Understanding Domestic Savings in Chile

Cerda, Rodrigo; Fuentes S. M., J. Rodrigo; García, Gonzalo; Llodrá, José Ignacio Working Paper Understanding Domestic Savings in Chile IDB Working Paper Series, No. IDB-WP-626 Provided in Cooperation with: Inter-American Development Bank (IDB), Washington, DC Suggested Citation: Cerda, Rodrigo; Fuentes S. M., J. Rodrigo; García, Gonzalo; Llodrá, José Ignacio (2015) : Understanding Domestic Savings in Chile, IDB Working Paper Series, No. IDB-WP-626, Inter- American Development Bank (IDB), Washington, DC, https://doi.org/10.18235/0000200 This Version is available at: https://hdl.handle.net/10419/146436 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc-nd/3.0/igo/legalcode Understanding Domestic Savings in Chile Rodrigo Cerda Rodrigo Fuentes Gonzalo García José Ignacio Llodrá IDB WORKING PAPER SERIES Nº IDB-WP-626 October 2015 Department of Research and Chief Economist Inter-American Development Bank October 2015 Understanding Domestic Savings in Chile Rodrigo Cerda* Rodrigo Fuentes** Gonzalo García** José Ignacio Llodrá*** * Pontificia Universidad Católica de Chile and Centro Latinoamericano de Políticas Económicas y Sociales, Universidad de Chile ** Pontificia Universidad Católica de Chile *** Centro Latinoamericano de Políticas Económicas y Sociales Cataloging-in-Publication data provided by the Inter-American Development Bank Felipe Herrera Library Understanding domestic savings in Chile / Rodrigo Cerda, Rodrigo Fuentes, Gonzalo García, José Ignacio Llodrá. p. cm. — (IDB Working Paper Series ; 626) Includes bibliographic references. 1. Saving and investment—Chile. I. Cerda, Rodrigo. II. Fuentes, Rodrigo. III. García, Gonzalo. IV. Llodrá, José Ignacio. V. Inter-American Development Bank. Department of Research and Chief Economist. V. Series. IDB-WP-626 Copyright © 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 (http://creativecommons.org/licenses/by-nc-nd/3.0/igo/ legalcode) and may be reproduced with attribution to the IDB and for any non-commercial purpose, as provided below. 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 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. Following a peer review process, and with previous written consent by the Inter-American Development Bank (IDB), a revised version of this work may also be reproduced in any academic journal, including those indexed by the American Economic Association's EconLit, provided that the IDB is credited and that the author(s) receive no income from the publication. Therefore, the restriction to receive income from such publication shall only extend to the publication's author(s). With regard to such restriction, in case of any inconsistency between the Creative Commons IGO 3.0 Attribution-NonCommercial-NoDerivatives license and these statements, the latter shall prevail. 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. http://www.iadb.org 2015 Abstract* This paper constructs time series data on savings per type of agent for Chile during the period 1960-2012. It is found that the economy’s average savings rate increased by 11 percentage points in the period 1985-2012 compared to 1960- 1984, with particularly pronounced growth in corporate savings. The evidence suggests that this increase was driven largely by the following measures: i) pension reform that introduced mandatory savings and private sector management, ii) banking reform, iii) tax reform, iv) capital markets reform and v) privatizations. JEL classifications: E21, N16 Keywords: Savings, Reform, Incentives, Chile * This paper was undertaken as part of the IDB Research Network Project on Understanding Domestic Saving in Latin America and the Caribbean. Cerda is Associate Professor, Institute of Economics, Pontificia Universidad Católica de Chile and Assistant Director of CLAPES UC. Email: [email protected]; Fuentes is Associate Professor, Institute of Economics, Pontificia Universidad Católica de Chile. Email: [email protected]; García is Associate Research, Institute of Economics, Pontificia Universidad Católica de Chile. Email:[email protected]; Llodrá is Associate Researcher, CLAPES UC. Email: jillodr[email protected]. 1 1. Introduction Funding is one of the main constraints on increasing investment in less developed economies. Most of these economies do not have full access to international capital markets and thus rely on their domestic savings capability. While evidence shows that saving rates are highly correlated with the level of per capita income, there might be policies that help increase domestic savings. Policies related to fiscal budget management, tax incentives to company savings and household savings are on the first line of possible policies to boost savings. However, the implementation of these policies might be offset by the agent’s reaction. As a matter of fact, substitution effects across agents might be important, i.e., tax incentives may raise private savings while reducing fiscal savings or a pension fund reform that increases mandatory savings may replace voluntary savings, etc. One of the main contributions of this paper is the construction of a time series data on savings per type of agent for Chile during the period 1960-2012. This updates previous studies by Bennett, Schmidt-Hebbel and Soto (1999), including changes in the methodology to adapt it to institutional changes. By using this new dataset, this paper illustrates the type of policies that could raise domestic savings in an emerging economy. In the evolution of savings rates in Chile, two important facts require an explanation. First, the economy increased the average savings rate by 11 percentage points in the period 1985- 2013 compared to 1960-1984 mainly due to a large change in the level (10 percentage points) of private savings, and an additional 1 percentage point from the public sector. While voluntary savings have been generally negative in Chile, they became even more negative after the 1981 pension reform that established a mandatory savings scheme. In addition, corporate savings increased significantly after the second half of 1980s. The replacement of voluntary savings by mandatory savings was established during the 1981 pension reform that changed from a pay-as- you-go system to an individually-funded system managed by the private sector and had an important effect on savings and investment (Corbo and Schmidt-Hebbel, 2003; Fuentes, 2013) The boost in the savings rate took place in 1985, a time when the economy was coming out of the largest financial and sovereign debt crisis since the Great Depression, characterized by the bankruptcy of several financial institutions and the intervention of many others, no access to international capital markets and low future growth expectations. At that time, Chile was implementing many important institutional reforms such as the private pension fund system, the 2 capital markets reform (1986), the reform of the banking legislation (1984) and an important tax reform (1984) that lowered corporate tax rate and avoided double income taxation, significantly simplifying the tax regime. In addition, there was a large wave of privatizations. In the following 12-years (1986-1997) the economy experienced its highest sustained growth in history. These reforms and sustained growth scenario may have had effects on household savings (part of the increase of growth was transitory) and on corporate savings (probably due to the increase in the marginal productivity of capital). The second fact to be explained is related to the change in composition of private savings. After several years of almost no corporate savings, this component became an important part of total savings, reaching an average of almost 10 percent of GNDI during the period 1986-2012. Several policies might have influenced the change in corporate savings, mainly the privatization process of the 1970s and 1980s of previously nationalized companies and changes to the tax system during the same period that provided incentives to corporate reinvestment. Our results show that the 1984 tax reform, the boost in the marginal productivity of capital and the deepening of the financial market were the main drivers of the dramatic increase in corporate savings. In addition, both mandatory savings and the fiscal system positively contributed, though to a lesser extent than the tax reform and the deepening of the financial market. We also found that tax incentives for retained earnings explained the changes in the composition of private savings. In fact, before 1984, there was double income taxation from corporate profits, with a large total burden (49 percent). The tax reform changed this system to a full imputation tax system that lowered corporate tax on accrued profits and delayed personal tax due to the distribution of profits to stockholders. Moreover, personal income tax was applied when dividends were received. The reform also cut personal tax rates. The paper proceeds as follows. Section 2 summarizes relevant literature to our case study. Section 3 describes a theoretical model that provides a setup to our empirical work. Section 4 focuses on methodological issues to build our macro saving data per institutional agent. Section 5 shows a glimpse of the data and documents the main facts related to our two questions. Section 6 presents the empirical analysis per type of agent using our macro data. In addition, we also include micro evidence on corporate savings due to their relevance in explaining the increase in savings since the mid-1980s. Using empirical evidence, Section 7 analyzes the policy implications of the results. Finally, Section 8 presents our conclusion. 