An estimation of CPI biases in Argentina 1985–2005 and its implications on real income growth and income distribution

Abstract

We use the shifts in Engel curves calculated from household surveys to estimate CPI biases in Argentina between 1985 and 2005. We find that real earning levels increased during this period between 4.3 and 5.7% faster per year than previously estimated. More surprisingly, relative to conventional wisdom, that income distribution has improved throughout this period.

Introduction

Argentina has always been considered a basket case. No better proof of this fact than the name of this collection which refers to Argentina’s exceptionalism, thus assuming that there is something unusual, “exceptional”, for good or bad, regarding Argentina’s economic performance.

It is a well-known fact that at the turn of the XXth century, Argentina was among the richest countries in the world¹ and that after WWII started a long period of economic decline.² While by the turn of the XXIst century, Argentina still was, in PPP terms, the richest among large Latin American countries, and it had lost significant ground relative to its peer group of a century ago. This long stagnation has become to some an apparently unavoidable fate, only to be interrupted occasionally by brief growth spurts that inevitably provided the stage for the following crisis (a process that has been dubbed “stop go” dynamics).³ In fact, studies about the Argentine perception of the business cycle indicate that Argentines tend to become pessimists in the midst of each economic boom as if anticipating the unavoidable next crisis (see Gabrielli and Rouillet 2003).

This stagnation and perennial process of going forward and backwards has permeated not only the economic sphere, but has also been relevant in politics, as Argentina witnessed a string of military interventions between 1930 and 1983. It is perhaps in this parallel dimension, where Argentines feel that real progress has been made since 1983, as nowadays, there is virtually no possibility of an interruption of the democratic political process. However, has this improvement in the political sphere been matched by a similar success in economic performance? Not in the collective imagination. Since the return of democracy, the country has experienced two hyperinflations, several defaults and restructurings of its debt, many large devaluations, periods of persistent high inflation, deflation, introduction of parallel currencies, and deep economic crises. This poor economic performance has implied a volatile evolution of its per capita GDP growth and a deteriorating income distribution, as shown in Fig. 1. It is the long period between the 70s and the first decade of the 21st century that has built the belief of a stagnant economy. Taking 1983, the year of the restoration of democracy, as a starting point, output per capita has grown only 1.5% when considering the period until 2009. However, the per capita income of 1983, with ups and downs, was left behind only in 2002–2003. The per capita income of 1980 was left behind only in 2005, i.e., 25 years later.

Fig. 1

Real GDP growth and income distribution. Source the Gini coefficient includes only Buenos Aires and its metropolitan area; it was computed using the Socioeconomic Database of Latin America and the Caribbean (CEDLAS-WB); the Real GDPpc are values reported in World Development Indicators (WB)

All the historical literature accepts this perspective as given,⁴ providing a cohesive and unanimous answer to the question about economic performance: Argentina’s exceptional bad performance since democracy is considered a stylized fact.

The purpose of this paper is to challenge this view. In fact, we want to challenge the view that economic performance during Argentina’s recent democracy has been dismal, both in terms of earnings growth as well as in terms of income distribution. Using the shift in the Engel curves to re-estimate the relevant price levels, we will argue that real earning growth has been steady and much bigger than measured and that income distribution has improved. If we are able to convince our readers of our results, our work would throw a completely new light on recent economic performance. Under this new light, the exceptionalism that has been the focus of the other papers in this special collection would appear to have been left behind already two decades ago. With commodity prices on a relatively strong footing, a region that appears to be increasingly in order, and large wealth increases from the development of mining, agricultural, and energy resources, the prospect for Argentina looks bright in the foreseeable future.

The outline of this paper is extremely simple. The next section explains the methodology to correct the bias in the price levels typically used to estimate real income growth; section results shows the estimation results; and the last section provides some final thoughts. Our conclusions are that Argentina’s exceptionalism is a presumption that still needs to be proven and that Argentina’s economic performance during our recent democracy, both in terms of income distribution and earnings growth, has been substantially better than accepted in the economic debate.

Methodology

It is standard to use income as the most relevant measure to estimate well-being. However, to obtain a comparable measure of income over time, it is necessary to deflate the nominal measures at each specific moment by a price series, most commonly the consumer price index (CPI). In the case of Argentina, in particular, the one used is that corresponding to the city of Buenos Aires and its metropolitan area. This a Laspeyres type index, with a fixed basket, and subject to a series of well-known biases.⁵

First, these indexes overestimate inflation, because they omit the effect of substitution between goods, changes in quality of the goods, and the impact of the availability of new products. Second, the use of a common price index may be a problem when building measures of income distribution, because it assumes that baskets are equivalent across all income groups.

