Inflation, Financial Development and Income Inequality

Carr, Jack; Chu, Kam Article Inflation, Financial Development and Income Inequality Kredit und Kapital Provided in Cooperation with: Duncker & Humblot, Berlin Suggested Citation: Carr, Jack; Chu, Kam (2005) : Inflation, Financial Development and Income Inequality, Kredit und Kapital, ISSN 1865-5734, Duncker & Humblot, Berlin, Vol. 38, Iss. 4, pp. 483-513, https://doi.org/10.3790/ccm.38.4.483 This Version is available at: https://hdl.handle.net/10419/293532 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. https://creativecommons.org/licenses/by/4.0/ Kredit und Kapital, 38. Jahrgang, Heft 4 Seiten 483-513 Inflation, Financial Development and Income Inequality By Jack Carr, Toronto, and Kam Hon Chu, St. John's* I. Introduction: Inflation, Financial Development and Income Distribution Traditional studies on the relationship between inflation and income distribution mainly focus on the impacts of inflation, both expected and unexpected, on income (re)distribution. The issue is quite extensively examined with reference to the experiences of many countries and the empirical results are also quite well documented (see for example Laidler and Parkin 1975, McCallum 1990, and Driffill, Mizon and Ulph 1990 and the references therein). Only until recently have economists looked into the possibility of income distribution as one of the determinants leading to higher inflation. These studies include Beetsma and van der Ploeg (1996), Carr and Chu (1996), Lippi and Swank (1996), Al-Marhubi (1997, 2000) and more recently Dolmas, Huffman and Wynne (2000). As shall be seen below, this paper adds to the literature by providing empirical evidence as well as a theory to explain how income inequality and financial development affect inflation. To highlight the contribution of this paper and how it differs from most of the major studies in the literature, we briefly recapitulate the major results of these studies in the next few paragraphs. In brief, the prevailing theories are based on a public interest argument applicable to democracies only and postulate that the poor benefit from inflation as a result of wealth redistribution. In sharp contrast, the theory of this paper does not rely on the assumption that politicians are benevolent social planners, and is thus applicable to non-democracies; it also contends that the poor rather than the rich suffer from higher inflation. * We would like to thank an anonymous referee, Norman Cameron, Andrew Filardo, Rod Hill, Chris McKenna, John Smithin, and participants at the Canadian Economics Association annual meetings for their helpful comments on earlier versions of this paper. We are responsible for all remaining errors. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 484 Jack Carr and Kam Hon Chu Among the first formal theories to explain the inequality-inflation link is the one by Lippi and Swank (1996), who reexamine and extend the Barro-Gordon (1983) model by interpreting it as a political model in which political parties care about both income distribution and output growth. If the central bank fails to make a policy commitment, then there is the well-known outcome that an inflationary bias will exist. However, if policymakers are concerned with income distribution only, there will be no inflationary bias in the sense that the optimal inflation policy depends on the distributional motives of policymakers. If lower income groups are more averse to unemployment and higher-income groups are more concerned with inflation, then policymakers catering to the distributional desires of lower-income groups will opt for high inflation. In this context, income distribution is a determinant of inflation because in equilibrium inflation depends on the targeted Gini coefficient. From a public-choice perspective and based on the median voter theorem, Beetsma and van der Ploeg (1996, hereafter BVP) argue that democratic governments are likely to levy inflation taxes in order to erode the real value of debt service, thus redistributing wealth from the rich to the poor. They also provide empirical evidence of a positive association between inflation and income inequality for a cross section of 23 democratic countries over the period 1960-85 to support their theory. Similarly, Dolmas, Huffman and Wynne (2000, hereafter DHW) recently argue that greater income inequality leads to higher inflation because of a desire by voters for wealth redistribution. In their overlapping generations model, higher income or wealth inequality may lead to greater pressure on the government to collect inflation taxes from the young in order to finance transfer payments to the old. Empirically, they also examine the impact of central bank independence on inflation (see for example Alesina and Summers 1993, among others). Their crosscountry regression results for 44 countries over the period 1960s-1980s indicate a positive relationship between income inequality and inflation for democracies, but not for non-democracies. They also conclude that democracies with more independent central banks tend to have better inflation outcomes for a given degree of inequality. While