the second focuses on the channels used by housing finance to impact growth and inequality and the third strand presents the impact of housing finance on growth and inequality reduction. In each of these strands, we distinguish between theoretical and empirical findings as well as SSA findings versus global findings Determinants of housing finance The literature on the determinants of housing finance so far has been empirical and has not focus on specific determinants of housing finance in SSA. This literature highlights three main determinants, namely, the level of development proxied by GDP per capita, the stock market development as well as informal finance. Most importantly it shows that government support or subsidy is not one of them. Therefore, a specific study for Africa may be needed where GDP per capita is actually very low and would certainly not make a difference. Badev et al. (2014) using a new set of data on the depth and penetration of mortgage markets across countries in the world found that: i) mortgage markets seem to develop only at relatively high levels of GDP per capita; ii) policies associated with financial system development (such as price stability, the efficiency of contractual and information frameworks, etc.) are also associated with mortgage market development, including; iii) well-functioning insurance market and better capitalization of stock market are strongly associated with mortgage market development. Moreover, Buckley and Madhusudhan (1984) 3

4 tested a model of the relationship between housing investment and GDP, anticipated inflation, changes in inflation and the extent of capital deepening across several developing and transition countries. They found that, holding all else constant countries with deeper financial markets invest relatively more in housing. Also, Haibin and Zhu, (2006), by analyzing the structure of housing finance markets and house prices in selected Asian countries, have found evidence that in economies with more flexible housing finance markets, house prices are more responsive to overall changes in market conditions, particularly equity price movements. The main explanatory variables used were GDP, bank credit, equity prices, short-term interest rates, consumer price index and exchange rate. In addition, Besley et al. (1992) have shown that informal finance matter in long term finance by showing that the allocative performance of Rotating Savings and Credit Associations (ROSCAS) is very high. The result highlights the need to verify and check if it can be used as an alternative in the context of Africa where this type of financial institution is very common. The fact that the formal housing finance system is limited in SSA countries reinforces the necessity to investigate the alternative of informal housing finance system. Finally, many studies have shown that government involvement is not one of the determinants. For example Badev et al. (2014) show that neither government subsidies, nor government support of housing financing through State bank dedicated to housing finance has been proven to spur this sector. In fact, many developing countries have had such banks for many decades, but their housing finance sector remains tiny Channels used by housing finance to impact growth and inequality In the literature, housing finance impacts economic growth by reducing the cost of capital, increasing savings, increasing tax revenues, increasing investment on education, reducing vulnerability, and increasing financial deepening. By reducing the cost of capital, housing finance can spur economic growth. In fact, Dubel (2007) provides a model analyzing the relationship between housing finance and housing affordability and show that a well-functioning housing finance reduces loan interest rates, which in turn results in greater affordability of housing with a potential positive impact on economic growth. By increasing savings housing finance can spur growth. According to Buckley (1996), many reasons explain why improving housing finance may lead to increased savings in the economy. First, the return to housing will likely provide positive returns. Second, housing provides the most secure collateral against market fluctuations and a positive yield over the long-run. Third, housing prices are less volatile than other asset prices. Fourth, the availability of housing improves labor mobility and therefore employment potential. Finally, the availability of affordable housing finance may lead to increased savings as potential homeowners save to make the required down payment and to maintain their asset. This saving will be the engine of investment and therefore for economic growth and development. By increasing tax revenues housing finance contributes to the development of public infrastructure and ultimately of economic growth and reduction of inequality. Hangen and Northrup (2010) and Econsult (2009) shows that activities related to housing has a positive impact on revenues of states and other local governments in the United States. In the United States as well as in other countries, greater revenues to local government often translate to better infrastructure for the population and ultimately to reduction of inequality. 4