In Argentina, consumption surveys are not very frequent. The last three were conducted in 1984–1985, 1996–1997, and 2004–2005, and where undertaken to update the basket in the CPI. However, the large time gap between updates may lead to significant biases, particularly if we consider the large structural changes undergone by the Argentina economy over the last 25 years (e.g., a large trade liberalization process).⁶ Thus, correcting for the biases produced in the CPI can change the evolution of real income, and correcting for the biases at different income levels can also change the evolution of income distribution during this period.⁷

These consumption surveys can be used to estimate the biases following the methodology of Costa (2001) and Hamilton (2001). In a nutshell, the methodology uses the assumption that Engel curves for food should be relatively stable. If this is the case, when the estimation of the Engel curves at different dates shows shifts, it is assumed that these correspond to CPI bias. To illustrate the point, consider two points in time between which the share of food in income declines with a stagnant earnings level. Under the assumption that the Engel curve is stable, this provides a presumption that CPI may be biased (overestimated in this case) as a falling income share is consistent with rising, not stagnant, income levels. Thus, the changes in the share, with some assumptions, may be linked to the CPI bias. Of course, the biases in the Engel curve are obtained after correcting for changes in relative prices and household characteristics.

In later work, Carvalho Filho and Chamon (2012) use semi-parametric models to extend the methodology to estimate the biases at different income levels, thus allowing to tackle the issue of income distribution.

We should clarify that in the previous work, identification was built from exploiting the differences across regions. In the case of Argentina, however, our data contained only one area (the metropolitan area of the city of Buenos Aires). Thus, our paper needs to innovate from a methodological point of view relative to the previous work, by finding a way to obtain identification when only data from one region are available, something we do using individual price indexes by household.

Estimating CPI biases

Following Costa (2001), the estimation strategy starts formally form the following equation:

$$w_{ijt} = \varphi + \gamma \left( {\ln P_{Fjt} - \ln P_{Njt} } \right) + \beta \left( {\ln Y_{ijt} - \ln P_{Gjt} } \right) + \sum\limits_{x} {\theta_{x} X_{ijt} } + \mu_{ijt}$$

(1)

where w _ijt is the ratio of food to non-food of household i, in region j at time t; P _Fjt is the true unobservable price of food in region j at time t; P _Njt is the true and unobservable price of non-food in region j at time t; Y _ijt is nominal income for household i, in region j at time t; P _Gjt is the true and unobservable general price level in region j at time t; X _ijt is a set of control variables for household i, in region j at time t; μ _ijt is a random term; and ϕ, γ, β, and the different θ _x are parameters.

If we call: $\prod_{Gjt}$ is the cumulative percentage growth of the observable CPI in region j, since time 0 and time t; $\prod_{Fjt}$ is the cumulative percentage growth of the price of food, in region j, between time 0 and time t; $\prod_{Njt}$ is the cumulative percentage growth of the price of non-food, in region j, between time 0 and time t; E _Gjt is the cumulative percentage increase in the measurement error in the CPI in region j, between time 0 and time t; E _Fjt is the cumulative percentage increase in the measurement error in the price of food, in region j, between time 0 and time t; and E _Njt is the cumulative percentage increase in the measurement error in the price of non-food, in region j, between time 0 and time t.We can rewrite (1) as

$$\begin{aligned} w_{ijt} &= \varphi + \gamma \left[ {\ln \left( {1 + \prod_{Fjt} } \right) - \ln \left( {1 + \prod_{Njt} } \right)} \right] + \beta \left[ {\ln Y_{ijt} - \ln \left( {1 + \prod_{Gjt} } \right)} \right]_{Gj0} \hfill \\ &\quad + \gamma \left[ {\ln P_{Fj0} - \ln P_{Nj0} } \right] - \beta \ln P + \gamma \left[ {\ln \left( {1 + E_{Fjt} } \right) - \ln \left( {1 + E_{Njt} } \right)} \right] \hfill \\&\quad - \beta \ln \left( {1 + E_{Gjt} } \right) + \sum\limits_{x} {\theta_{x} X_{ijt} } + \mu_{ijt} . \hfill \\ \end{aligned}$$

(2)

