Wealth, Financial Intermediation, and Saving in Latin America and the Caribbean

Transcription

1 Wealth, Intermediation, and Saving in Latin America and the Caribbean Ricardo Bebczuk Institutions for Development Sector Capital Markets and Institutions Division DISCUSSION PAPER Nº IDB-DP-406 October 2015

2 Wealth, Intermediation, and Saving in Latin America and the Caribbean Ricardo Bebczuk October 2015

3 Copyright 2015 Inter-American Development Bank. This work is licensed under a Creative Commons IGO 3.0 Attribution-NonCommercial-NoDerivatives (CC-IGO BY-NC-ND 3.0 IGO) license ( licenses/by-nc-nd/3.0/igo/legalcode) and may be reproduced with attribution to the IDB and for any noncommercial purpose. No derivative work is allowed. Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to the UNCITRAL rules. The use of the IDB's name for any purpose other than for attribution, and the use of the IDB's logo shall be subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC-IGO license. Note that link provided above includes additional terms and conditions of the license. The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the Inter-American Development Bank, its Board of Directors, or the countries they represent. Contact: Cesar Tamayo Tobon, ctamayo@iadb.org.

4 Wealth, Intermediation, and Saving in Latin America and the Caribbean Ricardo Bebczuk * Abstract After constructing a broad and novel database of 52 countries over , this study assesses the link between financial intermediation and saving. The study finds that the Latin American and Caribbean (LAC) region lags well behind other regions in terms of financial depth, as measured by gross private domestic financial assets. LAC countries also have a larger share of bank deposits and cash in the private sector portfolio, compared to non-bank assets (bonds and shares). Moreover, within the institutional investor industry, pension funds are relatively developed in the region, although they grew out of the compulsory pension systems in several countries that date back to the 1980s and 1990s. The findings also indicate that LAC countries have about 40 percent of gross private financial wealth invested abroad, but just 4 percent of gross private liabilities have that origin, which attests to region s obstacles in tapping international markets. The countries in general present a small share of household and business saving being intermediated through the financial system. In the specific case of bank deposits, just 5 percent of household saving and 3 percent of business saving are kept in the banking system. JEL Codes: G2, E21 Keywords: intermediaries, banking system, capital markets, institutional investors, household saving, corporate saving. * The author would like to thank Eduardo Cavallo, Edgardo Demaestri, Andy Powell, and César Tamayo for the their insightful comments and suggestions, as well as Máximo Sangiácomo for his expert advice on statistical and econometric issues. 1 Gross Private Foreign Assets are defined as Gross National Foreign Assets minus Central Bank Foreign Reserves.

5 Introduction While the bulk of the saving literature is focused on pinpointing the determinants of saving decisions, little effort has been devoted to tracking saving from its generation to its final uses. The latter issue is important because, after all, what matters is not saving per se, but what is done with it. This paper seeks to fill this gap for a broad set of countries, with special emphasis on the Latin America and the Caribbean (LAC) countries. To this end, a number of international and national datasets, listed at the end of the document, were employed to determine the use of saving. After eliminating countries and years with incomplete and/or unreliable data, we ended up with a sample of 52 countries over , including 18 LAC economies, plus 23 highincome OECD, eight upper-middle income and three lower-middle income economies (see Annex for the complete country list). To the best of our knowledge, this is the first analysis of this sort for the LAC region. Taking an international perspective, this paper addresses the following underresearched and key questions on the link between financial intermediation and saving: What is the depth and structure of the domestic financial system in LAC countries in terms of instruments and intermediaries? With a broader scope, what is the volume and allocation of private and national wealth, including both domestic and foreign assets and liabilities as well as financial and real assets (capital stock)? What is the relationship between those flows and stocks of wealth and the national and private saving rate? To what extent and in what instruments do households and non-financial firms invest their saving? To formalize the link between saving and financial intermediation, as well as other uses of saving, national accounts and balance of payments data will be used. However, unlike the traditional above-the-line items (revenues and expenses), our analysis will be concerned with below-the-line items, or financial accounts. Despite the lack of empirical applications directly comparable to the present one, the underlying macroeconomic stock-flow methodology is anything but new (see, for example, IMF, 2007, and Patterson and Stephenson, 1988). Among other uses, this methodology enables us to conduct a macroeconomic consistency exercise linking saving and its final applications. In this regard, Table 1 describes the various possible uses of gross national saving. Domestic saving is generated by three economic units (households, businesses, and the