the public-choice approach adopted by Beetsma and van der Ploeg as well as Dolma, Huffman and Wynne explains the inequalityinflation link in democracies, its robustness and applicability to nondemocracies remains to be seen.1 Based on cross-country OLS regression results for 72 countries over the period 1973-90, Al-Marhubi (1997) finds Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 485 that countries with greater income inequality have higher average inflation, even after controlling for other country-specific variables. His subsequent empirical paper (2000) shows that the association between inflation and income inequality holds for both democracies and non-democ- racies and also that the results are not due to reverse causation. Apparently, what is lacking in the existing literature is a theory to explain the inequality-inflation link among non-democracies. As a matter of fact, non-democracies on average have higher inflation and more unequal income distribution than democracies (see Table 1 below). In addressing the issue of a monetary constitution for the Americas, Carr and Chu (1996, pp. 290-2) observe that high inflation countries tend to have higher income inequalities whereas low inflation countries have more equal income distributions. Furthermore, the top six countries with the highest post-war inflation - namely Brazil, Argentina, Bolivia, Peru, Uruguay and Chile - are all non-democracies.2 One plausible explanation is that in these countries higher inflation rather than an increase in income taxes is chosen by the government to extract tax revenues to finance its expenditures or deficits because an inflation tax has a lower incidence on the rich people than a progressive income tax does. Higher inflation thus promotes the interests of higher-income groups. This paper formalizes and extends the above observation by investigating the relationship between inflation on the one hand and income distribution and financial development on the other. The effects of the latter two factors on inflation are not discussed in the survey papers by Laidler and Parkin (1975) and McCallum (1990). Using an overlapping generations model of Sargent and Wallace (1981), we demonstrate that higher income inequality can lead to a higher steady-state inflation when the financial system is underdeveloped - in the sense that not all economic agents have access to the financial market. Financial underdevelopment can be an outcome of economic development or government regulations restricting financial institutions from providing alternative 1 Interestingly, Hill (2000) uses the same dataset provided by Dolmas, Huffman and Wynne to show that the relationship between income inequality and inflation across democracies is not robust and that, contrary to their finding, income inequality appears to have a stronger relationship with inflation in non-democra- cies. As shall be seen, our paper also shows that the nexus between inflation and income inequality is stronger among non-democracies than democracies. 2 Changes in political regimes over time are inevitable. For example, Chile did not become a democratic country until the end of 1989 (see, for example, Banks et al. (1997)). It is thus classified as a non-democracy in our study. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 486 Jack Carr and Kam Hon Chu "money-like" financial instruments to the non-bank public. When the economy is financially underdeveloped, the government can engineer inflation as a means to finance its expenditures because some economic agents, the poor in particular, cannot hedge against inflation by switching their money holdings into other interest-bearing financial assets. The choice of high inflation regimes can be attributed to inefficient tax collection systems (see, for example, Cukierman, Edwards and Tabellini 1992), which tilt governments towards seigniorage as a principal source of revenue. In this paper, we explicitly show that financial "undevelopment" (i.e., the financial system is not developed such that both the rich and the poor hold money as a store of value) can lead to higher steadystate inflation when compared with "underdevelopment" (i.e., when only the poor cannot access the financial market and have to hold money). Furthermore, the steady-state inflation rate increases if the income of the poor when they are old decreases. But this condition also means higher income inequality. Hence, inflation can be positively related to income inequality and negatively related to the level of financial development. Our hypothesis is supported by empirical evidence. The OLS and generalized instrumental variable estimation (GIVE) results for a subsample of 56 non-democratic countries, out of a cross-sectional data set for 90 countries over the period 1950-92, show a positive relationship between average inflation rates and the Gini coefficients and a negative relationship between average inflation rates and the degree of financial development. Besides explaining income inequality as a factor leading to higher inflation, this paper can also be viewed as complementary to a recent strand of literature on growth, income distribution and politics, which