If we assume that the mismeasurement does not change across regions, we can rewrite (2) as

$$\begin{aligned} w_{ijt} & = \varphi + \gamma \left[ {\ln \left( {1 + \prod_{Fjt} } \right) - \ln \left( {1 + \prod_{Njt} } \right)} \right] + \beta \left[ {\ln Y_{ijt} - \ln \left( {1 + \prod_{Gjt} } \right)} \right] \hfill \\ &\quad+ \sum\limits_{j} {\delta_{j} D_{j} } + \sum\limits_{t} {\delta_{t} D_{t} } + \sum\limits_{x} {\theta_{x} X_{ijt} } + \mu_{ijt} \hfill \\ \end{aligned}$$

(3)

where $D_{j}$ and $D_{t}$ are dummies by regions and period, and

$$\delta_{j} = \gamma \left( {\ln P_{Fj0} - \ln P_{Nj0} } \right) - \beta \ln P_{Gj0}$$

(4)

$$\delta_{t} = \gamma \left[ {\ln \left( {1 + E_{Ft} } \right) - \ln \left( {1 + E_{Nt} } \right)} \right] - \beta \ln \left( {1 + E_{Gt} } \right).$$

(5)

Notice that $\delta_{t}$ is a function only of time. If we additionally assume that the biases for food and non-food items are similar, we can compute a measure of the general CPI bias from

$$\ln \left( {1 + E_{Gt} } \right) = - \frac{{\delta_{t} }}{\beta }.$$

(6)

From (6), we can compute $E_{Gt} = e{}^{{ - \frac{{\delta_{t} }}{\beta }}} - 1$ which is the measurement error between real inflation and CPI inflation. − E _Gt is the cumulative bias.The assumption that the bias for food and non-food are the same is not necessarily very realistic. However, under reasonable assumptions, our measure can be considered a lower bound for the estimate. From (5)

$$\ln \left( {1 + E_{Gt} } \right) = \frac{{\gamma \left[ {\ln \left( {1 + E_{Ft} } \right) - \ln \left( {1 + E_{Nt} } \right)} \right]}}{\beta } - \frac{{\delta_{t} }}{\beta }.$$

(7)

If food is a basic good with an income elasticity less than one (β < 0) and if the income effect is larger than substitution effect for food consumption (γ < 0),⁸ and under the reasonable assumption that the mismeasurement in non-food is larger than in food products, the first term in (7) is negative and our bias can be considered a lower bound. In other words, our measure would be underestimating the bias in the CPI.

So far, we have just described the estimation methodology used in the previous works. However, due to data limitations, we need to introduce some changes in the estimation procedure. Argentina has relatively few consumption expenditures that are publicly available and, as we mentioned above, we only had access to the Survey of Household Expenditures of 1985/1986 (Encuesta de Gasto de los Hogares 1985/86, EGH85/86), the National Survey of Household Expenditures 1996/1997 (Encuesta Nacional de Gasto de los Hogares 1996/97, ENGH 96/97), and National Survey of Household Expenditures 2004/2005 (Encuesta Nacional de Gasto de los Hogares 2004/05, ENGH 04/05). The EGH 85/86 took place in the city of Buenos Aires and its metropolitan area. For the ENGH 2004/05, we only have data for the city of Buenos Aires.

As a result, our data include only two regions, and thus, Eq. (3) becomes

$$\begin{aligned} w_{ijt}& = \varphi + \gamma \left[ {\ln \left( {1 + \prod_{Fjt} } \right) - \ln \left( {1 + \prod_{Njt} } \right)} \right] + \beta \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right)} \right] \hfill \\&\quad + \delta_{j} D_{j} + \sum\limits_{t} {\delta_{t} D_{t} } + \sum\limits_{x} {\theta_{x} X_{ijt} } + \mu_{ijt} \hfill \\ \end{aligned}$$

(8)

where D _j equals one for households belonging to the city of Buenos Aires.

In the literature, identification is obtained from regional variations, and thus, P _Fjt is the food price in region j, and P _Gjt is the general price index in region j. This gives several observations for each moment in time allowing estimating the coefficient on the time dummy. Unfortunately, we cannot follow this procedure here, because we only have price indexes for the entire sample (Buenos Aires and its metropolitan area). Even if we would have the regional price indexes, that of only two neighbor regions is clearly not good enough to identify the price relative effect and time dummy.