6 government) and can be invested in financial or real (physical) assets. uses, in turn, can take the form of plain currency, or cash, debt or equity. These financial instruments may be issued by domestic and foreign residents. In addition, they can be either intermediated or directly acquired by the saver; if the former, the financial intermediary can be a bank or similar institution or an institutional investor (pension funds, mutual funds, and insurance companies). Finally, the asset value can be recorded on a gross or a net basis: for financial assets, net assets are the difference between gross assets and gross liabilities, whereas net physical capital is gross capital minus depreciation. Table 1. Saving Uses Type of assets instrument Saving unit Asset value Nationality of issuer intermediary Gross Domestic Debt Households Banks Pension funds Stock Businesses Mutual funds Real Currency Government Net Foreign Insurance companies Nonintermediated Based on the above-described general outline, the paper will be organized in three sections. The first one will characterize the domestic financial system in LAC countries and other country groups, in terms of depth, intermediaries, and asset classes. The second section has a broader scope, incorporating into the analysis assets other than domestic financial assets, such as foreign financial assets and physical capital. Finally, and central to this paper s goal, the third section will investigate the link between saving decisions and the accumulation of this comprehensive array of assets. Some conclusions will close.

7 1. The Structure of the System in the LAC Region This section portrays the structure of the domestic financial system in the LAC region and other country groups from two different standpoints. First, we will evaluate the overall depth of the system and then will decompose gross private domestic financial assets by asset class and by intermediary type. Second, we will depict the structure of gross private financial liabilities. In terms of asset classes, the gross domestic financial assets of the private sector can be held in the form of currency, deposits, stock, and debt securities, the latter issued by either the private or the public sector. In turn, these assets can be intermediated by the financial system, including in this category, banks as well as non-bank institutions such as credit unions, cooperatives, and the like, or by institutional investors such as pension funds, mutual funds, and insurance companies. Some of these assets are not intermediated, but are purchased directly by the final investor, as in the case of currency and debt and stock securities not being managed by institutional investors. A detailed breakdown of intermediated and nonintermediated assets requires a wealth of information that is not readily available. For that reason, we will take a shortcut by calculating non-intermediated assets as the difference between total assets (strictly speaking, gross private domestic financial assets, as previously measured) and the sum of bank (and similar intermediaries) deposits and assets held by institutional investors. In sum, the above discussion boils down to the following identities classifying assets by class-equation [1]- and by intermediation status-equation [2]. Gross Private Domestic Assets = = Currency + Stock Market Capitalization + Outstanding Domestic Private Debt Securities + Outstanding Domestic Public Debt Securities + System Deposits [1] = System Deposits + Pension Fund Assets + Mutual Fund Assets + Insurance Companies Assets + Non-intermediated Assets [2] Table 1.1 shows financial deepening as measured by the holdings of gross domestic financial assets in the hands of the private sector (equation [1]). All individual LAC countries as well as the median for the LAC region and benchmarking country groups are displayed for the first and last year available in the dataset (2001 and 2011). Country group data reveals that the LAC region lags behind high-income and both upper- and lower-income economies. For 2011, gross

8 domestic financial assets amount to 61 percent, against 276 percent in high-income countries, 108 percent in upper-middle countries and 100 percent in lower-income economies. On top of that, comparing to 2001 figures, LAC has experienced a slower expansion than the other groups, with a change of 10 percentage points of GDP (from 51 percent to 61 percent). This increase reached 54 pp. for OECD and 25 and 4 pp. for upper- and lower-middle income economies, respectively. These summary values hide a large dispersion across the LAC region. In 2011, 8 out of the 18 countries have gross domestic financial assets for 90 percent or more of GDP, with Chile leading the list with 204 percent. On the other extreme, Venezuela, Dominican Republic, Paraguay, Uruguay, Ecuador, El Salvador, and Argentina are all below 50 percent. In Tables 1.2 and 1.3 we present, in terms of GDP and as a percentage of total gross domestic financial assets, the contribution of the various components. Two distinctive features emerge from these tables: for one, deposits represent a higher proportion of total assets in LAC (48 percent), upper-middle-income countries (46 percent) and lower-middle-income countries (50 percent) than in high-income OECD economies (34 percent). Similarly, investors show a greater preference for cash holdings: against the 3 percent observed in the OECD economies, this ratio is 8 percent, 9 percent, and 11 percent in LAC, upper-middle, and lower-middle-income countries, respectively. The previous figures are consistent with the conventional wisdom that more developed economies tend to have more capital market-based financial systems than less developed countries, where investors as a result of higher macro volatility, limited market liquidity as well as deficient corporate governance practices and thus fear of insider expropriation have more appetite for traditional and less sophisticated assets with low but predetermined returns, such as bank deposits and cash. Especially noticeable is the asymmetric development of the private bond market in OECD countries (48 percent of GDP and 17 percent of total assets) vis-à-vis the remaining groups, where this market is quite shallow or downright undeveloped. In the case of LAC, the median is 0 percent, although a few exceptions stand out, including Brazil (22 percent of GDP and 12 percent of total), Chile (15 percent and 7 percent), and Mexico (16 percent and 15 percent). Nevertheless, it must be taken into account that, despite this relative composition tilted to some degree toward capital market assets, developed countries surpass the other groups in terms of financial depth in every category, non-capital market instruments included. For instance, the deposits-to-gdp ratio is three times larger in OECD (94 percent) than in LAC