3 2. Related Literature Several studies attempt to analyze the determinants of private savings at the international level. Table 1 shows the results obtained in these studies, classifying them according to determinants following Loayza, Schmidt-Hebbel and Servén (2000) and Grigoli, Herman and Schmidt-Hebbel (2014). Edwards (1996) examines why saving ratios have been so uneven across countries, using panel data for 36 countries from 1970 to 1992. He found that the main determinants are i) per capita income growth, which is one of the most important determinants of both private and public savings; ii) an incomplete crowding-out effect from public to private savings; iii) a negative effect of government-managed social security systems on private savings, iv) a positive effect of financial development on private savings; and v) a negative correlation between foreign and domestic savings. Dayal-Gulathi and Thimann (1997) examine the empirical determinants of private savings for several economies in Southeast Asia and Latin America over the period 1975- 1995. The results show that social security systems, macroeconomic stability and financial deepening have been important in accounting for differences in the behavior of private savings between the two regions. Using a panel of 150 countries for the period 1965-1994, Loayza, Schmidt-Hebbel and Servén (2000) find positive effects of income and macroeconomic stability on private savings, and negative effects of higher interest rates and larger private domestic credit flows. Also, they find a negative effect of the dependency ratio and a negative but incomplete compensation effect between public and private savings. In relation to the pension system, Attanasio and Rohwedder (2003) and Attanasio and Brugiavini (2003) analyze the cases of the United Kingdom and the Italian reforms, respectively. They find a substitution effect between pension wealth and private savings. In a recent study, Grigoli, Herman and Schmidt-Hebbel (2014), using a panel data for 165 countries from 1981 to 2012, find a positive effect of income and macroeconomic uncertainty on private savings, where uncertainty is associated with precautionary savings. In contrast, there was a negative effect on the dependency ratio and the real interest rate. Public savings had a negative but small effect on household savings, while corporate savings had a larger negative effect. 4 Table 1. Determinants of Private Savings Analyzed in International Empirical Studies Variable Category Specific Variable Expected Sign Empirical Findings Income Income level Income growth Transitory Income Terms of Trade Ambiguous Ambiguous (+) (0 or +) 0 (5, 6, 13, 14) ; + (1, 2, 3, 4, 7, 16, 17); 0 or + (9) 0 (16); + (7, 9, 12, 14, 15, 17) 0 (7) 0 (14, 16, 17); + (2, 4, 6, 7, 12, 13) Rate of return Real interest rate Ambiguous - (7, 12); 0 (1, 3, 5, 6) ; + (2, 15, 17); 0 or + (16) Uncertainty Inflation, other measures of macroeconomic instability (+) - (4) ; 0 (1,2,3, 6, 12, 13, 16); + (7, 14, 17) Domestic borrowing constraints Private credit flows, broad money flows (-) - (7); + (3); + o - (15) Foreign borrowing constrains Current account deficit Foreign savings Capital flows restrictions (-) (-) (+) - (1, 2) - (14) 0 (7) Financial depth Bank credit stock Broad money stock Ambiguous Ambiguous - (5); 0 (7) 0 (7, 16); + (1, 3, 4, 13, 14) Demographics Old-age dependency Young-age dependency Urbanization (-) (-) Ambiguous - (2, 3, 4, 7, 12, 15, 16, 17) ; 0 (5, 6, 14) - (7, 15, 16); 0 (9) - (3, 16, 17) Income distribution Income concentration Ambiguous - (14); 0 (3, 9) Fiscal Policy Public sector saving Public sector budget balance Public consumption (-) (-) Ambiguous - (1, 3, 7, 12, 13, 16, 17) - (2, 5, 6, 8, 14) ; 0 (4) - (2, 6); 0 (8) Pension System Pay-as-you-go pension transfers to old Mandatory fully-funded pension system contributions Ambiguous (+) - (3, 4, 5, 10, 11) + (4) Households and firms Corporate saving effect on household saving 0 or (-) - (16, 17) Studies: 1. Corbo and Schmidt-Hebbel (1991); 2. Masson, Bayoumi and Samiei (1995); 3. Edwards (1996); 4. Dayal-Gulathi and Thimann (1997); 5. Bailliu and Reisen (1998); 6. Haque, Pesaran and Sharma (1996); 7. Loayza, Schimdt-Hebbel and Servén (2000); 8. López, Schimdt- Hebbel and Servén (2000); 9. Schimdt- Hebbel and Servén (2000); 10. Attanasio and Rohwedder (2003); 11. Attanasio and Brugiavini (2003); 12. De Serres and Pelgrin (2003); 13. Agénor and Aizenman (2004); 14. Gutiérrez (2007); 15. Horioka and Terada-Hagiwara (2012); 16. Bebczuk and Cavallo (2014); 17. Grigoli, Herman and Schmidt-Hebbel (2014). Research on trends and determinants of national savings for Chile can be divided into macroeconomic and microeconomic approaches. The main studies for Chile are summarized in Table 2. The first group of information related to Chile is a set of studies that built a “Chilean savings history.” This is the case of Morandé (1998), Hachette (1998), Agosin, Crespi and Letelier (1997), Agosin (2001) and Bennett, Schmidt-Hebbel and Soto (1999). Agosin, Crespi and Letelier (1997) found a partial compensation effect between voluntary and mandatory household savings, between household and company savings, and between household and public savings. According to Morandé (1998), foreign savings were relevant for the evolution of private savings in the mid-1980s, while thereafter national savings were supported by domestic savings. 11 [bt+1]: 0 = −λt+λt+1�1 + rt+1�1−τt+1 k�� (5) [ct]: 0 = βtuc(ct)−λt (6) [ct+1]: 0 = βt+1uc(ct+1)−λt+1 (7) Using equations (4) and (5), we get: (1−τt+1 c)Yk(t + 1)=υt+1 (8) where υt+1≡�1+rt+1�1−τt+1 k���1−τtd�ptI−�1−τt+1 d�(1−δ)pt+1 I �1−τt+1 d� is the user cost, similar to the formulation in Hall and Jorgenson (1967). Equation (8) states the consumer chooses how much to save through capital stock accumulation by equalizing the marginal product of capital with its user cost. If we assume a Cobb-Douglas production function, where α is the capital share, corporate savings as fraction of output can be written as in:1 Stc Yt=�(1+γ) vt+1 (1−τt+1 c) �1−τtc�−(1−δ)α �1−τtc�Yk(t)�ptI= (9) where γ is the growth rate of output. Equation (9) indicates that corporate savings increases the larger the aftertax marginal product of capital, while it decreases with the user cost of capital. From the household problem and assuming β�1 + rt�1−τtk��= 1 , we obtain the following expression for consumption: ct= r0�1−τtk�b0 P+r 1+r�∑wjL�1−τjp�+dj�1−τjd�+Mj �1+rj�1−τjk��j ∞ j=0 � (10) Thus consumption corresponds to a fraction of permanent income –defined as the present value of future income streams. 1 The marginal product of capital is 𝑌𝑘=𝛼�𝐾𝑡 𝑌𝑡�−1/𝜎, with substitution elasticity (𝜎) equal to one in a Cobb-Douglas production function, in addition equation (2) can be written as in 𝑆𝑡𝑐 𝑌𝑡=𝑝𝑡𝐼�(1+𝛾)𝐾𝑡+1 𝑌𝑡+1−(1 −𝛿)𝐾𝑡 𝑌𝑡� 12 Savings within the household are:2 Sth Yt= 1 + rt�btp+btg+Mt� Yt−Stc Yt−Stg Yt−ct Yt−Gt Yt (11) where Gt≡�gt+ rt�1−τtk�btg�. Equation (11) is not surprising. Note that if we add up corporate and government savings to its left-hand side, we obtain total savings (as fraction of national income), which are a function of consumption and government expenditure decisions. 4. Data Construction: Methodological Issues 4.1 Public Savings The public sector in Chile is usually classified into two main branches: the financial public sector and the non-financial public sector. The financial public sector includes the Central Bank and Banco Estado, one of the largest banks in the banking system. In fact, in 2013 it had 30.5 percent of total credits financing tertiary education and 20 percent of total mortgage loans (BancoEstado, 2013). The non-financial public sector has two main branches: i) the general government and ii) public firms. The general government consists of the central government, which includes ministries, armed forces and civil services that depend on the government such as the Budget Office (Spanish acronym: Dirección de Presupuestos, Dipres),the Internal Revenue Service (Spanish acronym: Servicio de Impuestos Internos, SII) or the National Custom Services (Spanish acronym: Servicio Nacional de Aduanas, SNA). The central government includes regional governments, public universities and local councils. In addition, there are currently 33 public firms in Chile according to information provided by the Ministry of Finance (Dipres3) that employed 49,455 workers in 2013. Their before-tax profits were 1.2 percent of GDP in 2013. Figure 1 shows the way the public sector is organized in Chile. 