Fortunately, while the specification assumes two types of goods, food and non-food, in reality, there are many goods within each of those categories. In the data, it is not feasible to compute a family specific food price index, but this is feasible for the non-food bundle. Thus, we construct a relative price between the food and non-food baskets at the household level. More precisely, we have that

$$P_{Fit} = P_{Ft}$$

(9)

$$P_{Nit} = \sum\limits_{k} {\lambda_{ik} P_{kt} }$$

(10)

where λ _ik is the ratio of expenditure in item k over overall spending on non-food items, for household i at time t.

Considering that λ _ik can be estimated from the individual data from the surveys, we can now rewrite (3) as

$$\begin{aligned} w_{ijt} &= \varphi + \gamma \left[ {\ln \left( {1 + \prod_{Ft} } \right) - \ln \left( {1 + \prod_{Nit} } \right)} \right] + \beta \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right)} \right] \hfill \\ &\quad+ \delta_{j} D_{j} + \sum\limits_{t} {\delta_{t} D_{t} } + \sum\limits_{x} {\theta_{x} X_{ijt} } + \mu_{ijt} \hfill \\ \end{aligned}$$

(11)

where ($\prod_{Nit}$) is the cumulative percentage growth of the price of non-food between time 0 and time t at the household level. This equation provides the estimates, as shown in Table 3.

A consequence of this strategy, however, is that the price index estimated at the family level may be correlated with the error term of the equation, and may pose an endogeneity problem, for example, if this price level is correlated with the taste for food. To deal with this problem, an alternative is to assign an arbitrary value for γ and then compute $w_{ijt} - \gamma \left[ {\ln \left( {1 + \prod_{Ft} } \right) - \ln \left( {1 + \prod_{Nt} } \right)} \right]$ as the dependent variable to estimate the bias. This circumvents the need to use the individual price level altogether. However, where could we take this coefficient from? If we use the coefficient estimated in Eq. (1) from Table 3 (0.038), the total cumulative bias reaches 59.5%, which is very similar to the 61% from Table 3. But better still is to use an exogenous measure of this coefficient. Costa (2001) obtains a coefficient of 0.046 for the United States when identifying the effect of relative prices from differences in regions. Repeating the exercise with 0.046, the cumulative bias reaches 59.4%. Using twice the coefficient for the United States (0.092), the cumulative bias reaches 58.9%. The main reason why changes in the γ coefficient do not significantly alter the results is that relative prices have not changed too much. Figure 2 shows the evolution of the relative price of food in terms of the general level between 1985 and 2005.

Fig. 2

Relative price of food in terms of CPI (Jan-1985 = 100). Source Own computations based on Índice de Precios al Consumidor en el Gran Buenos Aires (INDEC)

Because the price of food in terms of the CPI has fallen about 10% between the first and second surveys, and only 4% between the first and the third, to significantly alter the results, the coefficient should be extremely large. For example, to reduce the cumulative bias to half (i.e., to about 30%), the coefficient should be more than 40 times the estimated coefficient for United States. In short, our results appear to be extremely robust, independently of the methodology adopted.

Trebon (2008) has suggested that economies of scale in each household may affect the share of food to non-food and suggests a correction based on introducing the household size interacted with the time dummies (that identify the bias). In other words, he suggests estimating

$$\begin{aligned} w_{ijt} &= \varphi + \gamma \left[ {\ln \left( {1 + \prod_{Ft} } \right) - \ln \left( {1 + \prod_{Nit} } \right)} \right] + \beta \left[ {\ln Y^{pc}_{it} - \ln \left( {1 + \prod_{Gt} } \right)} \right] \hfill \\ &\quad+ \delta_{j} D_{j} + \sum\limits_{t} {\delta_{t} D_{t} } + \sum\limits_{t} {\psi_{t} (D_{t} *hhsize)} + \sum\limits_{x} {\theta_{x} X_{ijt} } + \mu_{ijt} . \hfill \\ \end{aligned}$$

(12)

While Trebon finds that this correction reduced CPI biases by as much as a half relative to the findings in Costa (2001) and Hamilton (2001) for the US, Sect. 3 shows that in our case, this correction does not change things.