9 countries (29 percent) and about twice as large as in middle-income countries. Even cash holdings are higher in OECD economies (8 percent) than in LAC (5 percent), although lower than in middle-income economies (10 percent in upper-middle and 11 percent in lower-middleincome countries). Table 1.1. Gross Private Domestic Assets (in percent of GDP) Country Change Argentina Bolivia Brazil Chile n.a. 204 n.a. Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

10 Table 1.2. Gross Private Domestic Assets: Composition by Asset Class, 2011 (in percent of GDP) Country Private Gross Domestic Assets Currency Stock Market Capitalization Outstanding Domestic Private Debt Securities Outstanding Domestic Public Debt Securities System Deposits Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

11 Table 1.3. Gross Private Domestic Assets: Composition by Asset Class, 2011 (in percent of GDP) Country Private Gross Domestic Assets Currency Stock Market Capitalization Outstanding Domestic Private Debt Securities Outstanding Domestic Public Debt Securities System Deposits Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

12 Economic units may invest for themselves by hoarding cash or buying bonds and stock in the capital market or by delegating asset management to banks or institutional investors (mutual funds, pension funds, and insurance companies). Table 1.4 shows the size of these three specialized intermediaries. It may come as a surprise that the median size of this industry is larger in LAC than in the OECD. However, the pension funds sector is the only reason for this. As it is well-known, a number of LAC economies adopted private pension systems during the 1980s, and especially during the 1990s. Since these systems are mandatory and they are still in a largely accumulation base (i.e., most members are young and decades away from retirement), assets under management have been growing at a fast pace. Considering voluntary saving only, as in the case of mutual funds and life insurance, the OECD fares far better than LAC. Table 1.5 compares the importance of bank deposits relative to the institutional investor industry. As the economy and the financial systems develop, the latter is expected to gain a more dominant presence at the expense of the former, as investors become more sophisticated and demand a broader set of assets. The table validates this view, showing that the median share of the institutional investor industry reaches 72 percent in OECD countries and 62 percent in LAC. However, the difference does not seem to be as significant as expected. The explanation, once again, lies in the development of mandatory pension funds in several LAC countries.

13 Table 1.4. Institutional Investment Industry, 2011 (in percent of GDP) Country Mutual Funds Assets Pension Funds Assets Insurance Company Assets Institutional Investors Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

14 Table 1.5. Gross Private Domestic Assets by Intermediary Type: Banks versus Institutional Investors, 2011 (in percent of GDP) Country Mutual Funds Assets Pension Funds Assets Insurance Company Assets Institutional Investors System Deposits Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

15 development is customarily evaluated from the liabilities rather than the assets side, as in previous figures. As a matter of fact, the most popular measure of financial deepening in the literature is the ratio of credit to the private sector to GDP. Table 1.6 shows the stock of private sector liabilities that is, bank loans plus outstanding bonds--in 2001 and With a ratio of 39 percent in 2011, LAC displays a much shallower debt market than OECD (168 percent) and upper-middle-income countries (72 percent), and only comparable with that of the lower-middle income country group. Relative to 2001, LAC also shows a sluggish expansion (from 27 percent to 39 percent, a 12 p.p. increase), which is far below that of OECD economies (up 44 p.p.) and upper-middle-income countries (up 55 p.p.). Within the LAC region, only Brazil, Chile, and Panama have private debt levels above 80 percent. On the lower side, an appallingly high fraction of LAC countries, including Argentina, Dominican Republic, Ecuador, Guatemala, Jamaica, Peru, Uruguay, and Venezuela do not exceed 30 percent. Table 1.7 splits domestic debt into bank credit and corporate bonds, and then the overall value with that of gross domestic financial assets. For LAC, the data shows an overwhelming dominance of bank credit over bonds, largely consistent with the bank-centered structure of the financial system in the region. Only in six countries does the bond market boast some significant activity, with just three countries with outstanding bonds over 3 percent of GDP: Brazil, at 22 percent, Chile, at 15 percent, and Mexico, at 16 percent. The OECD group, with a median at 48 percent of GDP, is the region with the deepest corporate bond market. The same Table 1.7 underscores an important point: the stock of liabilities (such as via private credit to GDP, as it is common practice) as opposed to assets may convey a radically different message when gauging financial development. Assets exceed liabilities by a median factor of 1.6 in both LAC and a similar value in the OECD and the upper-middle-income countries. In seven LAC countries, assets more than double liabilities. The lesson that follows, also applicable to the analysis in coming sections, is that not a single metric can accurately measure financial deepening, and the one to be chosen depends on the phenomenon under study.