2 We obtain the result by using the definition of household savings: 𝑆𝑡ℎ≡�𝑏𝑡+1 𝑝+𝑀𝑡+1�−�𝑏𝑡𝑝+𝑀𝑡�=𝑤𝑡𝑙𝑡�1−𝜏𝑡𝑝�+(1−𝜏𝑡𝑑)𝑑𝑡+𝑟𝑡(1−𝜏𝑡𝑘)�𝑏𝑡𝑝+𝑀𝑡�−𝑐𝑡 And we replace the definitions of government and corporate saving to finally obtain the result in (11). The definitions of government and corporate savings are: 𝑆𝑡𝑔≡𝑏𝑡+1 𝑔−𝑏𝑡𝑔=�𝑔𝑡+𝑟𝑡(1−𝜏𝑡𝑘)𝑏𝑡𝑔�−𝜏𝑡𝑐(𝑌(𝑘𝑡,𝑙𝑡)−𝑤𝑡𝑙𝑡)−𝜏𝑡𝑙𝑤𝑡𝑙𝑡−𝜏𝑡𝑘𝑟𝑡�𝑏𝑡𝑝+𝑀𝑡�−𝜏𝑡𝑑𝑑𝑡 𝑆𝑡𝑐=(1−𝜏𝑡𝑐)(𝑌(𝑘𝑡,𝑙𝑡)−𝑤𝑡𝑙𝑡)−𝑑𝑡 3 http://www.dipres.gob.cl/596/articles-64217_recurso_1.pdf 13 Figure 1. Public Sector in Chile Public savings will be calculated as the sum of the savings on the three big branches of the public sector (general government, public firms and central bank4). We next explain how we calculate savings in each of those three branches. 4.1.1 General Government Savings General government savings include those from the central government (ministries, armed forces, and civilian dependent institutions), regional government, local governments and public universities. This definition is the same used by Bennett, Schmidt-Hebbel and Soto (1999) who provide savings estimates from 1960 to 1997. Similarly, the same definition is available in the National Accounts by institutional sector compiled by the Central Bank of Chile from 1996 to 4 Next, we include BancoEstado as part of public firms. 14 2013. Thus we use data from Bennett, Schmidt-Hebbel and Soto (1999) and complement it with data since 1998 data published by the Central Bank of Chile. 4.1.2 Public Firms’ Savings According to Dipres, Chile had 33 public firms in 2013. Banco Estado is the single firm in the financial sector. Codelco, which produces 32 percent of Chilean copper production, accounted for 85 percent of total profits of public companies, while BancoEstado and ENAP accounted for 11.8 percent and 6.5 percent of total profits, respectively. The remaining public companies have negative profits, accounting for -0.04 percent of GDP. To obtain savings series from public firms, we used data from Dipres’ Annual Report on Public Finances (Spanish acronym: Informe de Estadísticas de las Finanzas Públicas). The report provided data related to before-tax profits 𝜋𝑡𝑃𝑃, depreciation 𝐷𝐷𝑝𝑡𝑃𝑃, transfers from the central government to public firms (𝑇𝑟𝑡𝐺,𝑃𝑃 ), and transfers from public firms to the central government (𝑇𝑟𝑡𝑃𝑃,𝐺). We were able to obtain data back to 1994 and we constructed public firm savings as non-distributed after tax profits plus depreciation and transfers from the government minus transfers to the government. Thus, public firms’ savings is calculated as in: 𝑆𝑡𝑃𝑃=𝜋𝑡𝑃𝑃+𝐷𝐷𝑝𝑡𝑃𝑃+𝑇𝑟𝑡𝐺,𝑃𝑃−𝑇𝑟𝑡𝑃𝑃,𝐺 (12) From 1960 to 1993, we used the data from Bennett, Schmidt-Hebbel and Soto (1999). That data followed similar methodology. 4.1.2 Central Bank Savings To construct Central Bank savings we obtained data from 1997 to 2013 from the Annual Financial Statements of the Central Bank. The data represent profits 𝜋tCB and depreciation 𝐷𝐷𝑝𝑡𝐶𝐶. In addition, we obtained data on capital contributions provided by the Ministry of Finance to the Central Bank, 𝑇𝑟𝑡𝐺,𝐶𝐶 , which were authorized by Law 20,128, approved in 2006. That legislation allowed a capital contribution up to 0.5 percent per year for five years. Those capital contributions effectively occurred in 2006, 2007, 2008 and 2009. Using that information, we constructed Central Bank savings according to: 𝑆𝑡𝐶𝐶=𝜋𝑡𝐶𝐶+𝐷𝐷𝑝𝑡𝐶𝐶+𝑇𝑟𝑡𝐺,𝐶𝐶 (13) 15 The savings series was completed with data from Bennett, Schmidt-Hebbel and Soto (1999) from 1960 to 1997. 4.2 Corporate Savings Corporate savings are basically calculated as non-distributed after-tax profits plus depreciation minus the profits of foreign companies: Corporate Savings = After-tax corporate profits – Distributed dividends + Private firms depreciation – Foreign firms profits. To estimate after-tax corporate profits we start by obtaining data from publicly traded private firms (Spanish acronym: Sociedades Anónimas Abiertas, SAA). We have different sources of information on SAA. First, we were able to obtain the public Standardized Quarterly Financial Reports (Spanish acronym: FECUs) of those firms. These are financial statements sent by publicly held firms to the Chilean Stocks and Securites regulator (Spanish acronym: Superintendencia de Valores y Seguros, SVS). There are two problems with these data. First, we were only able to obtain data from 1985 to 2007. Second, the data on dividends in the FECUSs are available only beginning in 2001. On the other hand, we have at our disposal the Economatica database,5 a system containing information on companies listed on the stock exchange market of the United States, Brazil, Argentina, Chile, Mexico, Peru and Colombia. The database provides many years of historical data on Quarterly Financial Statements, stock prices, corporate activities (dividends, splits, etc.) and most important shareholders. For Chile, Economatica has SAA data on stock prices, profits from 1990 to 2013 and dividends from 1997 to 2013. We also used an alternative source of information to construct the SAA dividends, 𝑑𝑑𝑑tSAA, gathering information from the SVS Monthly Bulletin from 1985 to 2000 at the SVS library. Between 1990 and 2007, SAA total profits data from the FECUs differ by only 0.3 percent per year, on average, from Economatica data. Similarly, the data on dividends from Economatica are consistent with data from FECUs or from the SVS Monthly Bulletin. 5 See https://economatica.com/en_anac_base-de-dados.html 16 For the remaining private companies, we estimate total corporate profits from the aggregate corporate tax revenues from Dipres reports. For dividends and withdrawals from the remaining companies, we used SII data beginning in 2006; this information was obtained through the provisions of the Transparency Act6). We also asked for the depreciation of privately owned companies which we obtain from Henríquez (2008), plus information from Dipres reports. Finally, we obtained information on foreign profits from capital accounts published by the Central Bank. The SAA represent a small fraction of firms in the economy (on average, almost 763 different firms on our sample. According to the SII, in 2013 the total number of private companies was 1,014,482, which includes firms paying corporate taxes plus individuals engaged in economic activities.7 While the number of SAA firms is small, the SAA’ s impact is large. In fact, according to the SII, 0.2 percent of companies with the largest sales in 2013 (mainly SAA), accounted for 68.4 percent of total sales in the economy and 25.6 percent of total employed workers, and they paid 40.8 percent of total workers’ compensation in that year. See Table 3 below. Table 3. Sales, Workers and Worker Compensation, 2013 Sales Number of firms Sales (US$, millions) Number of dependent workers Worker compensation (US$, millions) Large Firms 2.006 613,782 2,226,224 30,803 Total 1,014,482 897,372 8,709,993 75,494 Large firms as fraction of total firms 68.4% 25.6% 40.8% 4.3 Household Savings Household savings are calculated by the difference between total private savings and corporate savings. In turn, household savings can be divided into mandatory savings and voluntary savings. 6 Ley de Transparencia (Ley No. 20.285). 7 Individuals are owner of small firm and they use their personal national identification number in the SII to identify their firms. 17 4.3.1 Mandatory Savings Since 1981, the Chilean Pension System has been based on individual capitalization. Each member has an account in which his/her social security contributions are deposited. These are capitalized and earn the yield on the investments made by the administrators with the resources from the funds. When an individual retires, the fund is used to buy a pension, generally an annuity. If the fund owner dies, his/her surviving beneficiaries receive a pension. The amount of the pensions will depend on the amount of savings, thus there is a direct relation with personal effort, as well as with the return on the financial investment and the pension obtained. The Pension System is managed by private institutions known as Pension Fund Administrators (AFPs). As in the methodology proposed by Bennett, Schmidt-Hebbel and Soto (1999), mandatory savings are constructed as follows: Mandatory Savings = Payment of Mandatory Contributions + Other Increases – Management Fee – Total Benefits Paid – Other reductions + Assets Return • Payment of Mandatory Contributions corresponds to the mandatory contribution made by the households to the individual pension fund as a fixed share of its labor income.8 • Other Increases include Compensation Saving Accounts9 (CSA) and Additional Contributions.10 • Management Fee is the fee paid to the Pension Fund Manager. • Total Benefit Paid is the payment received by the retiree from the individual pension fund. There are two possible retirement arrangements: Programmed Withdrawal, agreed with the Pension Fund Manager, and Life Annuity, contracted with an insurance company. 