5 2.3. Housing finance, economic growth and reduction of inequality Studies suggest that housing investment has an impact on economic growth and employment. For example, Hongyu et al. (2002) find that compared to non-housing investment, housing investment has a stronger short-run effect on economic growth. They also find that housing investment has a long run impact on economic growth but not on non-housing investment. Chen and Zhu (2008) found that the relationship between housing investment and economic growth in China is different depending on which provinces are analyzed. Moreover, Erbas and Nothaft (2002) find that low income housing has a lower import component in production and also higher labor intensity. This implies that construction of low income housing will lead to greater employment and growth than the construction of middle or high income housing. Tipple (1994) by verifying the links between employment and housing development shows that investment in shelter is very effective for promoting employment, especially among lower-income groups; some of the benefits to the economy tend to be inversely proportional to housing cost meaning that low cost housing is more beneficial to the economy. In addition, to the empirical analysis of the relationship between housing and economic growth, there are some estimates of multiplier effects associated with construction in developing countries. For example, Uy (2006) cites that for every 1 peso spent on housing activities in the Philippines, an additional pesos are contributed to the GDP. Meanwhile, housing finance has the potential to increase investment on education, hence reducing vulnerability of the poor; it could also improve financial deepening and inclusion. This improvement of financial deepening is critical for poverty reduction in SSA. Housing finance could Increase investment on education and reduce vulnerability for the poor. Becker (1975) and Atkinson (1975) studied the link between investment in human capital and wealth distribution. An implication of these models is that income inequality will decrease as access to finance improves and that housing improves homeowner s borrowing capacity. Housing finance could lead to higher investment in human capital. By improving housing affordability, housing finance may improve education opportunities for the poor. Jacoby (1994) finds that lack of access to credit perpetuates poverty in Peru because poor households can t afford to provide their children with appropriate education. Jacoby and Skoufias (1997) find that without access to finance, shocks to income cause poor families to discontinue schooling for children; housing provides an asset that can be used to smooth shocks to income. More inclusive housing financial system can improve poor and low-moderate income household s access to finance their housing. Consistent with this view, Malpezzi (1999) suggests shifting from a housing finance perspective, where special circuits are used to mobilize short-term household deposits for longterm mortgages, to a perspective where housing finance is integrated with broader capital markets. Singh and Huang (2011) analyze data from Africa between 1992 and 2006 and find that financial deepening (as measured in part by credit to the private sector as a percent of GDP) is associated with less poverty and income disparities in African countries and that this is most important in early stages of financial development. Stronger property rights strengthen this relationship. Beck, Demirguc-Kunt, and Levine (2004) examine a broad cross country sample of 58 developing countries and find that financial development (as measured by the ratio of financial intermediation to the private section to GDP) reduces income inequality by disproportionately raising the incomes of the poor. Moreover, Singh and Huang (2011) found that poverty is inversely related to financial deepening. In addition, financial deepening reduces absolute levels of poverty but does not impact income inequality in a significant manner in 5

6 African countries. This suggests that various definitions should be examined to gain further insight into the relationship between housing finance and poverty and to capture the impact on the absolute poor. III. Stylized fact on housing finance in SSA 3.1. Data In order to highlight the stylized facts and to undertake the econometric analysis, we examine a sample of 54 African countries with data from African Development Indicators (ADI), and the Financial Development and Structure Database (FDSD), the Housing Finance Databases of the World Bank. The database summary and description is present in the appendix. The analysis is limited to to ensure more up to date results. Housing market and finance policy data are from the newest Housing Finance Databases of the World Bank Benchmarking the SSA housing market and finance policy: Typology and characteristics Benchmarking the SSA housing market and finance policy along with a comparative approach is essential to have a clear picture at the base of what have been done until now and what should been done. The first sub-section present a general point on all SSA countries and the second sub-section, specific views based on fundamental characteristics such as wealth, legal origins, political stability, regional context, and oil resources. SSA housing market and finance policy: A general point of view Overall, there is a growing banking sector in Africa since the liberalization two decades ago. Liberalization consisted for African government to approve new financial institutions legislation, and institutionalised private banking system, ending in some cases the State monopoly in this sector. SSA countries financial system is growing fast and becoming increasingly integrated into the global financial systems. At the core of the system are banks, followed by pension funds. The regular Financial Sector Assessment Program (FSAP) implemented in those countries generally confirmed that the banking system is well capitalized, liquid, and profitable. However, only few commercial banks clearly offer housing loans in the form of mortgages as a product for their customers. The majority of banks in Africa finance house acquisitions, not from housing loans, specifically but as investment, private and/or standard consumer loans. These latter products generally have high interest rates with short repayment periods. Some African countries have opted to establish specialized single-purpose nonbank mortgage lenders, or monoline lenders, most notably, Kenya, which has housing finance companies; Nigeria, which has primary mortgage institutions; and South Africa, which has financial service providers specializing in mortgage lending. Typically, these institutions have a narrow banking license limiting their activities and, in particular, restricting deposit collection. This means that they are usually reliant on wholesale funding on the liability side of their balance sheets. These institutions are particularly vulnerable during crisis period because their funding costs rose to a much greater extent relative to lenders with a deposit base. The banking and the financial system in most of SSA countries remain underdeveloped compared to other developing regions. In general, the ratio of private sector credit to GDP is less than 20 percent and financial access is lower (There is only 1.0 total branch per adult). The small size of national markets, the low level of incomes and the weak creditor rights and judicial enforcement mechanisms 6

7 could be the reason of this situation. Moreover, there is a need to improve the development level of the financial sector especially mortgage financing and the banking sector. However, there is good perspective since gradual financial deepening is underway in most SSA countries (Montfort et al., 2013). Although the mortgage market in Africa is still small by international standards, the sector is growing gradually and is attracting more and more attention from policy makers. Positive growth is now observed in some markets like Kenya and Uganda. For example Kenya's mortgage market grew by 37 percent in 2012, resulting in total up to 19,700 mortgages; while Uganda has shown signs of growth at the end of December 2012, representing about 0.98 percent of GDP which higher than the previous performance. According to Figure 1 bellow, South Africa and Namibia remain the market leaders in SSA in term of housing depth. figure 2 highlights practically the same classification with almost all value close to zero and a little difference since we have a few countries (Malawi, Djibouti, Chad, Mauritania, Mauritius, Tanzania) with higher level of mortgage penetration; the highest mortgage penetration rate in SSA which is 6.48 percent for Malawi is largely inferior to the one of several countries in the world including Sweden with almost 60 percent. 7

8 South Africa Namibia Kenya Botswana Senegal Rwanda Algeria Uganda Cameroon Nigeria Ghana Tanzania Chad Central African Republic Gabon Guinea Congo, Rep. Malawi Djibouti Chad Mauritania Mauritius Tanzania South Africa Angola Liberia Ghana Togo Rwanda Botswana Zambia Cameroon Kenya Uganda Central Afric. Rep. Mozambique Niger Mali Congo, Dem. Rep. Burkina Faso Madagascar Benin Gabon Congo, Rep. Nigeria Guinea Senegal Burundi 1,2 1,2 1,1 1,1 1,0 0,8 0,7 0,7 0,6 0,6 0,6 0,3 0,3 0,2 0,1 0,1 2,5% 2,3% 2,0% 1,2% 1,2% 1,0% 0,5% 0,4% 0,4% 0,2% 0,1% 0,1% 0,1% 0,0% 0,0% 2,8 2,3 2,1 1,8 1,8 1,7 3,6 4,3 30,8% 20,0% 5,7 5,4 5,4 6,5 6,3 6,3 6,1 Figure 1: Comparative mortgages depth across SSA countries (As a percentage of GDP for observation periods). 