16 Table 1.6. Gross Private Domestic Debt Liabilities, 2011 (in percent of GDP) Country Change Argentina Bolivia Brazil Chile n.a. 85 n.a. Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

17 Table 1.7. Gross Private Domestic Assets and Debt Liabilities, 2011 (in percent of GDP) Country Domestic Credit to the Private Sector Outstanding Domestic Private Debt Securities Private Gross Domestic Debt Liabilities Private Gross Domestic Assets Assets/Debt Liabilities Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

18 2. From Domestic Assets and Liabilities to Net National Wealth Broadening the scope of the analysis, three aspects need to be considered to move from gross domestic financial assets and liabilities to net national wealth, by introducing into the picture (i) foreign assets, (ii) the physical capital stock, and (iii) the public sector. The first step is to notice that the financial wealth of the private sector can be allocated to not only domestic but also foreign assets, and the same goes, obviously, for liabilities: 1 Gross Private Assets = = Gross Private Domestic Assets + Gross Private Foreign Assets [4] Gross Private Debt Liabilities = = Gross Private Domestic Liabilities + Gross Private Foreign Liabilities [5] Tables 2.1 and 2.3 report domestic and foreign assets as a share of GDP, whereas Tables 2.2 and 2.4 measure them as a proportion of total assets (domestic plus foreign). Against the background of the home bias literature (see, for example, Schmitt-Grohé and Uribe, 2014), these numbers suggest a relatively large share of foreign assets in investors portfolios in both LAC and other economies. As of 2011, 40 percent of gross private financial assets in LAC were kept in foreign instruments and a similar value was observed in OECD economies (38 percent); this percentage fell to about 25 percent in the other reference groups. In terms of GDP, LAC foreign assets moved to 41 percent, up from 21 percent in 2001, expanding their share in total from 29 percent to above 40 percent. This proportion has decreased in OECD countries, from 42 percent to 38 percent, and for upper-middle income economies, from 28 percent to 24 percent. This difference lends itself to two polar interpretations: the more benign would be that countries are more intensively exploiting the benefits of international diversification; or, it may be signaling a flight to quality on the part of domestic investors driven by a perception of heightened instability at home. Foreign assets constitute more than 40 percent of total assets in Argentina (47 percent), Panama (48 percent), Uruguay (49 percent), and Venezuela (71 percent). 1 Gross Private Foreign Assets are defined as Gross National Foreign Assets minus Central Bank Foreign Reserves. Note that foreign assets and liabilities include equity investments (direct and portfolio) and debt (bank loans and corporate bonds).

19 An apparent gap exists between foreign assets and liabilities. For LAC countries, liabilities amount to just 2 percent of GDP, with all countries below 6 percent, and Jamaica as the outlier, with 33 percent. Similarly, low values are found in upper-middle-income countries (1 percent) and lower-middle-income countries (6 percent), in sharp contrast to OECD (59 percent). This difference with the relatively high level of gross assets presented above simply illustrates the fact that saving decisions are considerably less constrained than borrowing decisions the former depend solely on our own expenditure decisions and the latter on the creditor s assessment of our creditworthiness. Since OECD countries have a better historical track record as borrowers than other nations to some extent tainted by the post-2008 crisis developments in some European countries the difference is not particularly puzzling. Reaffirming this conclusion, according to Table 2.4, foreign liabilities are also small relative to domestic liabilities, accounting for just 4 percent of total (26 percent in OECD countries). Based on informational barriers, borrowing constraints appear to be more binding in tapping overseas vis-à-vis local financial markets.