Income distribution effects

Following Carvalho Filho y Chamon (2012), we explore also the possibility that the amount of bias may change along the Engel curve thus allowing estimating different mismeasurements in earnings growth for different income levels. Using a semi-parametric specification and assuming, as before, that the biases are the same for the food and non-food bundles, we have that

$$\begin{aligned} w_{ijt} &= \varphi + \gamma \left[ {\ln \left( {1 + \prod_{Ft} } \right) - \ln \left( {1 + \prod_{Nit} } \right)} \right] \hfill \\ &\quad+ f_{t} \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Git} } \right)} \right] + \sum\limits_{x} {\theta_{x} X_{ijt} } + \mu_{ijt} . \hfill \\ \end{aligned}$$

(13)

The function $f_{t} \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Git} } \right)} \right]$ may be estimated non-parametrically using the differencing method of Yatchew (1997).To apply this method, we sort observations by income. The difference between two observations can be written as

$$\begin{aligned}& w_{ijt} - w_{i - 1jt} = \varphi + \gamma \left\{ {\left[ {\ln \left( {1 + \prod_{Ft} } \right) - \ln \left( {1 + \prod_{Nit} } \right)} \right] - \left[ {\ln \left( {1 + \prod_{Ft} } \right) - \ln \left( {1 + \prod_{Ni - 1t} } \right)} \right]} \right\} \hfill \\&\quad + f_{t} \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Git} } \right)} \right] - f_{t} \left[ {\ln Y_{i - 1t} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Gi - 1t} } \right)} \right] \hfill \\&\quad + \sum\limits_{x} {\theta_{x} \left( {X_{ijt} - X_{i - 1jt} } \right)} + \mu_{ijt} - \mu_{i - 1jt} . \hfill \\ \end{aligned}$$

(14)

As we have sorted by incomes, incomes are pretty similar so

$$\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Git} } \right) \cong \ln Y_{i - 1t} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Gi - 1t} } \right).$$

(15)

Assuming that $f_{t}$ is a smooth function:

$$f_{t} \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Git} } \right)} \right] \cong f_{t} \left[ {\ln Y_{i - 1t} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Gi - 1t} } \right)} \right].$$

(16)

Therefore, Eq. (14) becomes

$$\begin{aligned} w_{ijt} - w_{i - 1jt}& = \varphi + \gamma \left\{ {\left[ {\ln \left( {1 + \prod_{Ft} } \right) - \ln \left( {1 + \prod_{Nit} } \right)} \right] - \left[ {\ln \left( {1 + \prod_{Ft} } \right) - \ln \left( {1 + \prod_{Ni - 1t} } \right)} \right]} \right\} \hfill \\&\quad + \sum\limits_{x} {\theta_{x} \left( {X_{ijt} - X_{i - 1jt} } \right)} + \mu_{ijt} - \mu_{i - 1jt} . \hfill \\ \end{aligned}$$

(17)

Note that Eq. (17) is a linear function [with coefficients identical to those of (13)], so that we can consistently estimate it by OLS, and construct the linear part of the prediction of w _ijt , called $\hat{w}_{ijt}$, to arrive to

$$w_{ijt} - \hat{w}_{ijt} = f_{t} \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Git} } \right)} \right] + \mu_{ijt} .$$

(18)

If we take the right side of Eq. (18) as a dependent variable, we can estimate Eq. (18) by any common non-parametric method, and we choose to estimate it by local weighted regression method.After estimating $\hat{f}_{t}$, the cumulative bias may then be computed as the value of E _Git that solves for each household i at time t, the following equation:

$$\hat{f}_{t} \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right) - \ln \left( {1 + E_{Git} } \right)} \right] = \hat{f}_{0} \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right)} \right].$$

(19)

Intuitively, we may think that if the function f is constant in time, the value of f for a given income level must be the same independently of the time period used for its estimation.

To estimate the cumulative bias for households at time t, we went through the following steps. First, we selected the real income of households at time 0 that had $\hat{f}_{0}$ near the value estimated for each household at time t (that is $\hat{f}_{t}$). In fact, we selected two incomes at time 0 for each household at time t (those with income that were immediately higher and lower in terms of $\hat{f}$). Second, we computed the difference in real income between the two selected households. Third, we distributed linearly the difference according to the number of households from time t contained between the higher and lower bounds selected above (in terms of $\hat{f}$) from households at time 0. Fourth, we computed the real income from household in time t that it should have as per its share of food, adding to the income of lower (in terms of $\hat{f}$) the difference computed before. Fifth, we computed the bias from household i at time t, using the real income from household at time t, and the real income that it should as per its share of food. More precisely, what we do is to compute

$$E_{Git} = \exp \left[ {\ln Y_{it} - \ln \left( {1 + \prod_{Gt} } \right) - \left[ {\ln Y_{i0}^{{\hat{f}_{0}^{1} }} + \frac{{\left( {\ln Y_{i0}^{{\hat{f}_{0}^{2} }} - \ln Y_{i0}^{{\hat{f}_{0}^{1} }} } \right)}}{H}*h} \right]} \right] - 1.$$