8 It is a rate of 10 percent of labor income plus a fee for the fund manager. 9 November 1990 marked the beginning of compensation savings accounts (CSA) in order to give domestic workers compensation benefits in case the work relationship ended. 10 It is the contribution made by insurance companies to pension fund accounts for those workers who are declared impaired or deceased, and who are entitled to Disability Insurance. This contribution is equivalent to the amount resulting from the difference between the resources needed to finance the pension of reference and the amount accumulated in the pension fund account of the affiliate at the date of death or invalidity. If the difference is negative, the additional contribution will be zero. 18 • Other Reductions include withdrawals from the Compensation Saving Accounts and other obligatory decreases informed by the Pension Superintendent. • Assets Returns includes the dividends and interests gained by the assets that form the pension fund but exclude capital gains from changes in asset pricing. 4.3.2 Voluntary Savings Voluntary savings are estimated as a residual. They represent the difference between household savings and mandatory savings. 4.4 Other Variables There are some explanatory variables that are not readily available and it was necessary to obtain. The three most relevant are taxes, marginal productivity of capital and the interest rate. We now describe the methodology used to construct these variables. 4.4.1 Taxes The tax variables considered in this paper are the personal income tax, the reinvestment tax and the dividend tax. We built series of taxes from 1960 to 2012 with the information of the tax code from the National Library of Congress (acronym: BCN Biblioteca del Congreso Nacional). The personal income tax is the tax levied on personal earnings including labor and capital income. In Chile, it is the “Impuesto Global Complementario” (IGC) that taxes the entire personal base income and is the only progressive tax in Chile. Generally, other taxes are flat rates. We used the top marginal rate of the IGC to construct a time series of the personal income tax. These are available from the tax code, which has changed many times since 1960. By contrast, the reinvestment tax rate is built from the different tax levied on corporate profits but it corresponds to the tax rate the firm pays when it retains $1 of profits. Currently Chile has a unique corporate tax (the so-called first category tax or “impuesto de primera categoría”) but throughout history there were at least three additional taxes on non-distributed profits in different periods (from 1960-1963, 1968-1969 and 1975-1985). The dividend tax variable represents taxes on dividend income. Before the 1984 tax reform, corporate taxation was independent from the personal income tax, so the dividend tax was calculated similarly to a classical taxation system, i.e., as an additional tax on the cash flow 19 at the personal level. The 1984 tax reform allowed using the corporate tax paid as a withholding tax for the personal income tax. This is similar to a full imputation taxation system. Since 1984, the dividend tax corresponds to the IGC. 4.4.2 Marginal Productivity of Capital We estimated a proxy for the marginal productivity of capital by using an estimation of the capital stock corrected by the unemployment rate to capture the intensity of use over the business cycle. This series is published by Dipres. Assuming a Cobb-Douglas production function, our estimation of the marginal productivity of capital (MPK) is given by: K Y MPK α = where α represents the capital share that we assume equal to 0.4, Y is GDP and K is the capital stock corrected by the business cycle. 4.4.3 Interest Rate The real interest rate was built as the difference between the annual nominal interest rate of banking operations and the inflation reported on the period, divided by (1+inflation). Nominal interest rates series for the period 1960-1985 are those published by Chile’s Central Bank entitled “Interés Corriente para Operaciones no Reajustables” (regular interest for nonreadjustable operations), while for 1986-2012 the deposit interest rate for operations shorter than 90 days were used. 4.4.4 Financial Intermediate Ratio (FIR) The financial intermediation ratio (FIR) series is constructed as the sum of total deposits, mortgage liabilities, internal public debt, corporate bonds, public bonds and market capitalization. For the period 1960-2000 the main source of information is Díaz, Lüders and Wagner (2010). For the period 2001-2012, the series are updated using data from Chile’s Central Bank, the Superintendency of Banks and Financial Institutions, the Santiago Stock Exchange and Dipres. 20 5. A First Glance at the Data The savings rate in Chile fluctuated between 10 percent and 15 percent of national income during the 1960s and 1970s. Savings decreased to levels lower than 5 percent during the crisis of 1982, but as of 1984 national savings recovered, reached close to 25 percent of national income in 1989. Savings again declined to close to 22 percent after the 1998 Asian crisis but rebounded again to reach 28 percent in 2006 and 2007. Savings have subsequently fluctuated at around 24 percent. Figure 2 illustrates the evolution of savings in Chile since 1960. We see the evolution of both private and government savings. The large increase in the national savings rate during the 1980s is called the “saving miracle” according to some authors (see Agosin, Crespi and Letelier, 1997 and Agosin, 2001). As it can be seen, the “saving miracle” was due to a large and sustained increase in private savings. In addition, Chile had also a temporary jump in the saving rate in the mid-2000s, which was the result of an increase in public savings. Figure 2. National Savings Divided into Public and Private Savings as a Percentage of Disposable Income -.10 -.05 .00 .05 .10 .15 .20 .25 .30 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Public Savings (% GN5I) Private Savings (% GN5I) A Quandt-Andrews test shows a structural break for the mean of the national saving rate in 1987, while in the case of private savings the break took place in 1986. In those periods, the average rate of public savings increased by 1.7 percentage points, while the private saving rate 27 Figure 7. Fiscal Mining Revenue, Percent of GDP The study of the impact of the structural fiscal rule in savings is still unexplored. In this paper we provide some estimates of that impact. In addition, and following Bennett, Loayza and Schmidt-Hebbel (2001), who studied the compensation coefficients among all the components of private savings in Chile (household voluntary saving, firm savings and household mandatory savings), we will revisit this analysis with a longer time series. We will also study the compensation coefficient between public and private savings in the empirical analysis. 0 1 2 3 4 5 6 7 8 9 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 28 6. Empirical Results This section presents the result by type of agents: public and private savings, dividing the latter into corporate savings and voluntary savings. 6.1 Public Savings Public savings represent general government savings and thus include savings from the central government, public firms and the Central Bank. They are the sum of government revenues minus government expenditures, excluding public investment. To model public savings we include the following determinants: i) variables that allow us to distinguish between permanent and transitory fiscal revenues, ii) variables that might determine fiscal expenditures, iii) variables related to fiscal institutions and political regimes and iv) the real interest rate, in order to capture potential substitution and income effects. Fiscal revenues depend on general economic activity and mining. As a matter of fact, fiscal revenues were 23.7 percent of GDP in 2013, and mining fiscal revenues accounted for almost 2 percent of GDP. The rest of fiscal revenues are related to economic activity. To distinguish between permanent and transitory shocks, we calculate Hodrick-Prescott (HP) filters on two series: i) the copper price and ii) the level of real GDP. To obtain a proxy for a transitory shock on the copper price and on the GDP, we calculate deviations from the Hodrick-Prescott filters. We define copper price as the product of the international value of copper and the nominal exchange rate (peso/dollar), and we then deflated by the CPI to obtain a measure of the real copper price measured in Chilean pesos. Column (1) of Table 7 reports the results when we estimate public savings as a function of i) copper price deviations from