20 Table 2.1. Gross Private Domestic and Foreign Assets (in percent of GDP) Country Private Gross Domestic Assets, 2001 Private Gross Foreign Assets, 2001 Private Gross Assets, 2001 Private Gross Domestic Assets, 2011 Private Gross Foreign Assets, 2011 Private Gross Assets, 2011 Argentina Bolivia Brazil Chile n.a. n.a. n.a Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

21 Table 2.2. Gross Private Domestic and Foreign Assets (In percent of total) Country Private Gross Domestic Assets, 2001 Private Gross Foreign Assets, 2001 Private Gross Assets, 2001 Private Gross Domestic Assets, 2011 Private Gross Foreign Assets, 2011 Private Gross Assets, 2011 Argentina Bolivia Brazil Chile n.a. n.a. n.a Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

22 Table 2.3. Gross Private Domestic and Foreign Liabilities (in percent of GDP) Country Private Gross Domestic Debt Liabilities, 2001 Private Gross Foreign Debt Liabilities, 2001 Private Gross Debt Liabilities, 2001 Private Gross Domestic Debt Liabilities, 2011 Private Gross Foreign Debt Liabilities, 2011 Private Gross Debt Liabilities, 2011 Argentina Bolivia Brazil Chile n.a. n.a. n.a Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

23 Table 2.4. Gross Private Domestic and Foreign Liabilities (in percent of GDP) Country Private Gross Domestic Debt Liabilities, 2001 Private Gross Foreign Debt Liabilities, 2001 Private Gross Debt Liabilities, 2001 Private Gross Domestic Debt Liabilities, 2011 Private Gross Foreign Debt Liabilities, 2011 Private Gross Debt Liabilities, 2011 Argentina Bolivia Brazil Chile n.a. n.a. n.a Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

24 The last step to compute the net assets of the private sector second step is to compute net financial assets of the private sector as the difference between gross assets and liabilities, and then adding the stock of physical capital: 2 Net Private Assets = = Gross Private Assets Gross Private Debt Liabilities [6] Net Private Assets = = Net Private Assets + Capital Stock [7] Table 2.5 shows net private financial assets to GDP in 2001 and The median for LAC is 46 percent of GDP in 2011, similar to the value for upper- and lower-middle-income countries (48 percent and 44 percent, respectively). However, this ratio is about a quarter the OECD median (185 percent). Moreover these net assets display a modest growth of just 7 percentage points of GDP over the decade, which pales against the 46 p.p. expansion in high income OECD countries. Net private assets plus the stock of capital equals net private assets, with the latter encompassing both financial and real assets. As apparent from Table 2.6, the stock of physical capital makes for the main component of net private wealth. 3 The median net private assets amount to 320 percent of GDP in LAC, 604 percent in OECD economies, 342 percent in uppermiddle-income countries, and 345 percent in lower-middle-income countries. As with other wealth measures, LAC appears to underperform both the OECD and the other benchmark economies. Compounding this situation, the gap between LAC and the remaining groups has widened over the last few years, as can be inferred from the fact that the OECD countries display a much more substantial increase of net private assets over the decade (173 p.p. of GDP) than in LAC (just 2 p.p.). The corresponding figures for upper-middle and lowermiddle income are 21 and 79 percentage points of GDP. 2 Recalling that liabilities and assets within the private sector cancel out, the net financial wealth of the private sector is equal to the sum of currency (a public sector liability), government debt (another public sector liability) and net foreign assets (gross foreign assets minus gross foreign liabilities). The capital stock represents a net asset of the private sector. 3 Available capital stock measures do not include residential construction but only productive capital, so the total stock of capital is grossly underestimated.

25 Reinforcing previous conclusions, summary Table 2.7 reports for 2011 the various private (financial and overall) wealth measures introduced so far, making crystal clear the dramatic differences emphasized throughout this section. Clarity as to what is intended to be measured and studied should thus guide the choice for the most suitable wealth indicator. Table 2.5. Net Private Assets (in percent of GDP) Country Change Argentina Bolivia Brazil Chile n.a. Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

26 Table 2.6. Net Private Assets (in percent of GDP) Country Change Argentina Bolivia Brazil Chile n.a. 377 n.a. Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

27 Table 2.7. Private Sector and Non- Wealth Measures (in percent of GDP) Country Gross Private Domestic Assets Gross Private Assets Domestic Credit to the Private Sector Gross Private Domestic Debt Liabilities Gross Private Debt Liabilities Net Private Assets Net Private Assets Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median.