35% 30% 25% 20% 15% 10% 5% 0% Figure 2: Comparative mortgages penetration across SSA countries (Housing loan penetration). Note: Percentage of adult population with an outstanding loan to purchase a home. Unlike mortgage depth indicator, the penetration index refers to any provider of housing loans, including regulated financial institutions, microfinance institutions and other formal sources Source: Authors calculation using the World Bank data base on housing finance launched by Badev et al., 2014: Housing finance across countries: New data and analysis, WPS6756 Source: Authors calculation using the Global Financial Inclusion database (FINDEX) launched by Demirguc-Kunt, Asli and Leora Klapper (2012), Measuring financial inclusion: the global Findex Database, World Bank Policy Research Working Paper SSA is the region in the world where housing finance is the least developed. In fact, in SSA, many countries have less than 1 percent mortgage depth or don t have any mortgage market. Moreover,, its most advanced countries such as South Africa (30.8 percent) and Namibia (20.0 percent) in terms of mortgage market development, are far behind many countries in the world with deepen mortgage market like Denmark (109.8 percent). Figure 3 below shows that annual housing demand is increasing. This could be explained by increasing urbanisation rate. Figure 3: Annual housing demand evolution in SSA for (projection) period Source: Authors calculation using the World Bank data base on housing finance launched by Badev et al., 2014: Housing finance across countries: New data and analysis, WPS

9 Proportion of population access to mortgage market (%) Mortgage penetration (%) Many initiatives have been taken in Africa to support the development of the mortgage market. For example, the recent creation of national and regional mortgage refinancing institutions in Nigeria (2013), Tanzania (2011), and WAEMU sub-region (2010) will improve mortgage loans granting. It is then a potential way forward for integration between the housing sector and financial markets for more long-term liquidity, and at a reasonable cost. Unfortunately, as shown in figures 4 and 5 below, in almost all African countries, interest rates are very high, hampering its accessibility. Figure 4: Mortgage market access and housing finance policy in SSA. 70 CAP MAS 60 SOU 50 SEN GUB NAM BOT SWA y = -0,7327x + 22,677 SUD ANG R² = 0,0762 COM SIE ETH GAM MAD Mortgage policy (%) Figure 5: Mortgage penetration and housing finance policy in SSA SOU DJI CHA TOG MAL MAU ANG MAR 11,708 LIB BTS CMR UGA KEN CAR ZAM MAI NGR BEN GAB BUR CDR 14,5 BDI CON RWA GUI y = -0,0665x + 3,5968 R² = 0,0405 GHA MAD Mortgage policy (%) Regression Mortgage access = f( Mortgage policy) Linéaire (Regression Mortgage access = f( Mortgage policy)) Source: Authors calculation using the World Bank data base on housing finance launched by Badev et al., 2014: Housing finance across countries: New data and analysis, WPS6756. Note: The figure is a partial scatter plot, visually representing the regression of changes in the mortgage access (between average) on the mortgage rate ( average). The abbreviations next to the observations are the three-letter country codes as defined by the International Organization for Standardization. Mortgage penetration=f(mortgage policy) Linéaire (Mortgage penetration=f(mortgage policy)) Source: Authors calculation using the Global Financial Inclusion database (FINDEX) launched by Demirguc-Kunt, Asli and Leora Klapper (2012), Measuring financial inclusion: the global Findex Database, World Bank Policy Research Working Paper Note: The figure is a partial scatter plot, visually representing the regression of changes in the mortgage penetration (between average) on the mortgage rate ( average). The abbreviations next to the observations are the three-letter country codes as defined by the International Organization for Standardization. This general point of view is completed by a specific case study of the housing financial system of four specific African countries (Tanzania, Zambia, Cameroon, and Angola) presented in annex. The description highlight principally the fact that the formal housing finance system is not well developed and is bank-based; it cannot therefore respond to the demand from economic agents and we are not surprised to see that housing supply is not at the level or affordable to population housing need SSA housing market and finance policy typology and comparison: A decomposition approach In order to take into account the heterogeneity between SSA countries, the 48 SSA countries are decomposed into various groups depending on characteristics such as wealth, legal origins, oil-exporting, and sahel versus forestry and regional proximity (see figure 6). 9