trend and ii) GDP deviations from trend, variables that measure transitory shocks to the copper price and the GDP level. In addition, in column (1) we include the HP filters on both the copper price and the real GDP to provide a measure of permanent shock. We also include the lag of the dependent variable to capture potential dynamics and a variable related to the structural balance rule implemented since 2001. In that variable, we have the effective structural stance as a share of GDP. As explained above, the use of the structural rule allows the fiscal sector to save during economic expansions and dissave during economic contractions. Transitory shocks to the price of copper have positive and significant coefficients, while the coefficient on permanent shocks is non-significant and similar to the real GDP transitory component. Those results indicate that the 29 fiscal sector increases its expenditure when there is a permanent revenue increase in a similar magnitude to the permanent shock. In contrast, when there is a positive (negative) transitory copper price shock, the fiscal sector tends to save (dissave) the rainfall. The coefficient on the structural balance stance is positive and significant, meaning that a fiscal rule skewed to a structural surplus requires larger savings. As the fiscal structural stance is measured in points of GDP, the estimates indicate that a 1-point increase in the structural balance is associated with 1.2 additional percentage points of fiscal savings in the short run and 2.2 percentage points in the long run. Column 2 includes variables that could influence fiscal expenditure such as demographics (rural population, old age dependency rates), economic conditions (lagged unemployment rate that might incentivize larger fiscal expenditure programs), the real interest rate (to capture substitution effects or income effects in case the government is net creditor or net debtor) and dummy variables for the different presidential periods. As we include additional controls, both the transitory shock to real GDP and to the price of copper price are positive and significant, indicating that a fraction of the positive (negative) income shock is saved (dissaved). While the coefficient on GDP trend is non-significant, the coefficient on price of copper trend is positive and significant, indicating that a share of long-lasting (positive) copper price shocks is also saved. While we should have expected this last coefficient to be non-significant, the result might be due to our imperfect measure of the long-run copper price trend. As a matter of fact, since 2001 the government has convened an independent committee of economists and copper market experts who annually provide an estimate of the long-run copper price that is used to estimate fiscal revenues and determine fiscal expenditure. Figure 8 plots long-run cooper price estimates from the copper committee and from a Hodrick-Prescott filter. Price estimates from the copper committee are always lower than the estimate from the Hodrick-Prescott filter, indicating that decisions on the budget were based on a more conservative estimate of the long-run copper price than the one used in our regressions. Thus our estimate on the long-run copper price might be capturing at least part of a transitory shock. The interest rate has a positive impact on fiscal savings in line with the usual substitution effect, but it is non-significant since the structural balance rule still has a positive effect on savings. The coefficient on rural population is significant and positive, as generally more fiscal expenditure is needed in urban centers. Unemployment also has a non-significant coefficient. 30 This is not really surprising since, beginning in 2002, Chile has implemented an unemployment insurance scheme, and the scope of government programs related to unemployment is generally limited (in the case of the 1983 crisis, temporary job programs were created and later discontinued). Finally, the dependency ratio shows a negative and significant coefficient, as a country experiencing a demographic transition, with an increasing share of the elderly, has lower tax revenues but higher expenditure on the elderly. The coefficient on the lagged dependent variable becomes small and statistically non-significant. In column 3 we do not include non-significant variables and obtain our final model on public savings. The interest rate is significant and positive, and the remaining determinants that increase public savings are transitory shocks on both the real GDP and copper price, as well as the long-run price of copper, the structural balance and the rural population. The only determinant with a negative impact on public savings is the dependency rate. 31 Table 7. Public Savings Dependent Variable: Public Savings 1961 – 2012 (1) OLS (2) OLS (3) OLS Constant 0.049 -5.090 -1.529** (0.18) (-2.69) (-2.74) Public Savings (t-1) 0.476** 0.052 (4.57) (0.40) (Copper price - HP Copper price)/HP Copper price 0.102** 0.035* 0.053** (4.91) (1.85) (3.21) Ln(HP Copper price) -0.014 0.137** 0.094** (-0.63) (3.46) (3.18) (GDP - HP GDP)/HP GDP 0.126 0.415** 0.346** (1.47) (5.96) (6.53) ln(HP GDP) 0.013 0.148 (1.21) (1.69) Structural Balance Target 0.013* 0.022** 0.022** (1.86) (4.55) (4.53) Interest rate (90 to 360 days) 0.0004 0.0007** (1.120) (3.41) Rural Population 0.020** 0.004** (2.74) (2.02) Unemployment (t-1) 0.129 (0.77) Dependency rate -0.065** -0,039** (-2.36) (-2.24) R squared 0,578 903 0,876 Breusch-Godfrey Serial Correlation LM Test 0,44 0,61 0,68 Heteroskedasticity Test: ARCH 0,04 0,64 0,56 Normality test 0,00 0,49 0,9 Observations 51 51 51 Note: Significance at levels of 5 %(**) and 10%(*). T-statistic in parenthesis. a, b and c: p-values are reported. We also include dummy variables for presidential periods. 32 Figure 8. Long-Run Copper Price Estimates, USD$ 6.2. Private savings Private savings comprise corporate savings and household savings; the latter includes both mandatory and voluntary savings. In this section, we use as reference the conceptual framework presented in Section 3. We will proceed with the analysis of savings by each private agent: corporate savings and household savings. Later, we study total private savings behavior. 6.2.1 Corporate Savings As discussed in Section 4, corporate saving is mainly defined as retained earnings plus depreciation. This is basically the part of cash flows used to re-invest in the company. In other words, company owners have the alternative of i) receiving dividends and withdrawing them on their own or ii) saving through the company. In our conceptual framework, firms maximize the present value of cash flows, obtaining the condition that marginal productivity of capital after tax is equal to the cost of use of capital corrected by the corresponding tax rates. It is almost 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 US$ Copper Committe Hodrick-Prescott 33 impossible to construct a long time series of cost of capital for an economy like Chile, since there are no good statistics on the components of the cost of capital. Therefore, we will take the advantage of the first order condition of the maximization process and we will use the marginal productivity of capital net of corporate taxes. Table 8 shows the results of the regression for corporate savings. The first column presents a modified version of equation (9), where the after-tax marginal product of capital is divided into two components: i) the marginal product of capital and ii) the reinvestment tax rate. None of these variables appear to be statistically significant. In the following columns the corporate savings in t depend on the after corporate-tax marginal productivity of capital (ATMPK) in t and t+1, which is consistent with the conceptual framework. As expected, the contemporaneous ATMPK is not statistically significant in any of the specification, while the one period ahead is not statistically significant in the simplest specification. When we drop the contemporaneous ATMPK then the ATMPK in t+1 becomes positive and statistically significant. Columns 2 through 4 show the strength of the result when controlling other variables. The financial intermediation ratio is positive and statistically significant. This variable captures the financial market deepening, ensuring that firms can have the option of separating investment decision from financing and dividends decisions. Other variables such as the personal income tax and the growth rate of the economy are not statistically significant. Furthermore, the inclusion of these variables does not change the effect of ATMPK on corporate savings. These results are very important in order to explain the large increase in the investment rate in Chile financed with corporate savings. The Chilean economy experiences a process of rapid growth since the mid-1980s, which is shown in the large increase in the ATMPK. In addition, in 1986, there is a structural