28 Finally, net national assets (or national wealth) equal the sum of the capital stock and net national foreign assets: 4 Net National Assets = = Capital Stock + Net Foreign Assets [8] As reported in Table 2.8, net national assets are lower than net private assets in all four country groups, as a result of the public sector s debtor position in international markets. Table 2.8. Net National Assets (in percent of GDP) Country Change Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income Note: Country group values correspond to the group s median. 4 It must be borne in mind that, at the national level, private assets against the public sector (namely, currency and government debt) cancel out, leaving as net national wealth just the capital stock and net (gross assets minus gross liabilities) assets against the rest of the world.

29 3. Linking Net Wealth and Saving The previous sections have described the level and composition of national assets, distinguishing gross from net, domestic from foreign, and financial from real asset holdings. This last section is devoted to the ultimate goal of this paper, namely, assessing the relationship between such asset holdings and private and national saving. Basic national and balance of payment accounting implies that the following relationships must hold between some of the above variables and national and private saving (where Δ stands for year-to-year total change) (IMF, 2007; Schmidt-Grohé and Uribe, 2014): Δ Net National Assets = Net National Saving [9] Δ Net Private Assets = Net Private Saving [10] Net National Assets = Σ t Net National Saving [11] Net Private Assets = Σ t Net Private Saving [12] Simply put, equations [9] and [10] state that net national private saving equals the change in net national private wealth. In turn, equations [11] and [12] present the stock version of those flow identities, highlighting the fact that the stock of wealth is just the accumulation of net saving over time. Our analysis in earlier sections has revolved around stock indicators because the available information on financial assets is normally presented in the form of stocks. flows can be easily obtained by computing the period-to-period change. However, it must be noted that in practice the left- and right-hand sides of these four equations can yield largely different values due to the following reasons: Different statistical methods are applied to estimate financial wealth and saving, with wealth being mostly measured via financial flows and stocks, and saving via production and expenditure accounts. Available saving measures are typically gross of consumption of fixed capital, whereas the capital stock is net of depreciation. The capital stock accounts only for productive capital, but not residential construction, which is a major asset in household wealth portfolio.

30 Wealth is noticeably affected by capital gains and losses on outstanding assets and liabilities, which are not accounted for in current saving. Some financial and real transactions may go unrecorded because they take place in the informal market. This may include some forms of non-bank credit, trade credit, foreign currency purchases, the acquisition of real state assets, and others. All in all, the above accounting deficiencies may affect the consistency between changes in wealth and saving measures. Table 3.1 shows that both the private and the national saving rates in the LAC region has been generally lower than in other regions. In what follows, we conduct several regressions to check if a positive nexus exists between gross saving rates and net wealth, as put forward in equations [9] through [12]. The only legitimate dependent variables are net private and national assets, as these are the only variables that have a clearly identifiable relationship with saving there is no theory or accounting identity linking other financial indicators presented throughout the paper with saving, and lacking such theoretical background there is no way to interpret a regression on saving. Just for completeness we have added net private financial assets, even though this variable bears no systematic relationship with saving either, and the same applies to the link between private saving and national wealth. By the same token, although saving should be correlated to wealth flows (calculated as the annual change in stocks), the pervasive measurement errors likely to be encountered make it advisable also to run regressions of wealth stocks on average saving rates, under the presumption that this stock-based analysis may be less sensitive to statistical noise. In the same spirit of mitigating these errors by adopting a longas opposed to a short-term framework, we have adopted a cross-section rather a panel regression approach. 5 Also, even though those regressions are based on identities, we check robustness by controlling for some measures of macroeconomic and institutional development and stability, including the GDP growth rate, the inflation rate, per capita GDP, and the Rule of Law index. Our results, presented in Tables 3.2 through 3.4, point to a strong link between average saving rates and wealth stocks, and a mostly weak link with wealth flows. 5 Exploratory and unreported panel exercises yield non-significant results in all cases.

31 Table 3.1. Gross Private and National Saving (in percent of GDP) Country / Region Gross Private Saving Note: Country group values correspond to the group s median. Gross National Saving Argentina Bolivia Brazil Chile Colombia Costa Rica Dominican Rep Ecuador El Salvador Guatemala Honduras Jamaica Mexico Panama Paraguay Peru Uruguay Venezuela LAC High-income OECD Upper-middle-income Lower-middle-income