10 Figure 6: Poverty level across different types of SSA countries. 3,0 2,5 2,0 1,5 1,0 0,5 0,0-0,5-1,0-1,5-2, Oil exporters (LHS) Fragile states (LHS) Actual poverty rate in SSA (percent of population, RHS) Frontier markets (LHS) Others (LHS) Source: Authors calculation based WDI data base. Thus, it is justified to check if there is also a difference in terms of mortgage market policy and development across SSA countries. Additionally, we compare the effective performance of housing finance policy, depth, penetration, housing need, cheapest price, and input price with the benchmark performance. Wealth based decomposition From a wealth level point of view, as expected income level affects housing finance indicators. In fact, Table 1 shows that Upper middle income countries have better performance of housing finance indicators. Table 1: Lower income-middle income-upper income housing market and finance policy statistical comparison Housing finance policy (Mortgage rate) Housing finance depth (Mortgage debt to GDP / credit to GDP) Housing finance penetration (Financial inclusion indicator of percentage of adult population with outstanding loan to purchase a home) Housing finance access (Proportion of population who can access mortgage rate in SSA) Housing demand (Annual housing demand) Cheapest housing price (Price of the cheapest, newly built house by a formal developer or Mean Std. dev. Benchmark Comment Low 15,77 8,25 Performance order Lower middle 16,12 4,99 Low is: UM, L, LM. Upper middle 12,94 1,74 Low 0,23 0,63 Performance order Lower middle 0,37 0,65 High is: UM, LM, L. Upper middle 13,29 14,69 Low 2,11 2,21 Performance order Lower middle 2,17 2,15 High is: UM, LM, L. Upper middle 2,57 2,49 Low 2,38 3,87 Lower middle 16,26 20,26 Upper middle 35,72 11,41 Low 112,73 127,9 Lower middle 115,33 201,33 Upper middle 23,67 34,58 Low Lower middle Upper middle High Low Low Performance order is: UM, LM, L. Performance order is: UM, L, LM. Performance order is: UM, LM, L. 10

11 contractor) Housing input price (Cost of standard 50kg bag of cement in US$) Low 12,21 5,1 Lower middle 10,42 1,96 Upper middle 7,61 0,53 11 Low Performance order is: UM, LM, L. Source: Authors calculation based on a collection of data from a survey, the World Bank and Housing Finance in Africa yearbooks 2010, 2011 and Available at: (accessed: 08/09/2014). Notes: All measures are in US$. The countries included in each group are: (1) Lower income countries: Guinea-Bissau, Guinea, Sierra Leone, Liberia, Mali, Burkina Faso, Togo, Benin, Niger, Chad, Central African Rep., Dem. Rep. of Congo, Uganda, Kenya, Ethiopia, Eritrea, Somalia, Tanzania, Mozambique, Zimbabwe, Madagascar, Comoros; (2) Lower middle income countries: Cape Verde, Senegal, Mauritania, Ivory Coast, Ghana, Nigeria, Cameroon, Congo, Angola, Zambia, Lesotho; (3) Upper middle income countries: Gabon; Namibia, Botswana, South Africa, Equatorial Guinea. Legal origins based decomposition It is important to check in an analytical and quantitative point of view if there is a significant difference of the state of housing market and finance policy by taking into consideration the legal origins of countries. The main legal origins in SSA considered are the French (code civil) and British (common law) systems. Table 2 shows that countries with French based legal system perform better on mortgage rate policy while countries with British based legal system perform better on Housing finance depth, housing finance penetration, and housing finance access. Table 2: French and British legal origin countries housing market and finance policy statistical comparison Housing finance policy (Mortgage rate) Housing finance depth (Mortgage debt to GDP / Credit to GDP) Housing finance penetration (Financial inclusion indicator of percentage of adult population with outstanding loan to purchase a home) Housing finance access (Proportion of population who can access mortgage rate in SSA) Cheapest housing price (Price of the cheapest, newly built house by a formal developer or contractor) Housing input price (Cost of standard 50kg bag of cement in US$) French speaking English speaking French speaking English speaking French speaking English speaking Mean Std dev. Benchmark Comment 15,91 7,25 French speaking countries Low perform better than 16,16 6,3 English speaking countries 0,14 0,45 3,92 8,99 1,88 2,25 3,26 2,27 High High English speaking countries perform better than French speaking countries English speaking countries perform better than French speaking countries French 10,38 16,45 English speaking countries speaking High perform better than French English speaking 10,96 17,58 speaking countries French Both tend to have the same speaking Low performance English speaking French 12,29 5,3 Both tend to have the same speaking Low performance English 10,49 3,94 speaking Source: Authors calculation based on a collection of data from a survey, the World Bank and Housing Finance in Africa yearbooks 2010, 2011 and Available at: (accessed: 08/09/2014). Notes: All measures are in US$. The countries included in each group are: (1) French speaking countries: Angola, Benin, Burkina Faso, Burundi, Cameroon, Cape Verde, Central African Rep., Chad, Comoros, Congo, Ivory Coast, Dem. Rep. of Congo, Djibouti, Equatorial Guinea, Gabon, Guinea, Guinea-Bissau, Madagascar, Mali, Mauritania, Niger, Sao Tome & Principe, Senegal and Togo; (2) English speaking countries: Botswana, Eritrea, Ethiopia, Gambia, Ghana, Kenya, Lesotho, Liberia, Malawi, Mauritius, Mozambique, Namibia, Nigeria, Rwanda, Seychelles, Sierra Leone, Somalia, South Africa, Sudan, Swaziland, Tanzania, Uganda, Zambia and Zimbabwe.