purports to show that higher income inequality is harmful to economic growth. Examples of these studies are Alesina and Rodrik (1994), Bertola (1991) and Persson and Tabellini (1991, 1992). The common vein in their arguments is that higher income inequality causes the government to impose a higher equilibrium tax rate so as to redistribute income, and economic growth is reduced as a result of higher taxation on capital. Another study by Alesina and Perotti (1996) argues that higher income inequality leads to social discontent and hence higher political instability, which has an adverse impact on investment and growth. Our study here does not directly address the issue of income distribution and economic growth. But if we believe that high inflation is harmful to growth, as many studies have shown (e.g., Andres and Hernado (1997), Kormendi Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 487 and Meguire (1984)), then our study suggests an alternative channel through which income inequality deters economic growth. This paper proceeds as follows. In the following section, we use an overlapping-generations model to derive the relationship between inflation on the one hand and financial development and income distribution on the other. In section III, we discuss the political economy of inflation and income distribution and explain why governments, particularly in non-democracies and less-developing countries, tend to choose inflation taxes to finance their expenditures and why they impose regulations on the financial sectors. In the penultimate section, we provide cross-coun- try empirical evidence to support our hypothesis. The last section summarizes and concludes. II. An Illustrative Model 1. The Setup We apply an overlapping generations model of Sargent and Wallace (1981) to illustrate our arguments. There are two different classes of economic agents - the rich and the poor - in the economy. Each economic agent lives for two-periods. At any time t > 1 there are born Ni(t) identical poor people who are endowed with a\ units of good when young and a2 units when old, where a\ > a2. On the other hand, at any time t > 1 there are born N2{t) identical rich people who are endowed with /31 units of good when young and f32 when old, where ¡3\ > 02. By definition, Pi > (32 > QI > a2. For simplicity, we consider the case where populations of the rich and the poor grow at the same rate n, i.e., Ni{t + 1) = iVi(t)(l + n), for all t and i = 1,2. Given this setup, the Gini coefficient of this economy is iVi2K " a2)(l + n) + IV22(A - + n) + 2NX iV2(ft + P2 ~ - <*2) + Ni ^ n(02 +3 ft -3 - a2) + N, N2 - a,) 1 ] ~ (iV1+iV2)(n + 2)[iV1K+a2)+iV2(ft +&) + n(Ni <*i+N2 A)] The above equation appears to be complicated and formidable. Nonetheless, for the purpose of our analysis to follow, it suffices to focus on how the Gini coefficient changes in response to a change in a2. Turning to the household utility maximization problem, the representative household is assumed to have the following logarithmic utility function: Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 488 Jack Carr and Kam Hon Chu (2) h h C (t),C (t+1) h h • Ine (t) + lnc (t+1). where c£(s) is the consumption of agent h of generation t in period s. The representative household is assumed to maximize the above utility function subject to the inter-temporal budget constraint: (3) ct(t+ 1) 1 + r(t) w (t+1) 1 + r(t) where is the endowment of agent h of generation t in period s and r(t) is the interest rate on consumption loans. The superscript h denotes whether the economic agent is rich or poor. The solution to the representative household's maximization problem leads to the following savings function: (4) s (t)=/h[l + r(t)]=- h W (t+1) W (t)~ ' 1 + r(t) If money is the only form of store of value in this economy, we have the following aggregate saving function:3 (5) M(t) P(t) N1+N2 H = E * w. where M(t) is currency and p(t) is the price level. Or in per worker (young capita) terms, (6) N1 +n2 h M(t) ? p(t)N(t) N(t) where N(t) = JVi(t) + JV2(t). = /(l + r(t)). 3 This is apparently unrealistic as households can hold their savings in portfolios of different assets. However, a model is by definition a simplification of reality. This simplifying assumption makes our model analytically tractable without loss of generality. As long as households hold part of their savings in the form of money, the government or central bank can extract inflation taxes from money creation. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 489 On the other hand, the government is assumed for simplicity to finance its expenditures, G(t), by currency creation:4 x M(t)-M(t- 1) Alternatively, G(t) can be interpreted as the government expenditure in excess of current tax revenue. This assumption is appropriate for most less developing countries which find it less costly to collect taxes by inflation than by their inefficient tax systems. For simplicity, the rate of currency creation is assumed to follow a currency growth rule: (8) M(t)=zM(t-l),z> 1. The government's objective is to choose the currency growth rate 2 to maximize seigniorage (or what is equivalent to maximize G(i)) in a stationary equilibrium, i.e., N(t) ^ P(*)(l + n) subject to (10) 1 < z < z. where z is determined such that 1 /p(t) > 0 for all t, i.e., fiat currency is valued in a monetary equilibrium. 