change in the tax scheme that includes a reduction in the corporate tax rate plus the fact that reinvested profits do not pay the personal tax of the stockholders as was previously the case. That change in the tax system also affects considerably the ATMPK evolution. Finally, starting in the late 1970s there is a continuous development of the financial market, as reflected in access to loans and in a more active stock market due to a massive privatization of state-owned companies, the implementation of the private pension system and the development of the insurance market, among many other structural changes that took place at the time. 34 Table 8. Corporate Savings Dependent Variable: Corporate Savings 1961 – 2012 (1) OLS (2) OLS (3) OLS (4) OLS (5) OLS Constant -0.006 0.021 0.013 0.020 (-0.40) (1.09) (0.21) (1.09) Marginal productivity of capital -0.038 (-0.05) Reinvestment tax rate -0.014 (-0.16) Marginal productivity of capital (t+1) 0.470 (0.62) Reinvestment tax rate (t+1) -0.070 (-0.82) Marginal productivity of capital net of tax 0.021 -0.074 (0.04) (-0.16) Marginal productivity of capital net of tax (t+1) 0.538 0.428 0.382** 0.366** (1.24) (1.02) (1.78) (2.01) Financial intermediate ratio 0.017** 0.018** 0.017** (2.34) (2.81) (2.36) Personal Income Tax 0.008 (0.12) Income per capita growth rate -0.010 (-0.12) Corporate savings (t-1) 0.513** 0.374** 0.369** 0.367** (4.14) (2.75) (2.78) (2.86) R squared 0.684 0.710 0.710 0.710 Breusch-Godfrey Serial Correlation LM Testa 0.325 0.241 0.364 0.338 Heteroskedasticity Test: ARCHb 0.907 0.584 0.572 0.545 Normality Test: Jarque-Bera 0.506 0.789 0.787 0.782 Note: Significance at levels of 5%(**) and 10%(*). T-statistic in parenthesis. a, b and c: p-values are reported. 35 6.2.2 Household and Voluntary Savings Estimating an equation for household savings is a challenge because, while this variable is measured as the sum of voluntary plus mandatory savings, we do not have a direct measurement of the former. Voluntary saving is only a residual calculated as total private savings minus corporate savings minus mandatory savings, which may be considered noisy. We estimate an equation for total household savings using our conceptual framework; later we also estimate an equation for voluntary savings and show the validity of estimating an equation for the household as a whole. In our framework, household savings are a function of corporate savings, fiscal savings, transitory and permanent income and taxes. Besides the other two types of savings, we include more traditional determinants such as the growth rate of per capita income (to control transitory savings), real interest rates, dependence rate, inflation (measure of uncertainty) and unemployment (transitory shocks). Given that a household may decide whether to save in their own company or in another instrument, a problem of endogeneity is expected when estimating this equation. We used twostage least squares to estimate the equation of household saving, applying to corporate savings those variables used in the previous section as instruments. Table 9 presents the results for household savings. The first column shows the general equation estimated using IV. Corporate savings has a coefficient equal to -0.3, although it is not statistically significant at the classical level. On the other hand, government savings has a coefficient equal to -0.74 and it is statistically different than zero. The personal income tax has a negative coefficient but not statistically significant, while per capita growth rate (GDP12) is positive as expected. The coefficient on the real interest rate is negative, similar to what Loayza, Schmidt-Hebbel and Servén (2000) obtained. This is confusing since voluntary savings are negative, which means that households are net debtors and therefore the substitution and the wealth effect of an interest rate increase will reduce everyday consumption or, similarly, will increase savings. It is important to consider that the interest rate coefficient is statistically significant but very small. 12 We tried to estimate transitory income using the cyclical component of the HP filter, but the variable did not show a statistically significant effect. 36 In column 2 we provide OLS estimates, which should be consistent as we exclude corporate savings (which was not significant and was the source of potential endogeneity in column 1). The results are similar to the 2SLS, with some minor changes. The compensation coefficient of public saving is smaller than in the previous regression. Personal income tax has a negative coefficient but is now statistically different from zero. The interest rate enters with a negative sign with an even lower t-value, remaining non-significant. The interest rate was regulated during the 1960s and part of the 1970s. Thus, it is difficult to interpret financial transactions as a result of funds demand and supply in what was a competitive market at the time. In the mid-1970s there were concrete serious attempts to liberalize it, which could explain the negative coefficient on the interest rate in the regression. Therefore the third and the fourth columns of Table 9 show the same regressions as in the previous two columns, but using a shorter sample of 1977-2012. The results change in some important aspects. The conclusions regarding corporate and government savings are not different than the previous analysis, nor are they for the growth rate of per capita GDP. However, the magnitude and level of statistical significance of the personal income tax and the sign for the interest rate do change. We now obtain a positive sign for the interest rate, although it is not statistically significant. One could argue that household savings is hidden under the effect of voluntary savings since it is the sum of voluntary and mandatory savings. The last two columns of Table 9 show the same regressions, this time using voluntary savings as a dependent variable and mandatory savings as control, for the period 1977-2012. The results are almost exactly the same as in the previous two columns. The only important difference is the positive coefficient for the interest rate, which becomes statistically significant in the OLS estimation. The hypothesis that the coefficient of mandatory savings is equal to -1 cannot be rejected; this means that it is valid to estimate an equation for household savings. In summary, as expected, government saving and personal income tax negatively affects household savings, and transitory income positively affects household savings. Corporate savings negatively affects household savings, but the coefficient is not statistically significant when properly estimated by using IV. The interest rate enters with a positive coefficient in the equation, when we use a period that excludes interest rate rigidities. 43 explanatory variables in each period we calculate the contribution of each determinant to the increase in corporate savings. Table 13 shows these calculations. The Financial Intermediate Ratio as a percentage of GDP rose from 20 percent to 169 percent, increasing corporate savings by 4 percentage points. On the other hand, the increase of 6 percentage points in marginal productivity of capital net of tax (t+1) meant an expansion of 3.4 percentage points in corporate savings. This upsurge of the marginal productivity net of tax is explained by an increase of the MPK from 15 percent to 18 percent and a drop in corporate tax from 35.8 percent to 15.2 percent. In sum, the model predicts an increase of 7.41 percentage points of corporate savings post 1985, slightly lower than the actual change of 7.88 percentage points. Table 13. Contribution to Corporate Saving, Average Change between 1960-1984 and 1985-2012 Contribution to Corporate Savings (% GNDI) Contribution to Corporate Savings of each determinant Mean value for each period Long Run Coefficient* 1960-1984 1985-2012 Financial intermediate ratio (% GDP) 19.67% 169.32% 0.027 4.02% MPK net of tax (t+1) 9.76% 15.62% 0.578 3.39% Corporate savings predicted change 7.41% Corporate savings actual change 9.04% 16.92% 7.88% * Calculated as the current coefficient divided by one minus the coefficient estimated for the lag of corporate saving. Table 14 shows a similar exercise for household savings. Public savings shows a long-run coefficient of -0.80, which reflects the existence of the Richard equivalence effect, though without a complete compensation effect. This means that the increase in public savings from 4.64 percent to 5.56 percent of GNDI was offset by a decrease in household savings of almost 0.74 percentage points. The average increase in the growth rate of per capita income from 0.98 percent to 4.05 percent meant an increase in household savings of 1.03 percent of GNDI, where the long-run coefficient is 0.336, much lower than 1. Following the theory of Permanent Income, only a small fraction of increase in the average income was considered as permanent by