12 GDP growth (%) Gini coeficient (%) Other grouping We found that oil-exporting countries have better performance in terms of housing finance than oil nonexporting countries. The income level might be the better explanation of this situation. Besides, regional proximity is relevant for housing finance performances. We found that ECCAS present best performances in housing finance policy and demand; SADC have best performances in housing finance depth, penetration and access; ECOWAS have best performances in housing price and input price. EAC presents average performance in all housing finance characteristics Housing finance, inclusive growth, and inequality in SSA: a statistical analysis Recent literature states that in the world in general and in Africa in particular, more urbanized countries have lower inequality rate and that urbanization goes hand in hand with more prosperity (Fox and Sohnesen, 2012; Christiansen, 2013; World Bank, 2013). The process of urbanization should be accompanied by growth of affordable housing supply to respond to the consequent growing demand of housing. The development of the housing finance system could be an excellent tool for this objective. This statement is confirmed by figures 7 and 8 which show that mortgage penetration is positively correlated with economic growth but negatively correlation with the Gini coefficient. It means that better mortgage penetration spurs growth and reduces inequality. Housing finance can therefore boost inclusive growth and reduce inequality in Africa. Figure 9 and 10 shows that housing finances are positively correlated to two key indicators of national wealth for SSA countries. In fact, in SSA, Mortgage depth is positively correlated to higher GDP per capita growth. Also there is a nonlinear positive relationship between mortgage penetration and HDI index. It means that countries with greater mortgage penetration have a better level of human development. Figure 7: Mortgage penetration and economic growth in SSA ,5-1,5 y = 0,1383x - 1, R² = 0,0696-2,5 Mortgage penetration (%) Source: Authors calculation based on a collection of data from a survey, the World Bank and Housing Finance in Africa yearbooks 2010, 2011 and Available at: (accessed: 08/09/2014). Note: logarithm have been applied to both axes variables. -1 GDP growth function of mortgage penetration Linéaire (GDP growth function of mortgage penetration) Figure 8: Mortgage penetration and Gini coefficient in SSA 0,7 0,6 0,5 0,4 0,3 0,2 y = -0,489x + 0,4524 R² = 0,0143 0,1 0-0,02 0 0,02 0,04 0,06 0,08 Mortgage penetration (%) Gini coeficient function of mortgage depth Linéaire (Gini coeficient function of mortgage depth) Source: Authors calculation based on a collection of data from a survey, the World Bank and Housing Finance in Africa yearbooks 2010, 2011 and Available at: (accessed: 08/09/2014). Note: logarithm have been applied to both axes variables. Figure 9: Nexus between mortgage penetration and GDP per capita growth in SSA Figure 10: Mortgage penetration and HDI index 12

13 GDP per capita growth (%) HDI index (%) Source: Authors calculation based on a collection of data from a survey, the World Bank and Housing Finance in Africa yearbooks 2010, 2011 and Available at: (accessed: 08/09/2014). Source: Authors calculation based on a collection of data from a survey, the World Bank and Housing Finance in Africa yearbooks 2010, 2011 and Available at: (accessed: 08/09/2014). NB: Logarithm have been applied to both axes variable. Overall there is a positive link between housing finance and shared prosperity and between housing finance and poverty reduction. In order to complete this exploratory, we have presented some selected SSA country housing finance market typology in annexes. The case study focuses on four SSA countries from three different subregions. These countries are Cameroon for central and West Africa, Angola for central Africa, Tanzania for east Africa and Zambia for southern Africa IV. Empirical investigation and analysis 4.1. Regression models specification In order to answer the three questions of interest, three regression models are used. The first empirical investigation is based on a theoretical model of microeconomic study of the banking sector. Indeed, the SSA housing financial system is essentially constituted of banks and we think that a coherent study of housing finance determinants must consider the banking system. For this purpose, the theoretical model presented by Dehesa (2007) and replicated with modification by Nguena and Tsafack (2014) is used and augmented by taking into account the housing finance stylized facts in Africa. The model explains the theoretical link between financial deepening and factors related to sector banking and real economy operation by focusing on the credit market since there is a small proportion of lending over deposits and excess liquidity banks in Africa. From this theoretical model, we derive a reduced form of our first log-log empirical model to assess housing finance determinants; it is given by: HFIN it = + 1 MACAP it+ 2 Upop it + 3 GDPPC it + 4 GDPgrowth it + X it + it Where: 0,15 0,1 0,05-0,1-0,15 y = 0,0091x + 0,0233 R² = 0, , Mortgage depth (%) Shared prosperity function of mortgage depth Linéaire (Shared prosperity function of mortgage depth) , y = 0,0276x - 0,3351 R² = 0,0275-0,1-0,15-0,2-0,25-0,3-0,35-0,4-0,45-0,5 Mortgage penetration (%) Poverty reduction function of Mortgage penetration Linéaire (Poverty reduction function of Mortgage penetration) 13