2. Inflation, Financial Development, and Income Distribution Substituting the household saving function (Equation (6)) into the government budget constraint (Equation (9)), both in per capita terms, yields the following 4 We omit the important issue of central bank independence and simply assume that the central bank and the government together are a single entity. This assumption makes our theoretical analysis and empirical evidence more relevant as most countries, especially non-democracies, do not have independent central banks. Based on their empirical results, Dolmas, Huffman and Wynne (2000) argue that greater central bank independence seems to alleviate inflation pressure stemming from greater income inequality in a democratic society - the measure of central bank independence contributes to explaining cross-country differences in inflation, although the relationship is rarely significant in a statistical sense. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 490 Jack Carr and Kam Hon Chu _ G(t) M(t) - M(t - 1) 9 = mt)~ p(mt) (ii) In a state steady equilibrium, /(I + r(t)) = /(I + r(t + 1)) = ... = /(I + r), which implies p(t + 1 )/p(t) = z/( 1 + n) = 1/(1 + r). Therefore, the government's problem can be rewritten as MAX /(1 } (12) = /(*)( First, consider the case in which the economy is financially "undeveloped" such that fiat money is the only form of store of value. In this case, both the rich and the poor have to hold fiat money as a store of value in order to smooth out their consumption over their life. The solution to the government's maximization problem gives the following steady-state inflation rate: /iVi«i+iV2A ^ . (13) * = Then consider the case in which the economy is financially "underdeveloped" in the sense that only the rich can acquire other financial assets that serve as stores of value.5 This can be the case when we assume that the minimum amount for investing in these financial assets is /?i/2 and that the government imposes legal restrictions such that individuals cannot share investments. As a result, only the poor hold money as a store of value. The steady-state inflation rate in this case is 5 As already mentioned before, it is not the objective of this paper to model the portfolio choices of households. But there should be little doubt that the rich have better access to financial markets than the poor in the real world, particularly in less developing countries. For example, a recent report by World Bank (2001) states that the financial systems of developing countries are small and they provide fewer services at higher unit costs partly because of failure to exploit economies of scale (pp. 19-20) and also that low-income households are excluded from the international capital markets whereas large depositors can place their funds and make their investments abroad (pp. 180-181). Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 497 (17) CPI = a0+a1M + a2FD + a3 GINI + p.. where CPI is the rate of change of the Consumer Price Index, as a measure of the rate of inflation, M is the money supply growth rate, FD is a proxy variable for the degree of financial development, GINI is the Gini coefficient, as a measure of income distribution, and [i is a random disturbance term. Both CPI and M are in logarithmic form. If our theory is correct, we expect ai > 0,a2 < 0 and a3 > 0. Given the above specification, the coefficient ai measures the elasticity of inflation with respect to the money supply, whereas a2 and a3 measure the relative changes in the inflation rate with respective to absolute changes in the level of financial development and the Gini coefficient respectively. Cross-sectional data are used to test our hypotheses. The main sources of data, except the Gini coefficients, are International Financial Statistics (IFS) and World Tables published by the International Monetary Fund and the World Bank respectively. All data are annual average figures for the period 1950-1992. Some countries have shorter time series because data are not available. The average annual rate of change of the consumer price index (line 64..x in IFS) is used as a measure of inflation, whereas M is the average annual rate of money supply (line 34 in IFS). The best measure of financial development is probably the Financial Interrelations Ratio (FIR) proposed by Goldsmith (1969). This is the ratio of the sum of the value of all financial assets to the sum of the value of all real assets. A higher value of FIR reflects a higher degree of financial development. Unfortunately, most, if not all, countries do not have the relevant data series for us to construct the FIR. Instead, we use several other indicators as proxies for financial development. The first proxy is BMY, which is defined as the ratio of broad money (line 34 plus line 35 in IFS) to nominal GDP (line 99b). This is similar to the ratio of liquid liabilities of the financial system to GDP (labelled as LLY in their paper) as used as an indicator of financial depth in a study by King and