the households. The largest contribution to the increase in household savings comes from the reduction of Personal Income Tax from 65.9 percent to 45.5 percent, which induced an increase in household savings of 1.92 percent of GNDI. Finally, the real interest rate contributed -0.46 percent to the increase of household saving between both periods, where its negative value for 44 the first period was due to a long period of financial repression plus the hyperinflation of the 1970s. Table 14. Contribution to Household Saving, Average Change between 1960-1984 and 1985-2012 Contribution to Households Savings (% GNDI) Contribution to Household Savings of each determinant Mean value for each period Long Run Coefficient* 1960-1984 1985-2012 Public savings (% of GNDI) 4.64% 5.56% -0.798 -0.74% Income per capita growth rate 0.98% 4.05% 0.336 1.03% Personal Income Tax (t+1) 65.90% 45.50% -0.094 1.92% Real interest rate -8.92% 2.39% -0.041 -0.46% Households savings predicted change 1.76% Households savings actual change -2.23% -0.22% 2.01% * Calculated as one minus the coefficient estimated for the lag of corporate saving. Table 15 summarizes the results for private savings. Adding up, the models predict that private savings should have increased in 9.2 percentage points while it has actually increased 9.9 percentage points. The main drivers are the Financial Intermediate Ratio, the ATMPK, the income per capita growth rate and the Personal Income Tax. The improvement in the fiscal budget had a negative, but small effect on private savings. Table 15. Contribution to Private Saving, Average Change between 1960-1984 and 1985-2012 Contribution to Private saving (% of GNDI) Predicted Actual Corporate savings 7.41% 7.88% Households savings 1.76% 2.01% Total change 9.17% 9.89% 45 Table 16 shows the contribution of each determinant to the average change of public savings between similar periods used for private savings. The change on public savings was approximately 1 percentage point between 1960-1984 and 1985-2012, increasing from 4.63 percent to 5.58 percent. Public savings has two determinants that may predict an important decrease in public savings in the second period: i) old age dependency and ii) the decrease in rural population. Both effects induce larger fiscal expenditure, implying an expected reduction of public savings by around 11 percentage points. These effects are offset mainly by i) copper rainfall (the long run/permanent component) during the 2000s and ii) the government’s change in attitude towards savings in the midst of our sample, captured by the dummy variables for presidential periods. The fiscal rule is also an important explanatory variable, but because the rule has been in place since 2001, it has little impact on average fiscal savings of 1985-2012. We might hypothesize that the shift in approach to fiscal savings since the mid-1980s resulted in the implementation of the fiscal rule at the beginning of the 2000s. In summary, the model predicts an increase in 0.62 percentage points, which is compared with the actual increase of 0.95 percentage points. Table 16. Contribution to Public Savings, Average Change between 1960-1984 and 1985-2012 Contribution to Public Savings of each determinant Mean value for each period Long Run Coefficient Contribution to the Change in Public Savings (% GNDI) 1960-1984 1985-2012 Real GDP deviation over HP trend (%) 0,55% -0,63% 0,346 -0,41% Real Copper price deviation over HP trend (%) -3,19% -0,03% 0,053 0,17% Log(HP trend of real copper price) 18,02 18,38 0,094 3,32% TIR 90-180 days (Percent points) -0,01 7,26 0,001 0,51% Older adults / population (Percentage point) 5,24 7,29 -0,039 -7,94% Rural Population/population (Percentage point) 23,60 14,31 0,004 -4,15% Presidential Period 1971-1973 0,12 0,00 -0,102 1,22% Presidential Period 1990-1993 0,00 0,14 0,037 0,53% Presidential Period 1994-1999 0,00 0,21 0,074 1,59% Presidential Period 2000-2005 0,00 0,21 0,095 2,05% Presidential Period 2006-2009 0,00 0,14 0,137 1,95% Presidential Period 2010-2012 0,00 0,11 0,154 1,65% Balance Budget Fiscal Rule 0,00 0,06 0,022 0,13% Public Savings Predicted 0,62% Public Savings Actual 4,63% 5,58% 0,95% 46 8. Concluding Remarks This paper aims to explain two important facts about the Chilean economy. The first is how Chile substantially raised its savings rate—by almost 11 percentage points—in the period 1987- 2012 compared to 1960-1986. The public sector increased its savings by almost 1 percent of disposable income, while the private sector raised savings by10 percent of disposable income. Second, the large increase in private savings is mainly explained by corporate savings, which seem to substitute household savings. The most important message that results from studying the Chilean experience is that significant changes in savings rates can be achieved through the implementation of public policies. First of all, and by far, the most important variable was the development of the financial market, accounting for nearly 4 percentage points of the increase in savings as a fraction of GNDI. Fuentes (2013) argued that an important public policy that promoted the development of the financial system was the pension fund reform undertaken in 1981. Second, other public policies designed to improve the economy’s productivity jointly with the 1984 tax reform also impacted total savings, as the increase in the after-tax marginal productivity of capital boosted corporate savings by 3.4 percentage points of GNDI. The tax reform implemented in 1984 lowered the corporate tax rate but also included other changes in the tax code such as i) implementing a full imputation system, ii) eliminating double taxation and iii) allowing taxation at the personal level and not on an accrual basis when withdrawal of profits occurred. In this line, and as a third element, the reduction in personal income taxes seemed to increase household savings by 1.92 percentage points of GNDI. Besides public policy, another determinant is that only a fraction of the increase in the average per capita income growth rate was ween by households as permanent, so only a small part of the higher income was assigned to consumption (one third) and the remaining income (two thirds) would have increased household savings by 1.03 percentage points of GNDI. Finally, better public sector institutions are yet another source of savings. In Chile’s case, the implementation of the structural balance rule was a source of 0.7 percentage points of GNDI.14 We could try to use our results to disentangle the impacts of the 2014 tax reform on Chilean savings. We might obtain some ideas by using our estimates to determine the impact of reinvestment tax on corporate savings. With the recent tax reform, corporate tax increased from 14 Net of the negative impact on corporate savings. 47 20 percent to 27 percent, a value that will be reached in 2018. That increase in corporate tax should affect the taxation of reinvested profits. Although our model was not built to estimate the effect of the changes in the tax code system approved in 2014, our model predicts a long-run impact of -1.82 percent on aggregate corporate saving. We should however be cautious. As described in Box 1, the 2014 tax reform included many other changes other than just that of the corporate tax. Therefore our estimates may be just the lower bound of the true impact. 48 Box 1. The 2014 Tax Reform The Chilean tax system was a full imputation system in which firms paid a 20 percent corporate tax based on accrued base, while firm owners paid personal tax based on cash flows and they could use corporate taxes as credit for personal taxation. That tax system mimicked a system based on dividend taxation. The corporate tax was set at 20 percent, while the maximum personal tax rate was 40 percent. The tax reform approved in September 2014 eliminated the full imputation system and introduced two different tax systems. The first is based on the concept of attributed income, in which firm owners are attributed with firm profits and therefore must pay personal taxes even if they do not pay dividends from the firm. The second tax system is a partial imputations system in which firm owners can only partially use corporate taxation as credit for their personal taxes. The tax payer can choose between both systems. In the attributed system the corporate taxation will increase to 25 percent, while in the partial imputation system, corporate taxation will rise to 27 percent, and just 65 percent of corporate tax will be allowed to be used as credit for personal taxes. Those changes will be gradually implemented in a four-year period. The tax reform increased the limit on sales for firms to be classified as small and medium size enterprises (SME). Those firms pay taxes based on cash flows rather than on accrued profits. Starting in 2017, SMEs owned exclusively by natural persons (not by other firms) are exempt from paying taxes, but their owners are required to pay the personal tax on an attributed base. Those firms might also decrease from their tax base part of their investment cost. Finally, those firms could also delay in 90 days the payment of sales tax. The tax reform also lowered the top rate in the progressive scale of the personal income tax from 40 percent to 35 percent starting in 2017. The tax reform incorporated other changes as well. In the housing market, individuals, who were usually exempted from taxes on capital gains, will pay taxes if capital gains over their lifecycle are larger than US$ 300,000. In addition, subsidies for new housing valued between approximately US$ 80,000 and US$ 170,000 were eliminated. Subsidies remain for new houses valued at less than US$ 80,000. Taxes on tobacco and alcoholic beverages increased, while a tax was applied to the first sale of new vehicles. The tax rate depends on emissions. 