32 Table 3.2. Stocks of Private and National Wealth and Private Saving (1) (2) (3) (4) (5) (6) Explanatory Variables Net Private Assets Net Private Assets Net Private Assets Net Private Assets Net National Wealth Net National Wealth Gross private saving ** 5.269* 8.726*** 4.549** 5.836*** [2.130] [1.470] [3.113] [1.569] [2.024] [1.637] GDP growth *** [6.870] [7.640] [7.243] Inflation ** [1.290] [2.581] [2.515] Ln(GDP per capita) 51.41** 91.19*** 103.5*** [24.60] [20.16] [24.17] Rule of law ** [16.71] [26.63] [22.53] Constant * 319.3*** *** 200.4*** *** [37.45] [235.0] [56.36] [179.5] [42.16] [229.5] Observations R-squared Robust standard errors in brackets. *** p<0.01, ** p<0.05, * p<0.1

33 Table 3.3. Flows of Private and National Wealth and Private Saving (1) (2) (3) (4) (5) (6) Explanatory Variables Δ(Net Private Assets) Δ(Net Private Assets) Δ(Net Private Assets) Δ(Net Private Assets) Δ(Net National Wealth) Δ(Net National Wealth) Gross private saving 0.145* 0.187** 0.359** 0.497*** [0.0851] [0.0734] [0.149] [0.159] [0.160] [0.187] GDP growth *** * [0.408] [0.575] [0.587] Inflation * [0.107] [0.264] [0.218] Ln(GDP per capita) [1.208] [1.390] [3.152] Rule of law [1.208] [1.626] [2.804] Constant [1.498] [11.59] [2.591] [15.14] [2.808] [32.09] Observations R-squared Robust standard errors in brackets. *** p<0.01, ** p<0.05, * p<0.1

34 Table 3.4. Flows and Stocks of Private and National Wealth and National Saving (1) (2) (3) (4) Explanatory Variables Net National Wealth Net National Wealth Δ(Net National Wealth) Δ(Net National Wealth) Gross national saving 6.684*** 7.694*** [1.583] [1.402] [0.153] [0.148] GDP growth [8.233] [0.770] Inflation [2.245] [0.226] Ln(GDP per capita) 73.32*** * [24.34] [3.145] Rule of law * 4.967* [20.97] [2.614] Constant 138.0*** ** * [35.37] [230.3] [3.086] [29.87] Observations R-squared Robust standard errors in brackets *** p<0.01, ** p<0.05, * p<0.1

35 4. Households and Non- Firms: Gross Domestic Assets and Saving This section delves into the amount and composition of the financial assets owned by households and non-financial firms, as well as its relationship with the saving rate of these two sectors. It must be noted from the outset that the proper construction of this sort of data requires detailed financial accounts by the institutional sector that are unavailable for most LAC countries and emerging economies in general. In circumventing this limitation we have used the identities applied in previous sections along with other data and some assumptions to be made explicit as we move forward. We start from equations [1] and [2], which break down financial assets by instrument (currency, stock, public debt, private debt, and deposits) and by intermediary (banks and similar institutional investors and direct non-intermediated asset holdings), respectively. For the sake of clarity, both equations are reproduced below: Gross Private Domestic Assets = = Currency + Stock Market Capitalization + Outstanding Domestic Private Debt Securities + Outstanding Domestic Public Debt Securities + System Deposits [1] = System Deposits + Pension Fund Assets + Mutual Fund Assets + Insurance Companies Assets + Non-intermediated Assets [2] International and national financial statistics for the most part do not split financial assets by sector in emerging countries. 6 However, we have been able to make some progress on this front by resorting to other datasets (see Annex for details) and adopting a few assumptions, namely: a. The recently assembled IMF s Access Survey incorporates information on household deposits with commercial banks, as well as total deposits. The information covers about 30 countries in our sample for For years without data, the closest available figure was employed. For countries without data, we used the average on the respective year for the countries in the same group (LAC, high-income OECD, upper- and lower-middle- 6 Although that information indeed exists in most countries, it is not usually disclosed to the general public.