Levine (1993). These two indicators are in line with the traditional practice, such as Goldsmith (1969) and McKinnon (1973), which uses the size of the formal financial sector relative to the size of the economy as a measure of financial development or financial depth. The provision of financial services is assumed to be positively related to the size of the financial sector. While such an indicator may be appropriate in the study of financial development and economic growth, it is not quite appropriate, Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 498 Jack Carr and Kam Hon Chu or to a certain extent distortionary, for our present study because the growth in broad money could be due to an increase in currency rather than financial development. If currency accounts for a large proportion of broad money, we may observe a positive correlation between inflation and the ratio of broad money to GDP instead of a negative correlation as predicted by our theory For comparison, we will still produce the empirical results based on the ratio of broad money to GDP as a proxy for financial development despite the potential distortionary effect as mentioned above. The second proxy for financial development used in this study is QMY, defined as the ratio of quasi-money to GDP. This indicator is used by Brodsky and Finnerty (1994) to study the relationship between financial depth and human development. Quasi-money includes time, savings, and foreign currency deposits held at depository institutions such as banks (line 35 in IFS). Well developed financial markets are expected to have high levels of quasi-money. Furthermore, a high level of quasi-money is associated with a wider array of financial products as well as a greater financial depth. The underlying idea is essentially similar to the first indicator except that non-interest bearing currency and demand deposits are excluded. Therefore, this indicator is expected to serve as a more reliable indicator of financial development than the first one. While quasimoney grows in nominal value over time due to inflation, it is also the case for nominal GDP. Therefore, the ratio of quasi-money to GDP can be taken as a measure of the size of financial sector relative to economic activity. A higher ratio of quasi-money to GDP is thus taken as a higher degree of financial development or financial depth. The last proxy we use in this study does not rely on the strong assumption that the degree of financial development is directly proportional to the size of the financial sector. Instead, it is based on a "stylized" pattern observed in the development of the financial sector in almost all economies (see, for example, Dow and Earl 1982). This pattern is the switch of using commodity money, or currency, to an alternative, more convenient, asset-like financial instrument, such as bank checking deposits, in the exchange process as both the economy and the financial system develops; and further financial innovation and development saw the emerging of money-like instruments such as savings and time deposits. Based on this pattern of financial development, we construct the ratio of quasi-money to currency (QMC) as an indicator of financial depth. The higher the value of this ratio, the higher the degree of financial development. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 499 The Gini coefficient is used as a measure of income inequality. A higher value for the Gini coefficient is assumed to represent higher income inequality, although it is well known that two different income distributions can have the same Gini coefficient. Data on the Gini coefficients are obtained directly or indirectly computed from various sources. The main data source is Deiniger and Squire (1996). Other sources include Jain (1975), Sundrum (1990), United Nations (1985), and various issues of World Development Reports. For countries with more than one value of the Gini coefficient reported for the sampling period 1950-1992, the average figures are used in the regressions. While some countries experienced changes in income distribution over the sampling period, their changes are not dramatic. Moreover, the pattern seems to be relatively stable over time, i.e., the rank orders of inequality appear to change only a little over the sampling period. Countries with high initial income equality, as ranked by the Gini coefficients, remain to have high income equality throughout the sampling period. Based on available data, a total of 90 countries (see Table 1 for a list of these countries) are included in our sample. To facilitate the reader to assess the impacts of the level of financial development and income distribution on inflation, we first report in Table 2 the OLS results with money supply growth omitted as an explanatory variable. In all cases the coefficient of the Gini coefficient has the predicted positive sign and is statistically significant, except in the QMY equation where it is only marginally significant. Similarly, the coefficient of the level of financial development has the predicted negative sign and is statistically significant in all cases except in the BMY equation. However, in terms of R2 a