14 1. HFIN is a housing finance indicator, MACAP a financial market indicator, Upop the ratio of urban population to total population, GDPPC the GDP per capita which is an indicator of the level of development of the country, GDP growth; and X is a set of control (see appendix for more detail). 2. are parameters, B a matrix of parameters, it the disturbance term; and the other variables are defined in appendix. The second model assesses the linkage between housing finance and economic growth. It is based on theoretical argument developed in the literature review and specific finding about the SSA housing market presented in section 3. Starting by the theoretical economic growth model developed by Mankiw et al. (1992), we have defined several variables including housing finance indicator which were assumed to impact the real block. Theoretical predictions advocate that financial development (including housing finance development) contributes directly to inclusive growth and inequality reduction: first, in a direct way through savings, insurance services and access to credits that can enhance the productivity of assets the poor by allowing them to invest in new technologies, or investing in education and health. Financial development can improve opportunities for the poor to have access to formal finance (Jalilian and Kirkpatrick, 2001); there is also a multiplier effect 5 during and after housing construction through job creation. Second, financial system enables the poor to access financial services, particularly credit and insurance risk, enhancing the productive assets of the poor, by improving productivity and increasing the potential to achieve sustainable gains (Jalilian and Kirkpatrick, 2001). It is given by ( ) ( ) ( ) (2) Where are parameters, it the disturbance term. The set of control variables account for all relevant determinants of GDP per capita given by the neoclassical growth theory such as labor, capital, etc. In addition it accounts for legal origin and regional consideration highlighted in section 3. The third model assesses the linkage between housing finance and inequality. It is based on theoretical arguments developed in the literature review and specific findings about the SSA housing market presented in section 3. Housing finance can help reduce poverty through mainly economic growth and job creation. Additionally, housing finance development can theoretically help reduce poverty through the market of labor (unleash recruitment, reduce unemployment, increase household revenue through salaries and thus reduce poverty); the market of goods and services (increase in household services and good consumption and thus reduce poverty through household business revenue increase) and the housing market (reduction of housing price, increase of household finance capacity and thus reduce poverty). 5 Multiplier Effects from Housing during Construction: the first impacts are the jobs and spending that occur when the housing is constructed. Jobs created during this phase include the construction workers who are building the structure as well as workers in their firms who support them, such as office managers, cost estimators, and accountants. Multiplier Effects from Housing after Construction: after construction, housing is occupied and further economic impacts occur. If the housing units are rented, rental payments contribute to the economic activity. The household income of all housing residents contributes to the economic activity when households spend their disposable income on items such as food, clothing, transportation, and health care. 14

15 As the financial system becomes healthier and more competitive, it may have more capacity and desire to bear the high costs of small credits (Rajan and Zingales, 2003). For instance, in Latin America commercial banks have begun to make pooled loans available to the poor, as microcredit institutions have been doing (Mosley, 1999). Also, the evolution of informal credit, often the only source of borrowing for poor people, is made easier by the growth of a formal financial system that provides informal institutions with opportunities for profitable deposits. Moreover, access to credit enables the poor to smooth their consumption, thus reducing their vulnerability to exogenous shocks and building human capital. Following this presentation of the theoretical relation, we based our empirical model on the model developed by Davis and Nieuwerburgh (2014) and Acemoglu (2009); he built a household centred partial equilibrium model by considering of human