36 income). The rest of total deposits are assumed to be owned by institutional investors, when operating in the country, and non-financial firms. 7 Deposit holdings by pension funds come from portfolio compositions (as presented in the AIOS database for LAC and the OECD Global Pension Database for other economies). Based on availability, non-lac data corresponds to 2012 and to 2011 for LAC, repeating the same proportion for other years. 8 For insurance companies, the OECD Global Insurance Database was used for countries in that group. For other countries, given the lack or unreliability of other sources, we assumed that insurance companies allocate into deposits the same share of their total assets as pension funds. For all countries, the portfolio choices of mutual funds, on which no data is available, is supposed to be the same as that of pension funds. The deposits of non-financial firms are calculated as a residual. 9 b. Assuming that non-financial firms do not engage directly in securities holding and trading, and only maintain financial assets to cover for current and future liquidity needs in the form of bank deposits, we distribute securities (stock as well as public and private debt) between households and institutional investors. 10 In these cases our primary source is once again the portfolio structure of institutional investors and the allocation rules and assumptions are the same as those spelled out for deposits. c. Finally, we assume that cash (currency in circulation) is only demanded by households for transaction purposes. Table 4.1 shows gross private domestic financial assets to GDP by household, institutional investor, and non-financial firms. Table 4.2 reproduces the same figures as a share of the total. Over and above the sharp differences in total assets in terms of GDP, Table Because information is lacking on government and foreign deposits, we assume that these are zero. This assumption should not cause on average a major distortion in our conclusions for LAC countries and emerging economies in general, as we do not expect the government to own large bank deposits, except for some specific state-owned banks in some countries. Foreign deposits are also likely to be low, as these investors tend to choose other vehicles, such as direct investment or capital market securities. 8 The procedure to obtain the share of total deposits in the hands of pension funds consists of multiplying the share of deposits by total assets under management and then dividing by total commercial bank deposits. The same has been done with other institutional investors, data permitting. 9 This imputation of deposits by sector led in a few cases to inconsistent results, for instance, that the sum of household and institutional investor deposits which are estimated from different data sources exceeded total deposits. In those cases, encountered in a reduced number of emerging and developed countries, we kept the value for household deposits and assigned the rest of total deposits to the institutional investors, leaving firms deposits equal to zero. 10 Here we are assuming that banks and other intermediaries do not possess securities of this kind. This may not be a realistic assumption, as banks may own some securities, especially government bonds. Lack of data prevents us from deducting these holdings. In fact, the IMF s International Statistics report a bank balance sheet item labeled Claims on General Government, but it includes both loans and bonds net of government deposits. This global concept yields sometimes negative values and other times values that exceed the total public debt outstanding, making it a hardly dependable measure for our purposes.

37 suggests that households are the main owners of financial assets, accounting for a median 65.6 percent in LAC, comparable with 70.2 percent in the OECD, 77.3 percent in upper-middleincome countries and 83.2 percent in lower-middle-income countries. This pattern is common to all LAC countries except for Venezuela (41 percent). The institutional investor portfolio amounts to a median 11.1 percent in LAC, which pales compared to the 19.7 percent among OECD members, but surpasses a share that is below 5 percent in other middle income nations. As the portfolio of institutional investors ultimately belongs to households, the latter possess a median of 83 percent in LAC and even higher levels (between 87 percent and 92 percent) in the other groups. However, it must be remembered that most of the institutional investor activity in the LAC region is explained by mandatory, not voluntary, pension fund contributions. In turn, nonfinancial firms hold a median 17 percent in LAC, 8.2 percent in the OECD, and about 13 percent in the other country groups. From a related angle, Table 4.3 splits household assets by whether those funds are managed by deposit-taking intermediaries, institutional investors, or directly by the household, that is, not invested through an intermediary but consisting of purchases of stock, private and public debt, and the accumulation of cash balances. In LAC, 53.7 percent of household assets are not intermediated, 31 percent are managed by banks, and 15.2 percent by institutional investors. These proportions are relatively similar in the OECD (47.2 percent, 32 percent, and 20.8 percent). Direct investments are even more prevalent in upper-middle-income (63.3 percent) and lower-middle-income economies (71.6 percent). The distribution of the stock of each asset class (deposits, public and private debt, and stock) across sectors is highlighted in Table 4.4 to Table 4.7. According to the country group medians at the bottom of Table 4.4, 50.9 percent of deposits are household-owned, as opposed to 41.1 percent in the hands of non-financial firms, and just 4.5 percent included in the portfolio of institutional investors. This of course hides an ample variation across countries. For instance, in countries with a noticeable presence of pension funds, institutional investors display a higher participation in total deposits, peaking at 32 percent in Chile and 43 percent in El Salvador. In other cases, business deposits are well above the average, as in Colombia (72.1 percent), Venezuela (68 percent), and Panama (58.3 percent). This composition tilted towards households is also observed in the OECD (70.9 percent), and middle-income countries (around 60 percent), with a small share belonging to institutional investors (in all cases below 4 percent). Most of the stock market capitalization in LAC appears to be owned by households (92 percent, against just 8 percent for institutional investors). Similar proportions are found in middle-income countries, while institutional investors rise to 29.9 percent in the OECD. In the