large proportion of the variations in inflation is unexplained by these two variables.13 As expected, money supply growth is the main factor leading to higher inflation. This is revealed by the OLS results tabulated as Table 1-3 for the entire sample, the subsample of non-democracies and democracies respectively. In each table, the first set (column) of results is based on using BMY as a proxy for financial development, whereas the second and third set of results are respectively based on using QMY and QMC respectively as a proxy for financial development. In all cases, the coefficients of the money supply growth variable have the correct positive sign 13 We would like to remind the reader that these results are for reference purpose only and should be interpreted with qualifications because they are based on a "mis-specified" model with an omitted variable. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 500 Jack Carr and Kam Hon Chu and are statistically significant. For hypothesis testing, White's (1980) heteroscedasticity-consistent standard errors are also reported to account for heteroscedasticity, which are detected in some cases. While the coefficients of the level of financial development have the predicted negative sign in all cases, they are not all statistically significant. The results for the Gini coefficient are mixed: the parameters are statistically insignificant in most cases, not to mentioned that in some cases they have the wrong negative sign as predicted by our theory. Nonetheless, the results for non-democracies (Table 4), notably those for the QMC equation, lend support to our theory As our theory predicts that inflation and income inequality are positively related, the estimates for the BMY and QMY equations are also significant at the 10% level if onetailed tests are used. Though efficient, the OLS estimates are bias and inconsistent because the money supply growth is endogenously determined according to our theory and this is also confirmed empirically.14 For this reason, we apply the generalized instrumental variable estimation (GIVE) technique to Equation (17) to obtain consistent estimates.15 For the money supply growth, we use the (log of) average annual growth rate of government expenditures at current market prices (line 91 or 9 If IFS) as an instrument. Government expenditure growth is a suitable instrument because it is a major driving force behind the government's need to finance its expenditures by inflationary taxes, as suggested by Equation (7).16 Both variables FD and GINI are taken as exogenous variables and used as instruments themselves.17 Furthermore, all the FD proxies are used as extra instruments so that we have the degree of freedom to test the joint 14 This endogeneity problem is empirically confirmed by an omitted variable (OV) version of Hausman (1978) specification test, the results of which indicate contemporaneous correlation between the error and the money supply growth variable. See for example Kennedy (1998, p. 151) for details of the test procedures. is Given our finite sample size, however, both the OLS and IV results are reported for comparison. !6 Theoretically, the monetization of government deficit rather than government expenditure growth is the ultimate factor. But empirically, the latter is a better instrument. For example, Protopapadakis and Siegel (1987) found little relation between budget deficits or public debt and inflation or money growth for 10 industrial countries. I? Theoretically, inflation can affect income inequality and hence the variable GINI can be endogenous. However, this is not a problem empirically, at least in our sample, because the results of another OV version of Hausman test indicate that the variable is not endogenous. For details of the testing procedures in this case, see for example Kennedy (1998, pp. 174-5). Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 501 hypothesis of validity of the instruments and correct model specification.18 As heteroscedasticity is detected in the OLS regressions, White's (1980) heteroscedasticity-consistent covariance matrixes are used as the weighting matrixes in the GIVE estimation. The results are reported in Tables 6-8. As in the case of the OLS results, the results for non-democracies (Table 7) support our theory, as indicated by the correct signs and statistical significance of the explanatory variables in almost all equations. The test statistics proposed by Davidson and MacKinnon (1993, pp. 232-7) do not reject the joint hypothesis that the instruments are valid and the model is correctly specified. In contrast, for democracies the parameter estimates of the Gini coefficient are all statistically insignificant, not to mention that some have the incorrect negative sign. Overall, our empirical results lend support to our theoretical hypothesis that inflation is related positively to income distribution but negatively to the level of financial development, notably for the sample of non-democratic countries. The results are in line with the findings of Al- Marhubi (1997, 2000) based on a sample of both democratic and nondemocratic countries. On top of these regression results, it is