capital investments with imperfect credit markets (as it is the case in practically all the SSA region). Thus the third empirical model is given by: ( ) ( ) ( ) (4) Where are parameters, the disturbance term; and INQ an inequality indicator The set of control variables account for all relevant determinants of inequality given by the development theory such as employment, education, gender inequality, etc. It also accounts for key findings highlighted in section Estimation Strategy The choice of panel data analysis gives us the advantage of having a reasonable size of sample and time series data for analysis, which could not have been performed on each of the individual countries. Additionally, the double dimension of panel data allows us to take simultaneously into account the dynamic behavior and possible heterogeneity across countries, which is neither possible with time series nor with cross-sectional data. We will consider both static and dynamic specification. To estimate the static model specification, we carry specification and robustness check including stationary and cointegration test; based on the result, we will apply either the FGLS (Feasible Generalized Least Square) in case of presence of heteroscedasticity or XTPMG Stata command in case of nonstationary problem. To estimate the model, we use the System Generalized Method-of-Moment (GMM) estimator developed by Blundell and Bond (1998). The estimator combines two set of equations. The first set includes first-differenced equations where the right-hand-side variables are instrumented by the levels of the series lagged one period or more. The second set consists of the equations in levels with the righthand side variables being instrumented by lagged first of higher-order differences. This estimator has several advantages. 6 It takes into account country-specific effects, while addressing issues associated with endogeneity, measurement errors, and omitted variables. By exploiting internal instruments, the System GMM estimator removes the often hard task of identifying valid external instruments consisting of variables that are correlated with the endogenous explanatory variable but not with the error term of the equation. The validity of these internal instruments (lagged variables in level and first differences) was not rejected. As suggested by Arellano and Bond (1991), and Blundell and Bond (1998), a Sargan/Hansen test of overidentifying restrictions and a serial correlation test were carried out. In both instances, the null hypothesis could not be rejected (the instrumental variables are not correlated with the residual, and the errors exhibit 6 Bond, Hoeffler, and Temple (2001) offer a good overview on GMM estimation of empirical growth models. 15

16 no second-order serial correlation). In addition, to limit the risk of over-instrumentation, we keep the number of instruments to the minimum by using as instrument only the first valid lagged value of the right-hand side variables. We assume that financial variables are endogenous, therefore are instrumented by their second lag value, while the other variables, treated as predetermined, are instrumented by their first lag value. We will also test the existence of threshold if necessary Results Analysis and discussion We present below estimation results of the three models. Housing finance determinants Table 6 below presents results on housing finance determinants. It shows that the stock market capitalization, the growth rate of the urban population, and the dummy variable of lagged conflict are key determinants of housing finance in SSA. Stock market capitalization and the growth rate of urban population are the strongest positive determinants, while recovery from conflict is not significant for some specifications. In addition, some variables seem to be hampering housing finance depending on the economic environment. These variables are: the credit to the economy, which is a proxy of the development of the banking sector and Trade openness, which may masque the low housing finance in oil exporting countries. Table 6: Housing finance determinants (Housing finance depth) Model A1: Housing finance determinants (Housing finance depth) (1) (2) (3) (4) (5) Market capitalization 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** ( ) ( ) ( ) ( ) ( ) Urban population growth * ** * * ** ( ) ( ) ( ) ( ) ( ) Conflictdummy t ** 0.008** 0.007** (0.003) (0.003) (0.003) (0.002) (0.003) Investment stock (0.002) (0.002) (0.002) (0.002) (0.002) GDP per capita (0.002) (0.002) (0.002) (0.002) (0.002) GDP growth (0.002) (0.002) (0.002) (0.002) (0.002) Voice accountability (0.0001) (0.0001) (0.0001) ( ) (0.0001) Domestic credit to private sector ** (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) Trade openness ( ) ( ) ( ) ( ) ( ) Inflation (0.0002) (0.0002) (0.0002) (0.0001) (0.0002) Interest rate (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) Transfers 8.95e e e e e-20 (1.15e-19) (1.1e-19) (1.1e-19) (9.3e-20) (1.1e-19) Kaolegal ** (0.002) Kaosahel (0.005) 16