49 References Agénor, P. and J. Aizenman. 2004. “Savings and the Terms of Trade under Borrowing Constraints.” Journal of International Economics 6392): 321-340. Agosin M.R. 2001. “What Accounts for the Chilean Saving ‘Miracle’?” Cambridge Journal of Economics 25: 503-516. Agosin, M.R., G. Crespi and L. Letelier. 1997. “Análisis sobre el Aumento del Ahorro en Chile.” Documento de Trabajo R-309. Washington, DC, United States: Inter-American Development Bank. Attanasio, O., and A. Brugiavini. 2003. “Social Security and Households’ Saving.” Quarterly Journal of Economics 118(3): 1075-1119. Attanasio, O., and S. Rohwedder. 2003. “Pension Wealth and Household Saving: Evidence from Pension Reforms in the United Kingdom.” American Economic Review 93(5): 1499- 1521. Bai, J., and P. Perron. 1998. “Estimating and Testing Linear Models with Multiple Structural Changes.” Econometrica 66(1): 47-48 Bai, J., and P. Perron. 2001. “Computation and Analysis of Multiple Structural Change Models.” Journal of Applied Econometrics 18: 1-22. Bailliu, J., and H. Reisen. 1998. “Do Funded Pensions Contribute to Higher Savings? A Cross- Country Analysis.” Review of World Economics 134(4): 692-711. BancoEstado. 2013. “Un Banco de Todos y para Todos: Memoria Anual—Estados Financieros 2013.” Available at: http://www.corporativo.bancoestado.cl/docs/default-source/memorias-bancoestado/memoria_2013.pdf?sfvrsn=0 Bebczuk, R., and E. Cavallo. 2014. “Is Business Saving Really None of Our Business?” Working Paper IDB-WP-523. Washington, DC, United States: Inter-American Development Bank. Bennett, H., N. Loayza and K. Schmidt-Hebbel. 2001. “Un Estudio del Ahorro Agregado por Agentes Económicos en Chile.” In: F. Morandé and R. Vergara, editors. Empirical Analysis of Saving in Chile. Central Banking Series, Economic Analysis and Policies, Volume 1. Santiago, Chile: Central Bank of Chile. Bennett, H., K. Schmidt-Hebbel and C. Soto. 1999. “Serie de Ahorro e Ingreso por Agente Económico en Chile 1960-1997.” Discussion Paper 53. Santiago, Chile: Central Bank of Chile. 50 Bustos, A., E. Engel and A. Galetovic. 2004. “Could Higher Taxes Increase the Long-Run Demand for Capital? Theory and Evidence for Chile.” Journal of Development Economics 73: 675-697. Butelmann, A., and F. Gallego. 2000. “Household Saving in Chile: Microeconomic Evidence.” Economía Chilena 3(1): 5-24. Butelmann, A., and F. Gallego. 2001. “Estimaciones de los Determinantes del Ahorro Voluntario de los Hogares en Chile.” In: F. Morandé and R. Vergara, editors. Empirical Analysis of Saving in Chile. Central Banking Series, Economic Analysis and Policies, Volume 1. Santiago, Chile: Central Bank of Chile. Cerda, R., and F. Larraín. 2005. “Inversión Privada e Impuestos Corporativos: Evidencia para Chile.” Cuadernos de Economía 42: 257-281. Cerda, R., and F. Larraín. 2010. “Corporate Taxes and the Demand for Labor and Capital in Developing Countries.” Small Business Economics 34(2): 187-201. Cerda, R., and J.I. Llodrá. 2015. “Impacto del Impuesto Corporativo sobre el Stock de Capital de Largo Plazo en Chile.” Santiago, Chile: Pontificia Universidad Católica de Chile. Mimeographed document. Cerda, R., and D. Saravia. 2009. “Corporate Tax, Firm Destruction and Capital Stock Accumulation: Evidence from Chilean Plants, 1979-2004.” Working Paper 521. Santiago, Chile: Banco Central de Chile. Cerda, R. et al. 2014. “El Fondo de Utilidades Tributables (FUT): Elementos para la Discusión.” Estudios Públicos 135: 39-87. Corbo, V., and K. Schmidt-Hebbel. 1991. “Public Policies in Developing Countries.” Journal of Development Economics 36: 89-115. Corbo, V., and K. Schmidt-Hebbel. 2003. “Efectos Macroeconómicos de la Reforma de Pensiones en Chile.” Santiago, Chile: Central Bank of Chile. Mimeographed document. Dayal-Gulati, A., and C. Thimann. 1997. “Saving in Southeast Asia and Latin America Compared: Searching for Policy Lessons.” IMF Working Paper 97/110. Washington, DC, United States: International Monetary Fund. De Serres, A., and F. Pelgrin. 2003, “The Decline in Private Saving Rates in the 1990s in OECD Countries: How Much Can Be Explained by Non-Wealth Determinants?” OECD Economic Studies 36(4): 117-153. 51 Díaz, J., R. Lüders and G. Wagner. La República en Cifras, 2010. EH ClioLab-Iniciativa Científica Milenio. URL: http://www.economia.puc.cl/cliolab Edwards, S. 1996. “Why Are Latin America’s Savings Rates So Low? An International Comparative Analysis.” Journal of Development Economics 51(1): 5-44. Fuentes, R. 2011. “Learning How to Manage Natural Resource Revenue: The Experience of Copper in Chile.” In: P. Coller and A. Venables, editors. Plundered Nations? Successes and Failures in Natural Resource Extraction. London, United Kingdom: Palgrave Macmillan. Fuentes, R. 2013. “The Private Pension System’s Contribution to Latin America Economic Development: Evidence for Chile.” Santiago, Chile: SURA Asset Management. Gallego, F., F. Morandé and R. Soto. 2001. “El Ahorro y el Consumo de Durables frente al Ciclo Económico: ¿Consumismo, Frugalidad, Racionalidad?” In: F. Morandé and R. Vergara, editors. Empirical Analysis of Saving in Chile. Central Banking Series, Economic Analysis and Policies, Volume 1. Santiago, Chile: Central Bank of Chile. Grigoli, F., A. Herman and K. Schmidt-Hebbel. 2014. “World Saving.” IMF Working Paper 14/204. Washington, DC, United States: International Monetary Fund. Gutiérrez, M. 2007. Savings in Latin America after the Mid-1990s: Determinants, Constraints, and Policies. CEPAL Serie Macroeconomía del Desarrollo 57. Santiago, Chile: Comisión Económica para América Latina y el Caribe. Hachette, D. 1998. “Ahorro Privado en Chile.” Cuadernos de Economía 35(104): 3-47. Haque, N.U., M.H. Pesaran and S. Sharma. 1999, “Neglected Heterogeneity and Dynamics in Cross-Country Savings Regressions.” IMF Working Paper 99/128. Washington, DC, United States: International Monetary Fund. Hall, R.E., and D.W. Jorgenson. 1967. “Tax Policy and Investment Behavior.” American Economic Review 57(3): 391-414. Henríquez, C. 2008. “Stock de Capital en Chile (1985-2005): Metodología y Resultados.” Estudios Económicos y Estadísticos 63. Santiago, Chile: Banco Central de Chile. Horioka, C.Y., and A. Terada-Hagiwara. 2012. “The Determinants and Long-Term Projections of Saving Rates in Developing Asia.” Japan and the World Economy 24: 128-37. 52 Hsieh, C., and J. Parker. 2007. “Taxes and Growth in a Financially Underdeveloped Country: Evidence from the Chilean Investment Boom.” Economía: Journal of the Latin American and the Caribbean Economic Association 8(1): 1-40. Jorgenson, D.W. 1963. “Capital Theory and Investment Behavior.” American Economic Review 53(2): 247-259. Loayza, N., K. Schmidt-Hebbel and L. Servén. 2000. “What Drives Private Saving across the World?” Review of Economics and Statistics 82(2): 165-181. López, H., K. Schmidt-Hebbel and L. Servén. 2000. “How Effective Is Fiscal Policy in Raising National Saving.” Review of Economics and Statistics 82(2): 226-238. Masson, P., T. Bayoumi and H. Samiei. 1995. “Saving Behavior in Industrial and Developing Countries.” In: Staff Studies for the World Economic Outlook. Washington, DC, United States: International Monetary Fund. Medina, J.P., and R. Valdés. 1998. “Flujo de Caja y Decisiones de Inversión en Chile: Evidencia de Sociedades Anónimas Abiertas.” Cuadernos de Economía 35: 301-323. Ministerio de Economía de Chile. 2007 and 2009. “Encuesta Longitudinal de Empresas.” Santiago, Chile: Ministerio de Economía. Morandé, F. 1998. “Savings in Chile: What Went Right?” Journal of Development Economics 57(1): 201-228. Schmidt-Hebbel, K., and L. Servén. 2000. “Does Income Inequality Raise Aggregate Saving?” Journal of Development Economics 61: 417-446. Vergara, R. 2001. “Determinantes del Ahorro Privado en Chile.” In: F. Morandé and R. Vergara, editors. Empirical Analysis of Saving in Chile. Central Banking Series, Economic Analysis and Policies, Volume 1. Santiago, Chile: Central Bank of Chile. Vergara, R. 2010. “Taxation and Private Investment. Evidence for Chile.” Applied Economics 42(6): 717-725.