a plain fact that non-democracies have in general higher average inflation rates and more unequal income distributions than democracies, as clearly revealed in Table 1. V. Conclusion In the recent, blossoming literature on income inequality as a cause of inflation, the prevailing view is that higher income inequality leads to higher inflation in democratic countries because governments implement inflationary policies to redistribute wealth from the rich to the poor. These theories, however, may not generalize to non-democracies, which on average have persistently higher inflation and more unequal income than democracies. is Apart from enabling testing the hypothesis, Kennedy (1998, p. 153) suggests that it would seem desirable to have two more instruments than explanatory variables for the IV estimator to have better properties in finite samples. Anyhow, in our case the empirical results for using the extra instruments are found to be very similar to those when the number of instruments is the same as the number of explanatory variables. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 502 Jack Carr and Kam Hon Chu Based on differential tax incidence, this paper provides an alternative explanation of how unequal income distribution can lead to high inflation. An inflation tax has a lower incidence on the high income group than an income tax. Persistently high inflation polices can be the combined result of high income inequalities, inefficient tax systems and political factors. Financial development is also a determinant of inflation because the government can extract more inflation taxes if the financial system is less developed. As our theory does not rely on the assumption of benevolent politicians or governments, it should have a wider applicability, especially to non-democratic countries. Empirically, our theory is supported by OLS and GIVE regression results for 56 non-democratic countries over the period 1950-1992, which indicate that inflation is positively related to money supply growth and the Gini coefficient but negatively related to the level of financial development. In the literature there are quite a lot of studies on the distributional consequences of inflation, but much less work on the reverse channel. From this perspective, this study can be regarded as making a small step forward in explaining how income inequality, together with financial development, affects inflation. Given the high degree of heterogeneity across countries, there is unlikely a general theory applicable to all countries. Our theory can at best be interpreted as being applicable to most non-democracies. Given the complex interdependencies among inflation, income distribution and financial development, the possibility of alternative theoretical interpretations of our empirical findings cannot be entirely ruled out. Further theoretical and empirical work is needed before we have a better understanding of the interplay between income inequality and inflation. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 503 Table 1 List of Countries and Summary Statistics (a) Democracies Australia Austria Belgium Bahamas Barbados Canada Colombia Cameroon Costa Rica Denmark Finland France Germany Greece Guyana India Ireland Israel Italy Jamaica Japan Mauritius Malaysia Netherlands Norway New Zealand Portugal Spain Sri Lanka Sweden Switzerland United Kingdom United States Venezuela (b) Non-Democracies Algeria Argentina Bangladesh Bolivia Brazil Botswana Chad Chile China Cyprus Dominican El Salvador Ecuador Egypt Fiji Gabon Ghana Guatemala Hong Kong Honduras Indonesia Iran Ivory Coast Jordan Kenya Korea Madagascar Malawi Mexico Morocco Nepal Niger Nigeria Pakistan Panama Peru Philippines Rwanda Senegal Seychelles South Africa Sudan Sierre Leone Singapore Suriname Taiwan Tanzania Thailand Trinidad & Tobago Tunisia Turkey Uganda Uruguay Yugoslavia Zambia Zimbabwe All Countries Non-Democracies Democracies Number of Countries 90 56 34 Inflation: Mean 21.6708 29.7105 8.4291 Standard Deviation 47.5389 58.7767 6.8439 Range 2.088-307.494 2.088-307.494 2.898-42.992 Gini Coefficient: Mean 0.4124 0.4351 0.3750 Standard Deviation 0.0869 0.0881 0.0716 Range 0.2598-0.6230 0.2890-0.6230 0.2598-0.5151 Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 504 Jack Carr and Kam Hon Chu Table 2 Effects of Income Distribution and Financial Development Dependent variable: log(CPI) Number of Observations = 90 log(CPT) log(CPT) log (CPI) log(CPT) Intercept 1.3325 (2.793)*** 1.2137 (2.401)** 2.0139 (3.725)*** 1.4951 (3.053)*** GINI 2.4863 (2.196)** 2.5463 (2.237)** 1.6503 (1.432)1 2.4228 (2.148)** BMY 0.2214 (0.729) QMY -1.5715 (-2.455)** QMC -0.0519 (-1.358)1 R2 0.0519 0.036 0.093 0.0503 F-statistic 4.8204** 2.6630* 5.5621*** 3.3560** S.E.E. 0.9284 0.9309 0.903 0.924 Log-likelihood -120.011 -119.737 -116.997 -119.067 Notes: 1. Figures in parentheses are t-statistics based on OLS standard errors. 2 | ** an(j *** reSpectively denote significance at the 20%, 10%, 5% and 1% levels. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 505 Table 3 OLS Results for All Countries Dependent variable: log(CPI) Number of Observations = 90 Proxy for FD BMY QMY QMC Intercept -0.8575 -0.6021 -0.8107 (-3.67)*** (-2.31)** (-3.48)*** [-3.69]*** [-2.31]** [-3.74]*** log(M) 1.0994 1.0778 1.0913 (20.42)*** (20.21)*** (20.39)*** [16.54]*** [16.31]*** [16.82]*** FD -0.0446 -0.5436 -0.0174 (-0.35) (-1.99)** (-1.09) [-0.32] [-2.17]** [-1.28]| GINI 0.4647 0.2273 0.4703 (0.96) (0.46) (0.98) [0.99] [0.48] [1.06]** R2 0.8332 0.8404 0.8353 F-statistic 149.24*** 157.16*** 151.43*** S.E.E. 0.3871 0.3788 0.3848 Log-likelihood -40.26 -38.3 -39.71 White Test 21.43*** 20.90** 16.40** Notes: 1. Figures in parentheses are t-statistics based on OLS standard errors whereas figures in brackets are those based on White's heteroscedasticity-consistent standard errors. 2 | * ancj *** respectively denote significance at the 20%, 10%, 5% and 1 % levels. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 506 Jack Carr and Kam Hon Chu Table 4 OLS Results for Non-Democracies Dependent variable: log(CPT) Number of Observations = 56 Proxy for FD BMY QMY QMC Intercept -1.2014 -0.9817 -1.2515 (-3.38)*** (-2.77)*** (-3.60)*** [-3.31]*** [-2.81]*** [-3.72]*** log(M) 1.1492 1.1322 1.1483 (16.19)*** (17.15)*** (17.15)*** [13.39]*** [14.10]*** [13.92]*** FD -0.0875 -0.837 -0.0446 (-0.54) (-2.18)** (-1.81)* [-0.55] [-2.92]*** [-2.77]*** GINI 0.887 0.7409 1.1383 (1.32)t (1.15) (1.71)* [1.50]t [1.25]t [2.07]** R2 0.8337 0.8468 0.8427 F-statistic 92.98*** 102.33*** 99.22*** S.E.E. 0.4357 0.4181 0.4237 Log-likelihood -30.86 -28.56 -29.29 White Test 19.20** 20.27** 10.74** Notes: 1. Figures in parentheses are t-statistics based on OLS standard errors whereas figures in brackets are those based on White's heteroscedasticity-consistent standard errors. 2 ** an(j *** respectively denote significance at the 20%, 10%, 5% and 1% levels. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50 Inflation, Financial Development and Income Inequality 513 Zusammenfassung Inflation, Entwicklung des Finanzsektors und Einkommensungleichheiten Im Gegensatz zu den meisten traditionellen Untersuchungen, die auf die Verteilungswirkungen von Inflation gerichtet sind, enthält dieser Beitrag eine theoretische Untersuchung, warum die Einkommensverteilung eine Determinante von Inflation sein kann. Unter Verwendung eines Modells sich überlappender Generationen zeigt dieses Papier, dass im steady State die Inflation bei einer „Unterentwicklung" des Finanzsektors, die dadurch gekennzeichnet ist, dass die Reichen und die Armen Geld nachfragen, höher ist als im Falle einer „Unterentwicklung", bei der nur die Armen Geld nachfragen. Ferner wird die Inflation dann höher sein, wenn das Einkommen der Armen im Alter niedriger ist. Somit kann eine ungleiche Einkommensverteilung im steady State zu höherer Inflation führen. Der Grund dafür ist, dass die Regierung mehr Inflationssteuern erheben kann, deren Inzidenz für die Reichen geringer ist als die Inzidenz einer progressiven Einkommensteuer. Dies mag erklären, warum insbesondere nicht demokratische Länder im politischen Prozess eher dazu neigen, eine höhere Inflation zuzulassen, als eine Steuerreform durchzuführen. Wir haben im Zeitraum von 1950 bis 1992 sowohl OLS- als auch GIVE-Schätztechniken auf die für 90 Länder erhobenen Daten angewandt. Unsere Hypothese wird von den Ergebnissen einer Stichprobe für 56 nicht demokratische Länder gestützt. Diese zeigt, dass Inflation positiv mit dem Geldmengenwachstum und dem Gini-Koeffizienten, jedoch negativ mit dem Niveau der Entwicklung des Finanzsektors korreliert. Résumé Inflation, développement financier et inégalité de revenus Contrairement à la plupart des études traditionnelles qui se focalisent sur les effets de distribution de l'inflation, cet article examine de façon théorique comment la distribution du revenu peut représenter une déterminante de l'inflation. A l'aide d'un modèle de débordement de générations, il est montré ici que l'inflation à l'état d'équilibre sous le «non-développement» financier - dans lequel autant les riches que les pauvres gardent de l'argent - est plus élevée que sous un «sous-développement» financier où les personnes pauvres seulement détiennent de l'argent. De plus l'inflation sera plus élevée si le revenus des personnes âgées pauvres est plus bas. Donc, une distribution de revenus plus inégale peut entraîner une inflation d'équilibre plus élevée. Ceci s'explique par le fait que le gouvernement peut extraire plus de taxes d'inflation, ce qui a une plus faible incidence sur les riches qu'une taxe progressive sur le revenu. Ceci pourrait expliquer pourquoi dans le processus politique, particulièrement dans les pays non-démocra- tiques, on préfère une inflation plus élevée à une réforme fiscale. Les techniques d'estimation OLS et GIVE sont utilisées ici pour des données internationales de 90 pays sur la période s'étendant de 1950 à 1992. Les auteurs soutiennent l'hypothèse que les résultats d'un sous-échantillon de 56 pays non démocratiques indiquent que l'inflation a un rapport positif avec la croissance de l'offre monétaire et au coefficient de Gini mais qu'elle est corrélée négativement au niveau du développement financier. Kredit und Kapital 4/2005 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.38.4.483 | Generated on 2023-01-16 13:22:50