(20)

Given that $Y_{i0}^{{\hat{f}_{0}^{1} }}$ is the income of the household with the lowest closest $\hat{f}_{0}$ to the household i at time t, and $Y_{i0}^{{\hat{f}_{0}^{2} }}$ is the income of the household with the highest closest $\hat{f}_{0}$ to the household i at time t, H is the number of households at time t that has an $\hat{f}_{1}$ between $\hat{f}_{0}^{1}$ and $\hat{f}_{0}^{2}$ and $h = 1 \ldots H$ is the order of these households sorted by $\hat{f}$.

Results

Data

As we mentioned above, Argentina has relatively few consumption expenditures that are publicly available. Thus, we only had access to the Survey of Household Expenditures of 1985/1986 (Encuesta de Gasto de los Hogares 1985/86, EGH85/86), the National Survey of Household Expenditures 1996/1997 (Encuesta Nacional de Gasto de los Hogares 1996/97, ENGH 96/97), and the National Survey of Household Expenditures 2004/2005 (Encuesta Nacional de Gasto de los Hogares 2004/05, ENGH 04/05). The EGH 85/86 took place in the city of Buenos Aires and its metropolitan area. For the ENGH 2004/05, we only have data for the city of Buenos Aires.

We start our analysis of this data in Fig. 3, with a brief illustration of some basic statistics for the three household surveys. There, we show expenditure shares on different types of goods as a function of income levels. Each curve depicts one the three surveys for which we have data.

Fig. 3

Source Own elaboration based on expenditure surveys

Basic Statistics

Some straightforward conclusions may be inferred from the figure. First, the relation between food and income is negative, indicating that food is a basic good. More so, not only can we see that the share of food falls systematically as we move upwards in income, but that the shares fall for each later survey. To the extent that Engel curves are stable, this would clearly indicate that income levels increased uninterruptedly throughout the period. With the exception of housing, the shares of the remaining composite goods tend to increase with income. For a non-Argentinean, perhaps, it is surprising how much Education expenditures increase with income, a result that originates on the much higher use of private education among higher income levels.

To check the consistency and quality of the data, Table 1 shows the main demographic characteristics for each survey. The table shows that the data are fairly homogenous, but that over the period of the three surveys, Argentina has experienced a reduction in household size, a larger share of females in the labor force, and an increase in the number of single parents’ households.

Table 1 Demographics

To compare the nominal variables, we use the CPI to bring them to a comparable basis (in the table, all prices are expressed in 1999 pesos). The table shows that, according to the data, income levels decrease quite sizably between the 85/86 wave and the 96/97 sample. Notice that during the same period (see Fig. 3), there is an unambiguous decline in the share of food for all income groups. It is this inconsistency (lower food share comes with higher, not lower income) that will be at the crux of our estimation of the CPI bias during this period. For the later period, incomes increase, while the food share continues to decline, so at this stage, it is unclear whether a bias exists or not.

Table 2 shows that data for the city of Buenos Aires only provide an even more striking finding: household income has fallen throughout in spite of declining food shares.

Table 2 Demographics, city of Buenos Aires only

Estimating biases

To estimate the bias in CPI measurement, we use Eq. (11) which allows to estimate the magnitude (as well as the statistical significance) of the bias. The results are shown in Table 3.

Table 3 CPI bias, robustness check

Columns (1) and (4) use expenditures as a proxy for permanent income. Columns (2) and (5) use current income. Columns (3) and (6) use current income as an instrument for expenditure. The second set of regressions adds a number of additional control variables.

The table shows that if we compare the 85/86–96/97 periods, we see similar measure biases across the estimations, with a cumulative bias of the order of between 58 and 65%. The large bias indicates an overestimation of the CPI of a whopping range between 7.5 and 9.2% per year. Considering that it is likely that the bias may not have occurred uniformly across years, this suggests a massive overestimation in particular years. On the contrary, when comparing the 96/97 and 04/05 periods, we find a relatively small bias, which is also, typically, not significant.

Considering the whole sample and spanning the entire democratic period, we find an average bias of between 4.3 and 5.7%, indicating that real earnings may have grown by this additional amount during the period, similar to the numbers found for Brazil and much larger than the numbers found for the US.

The fact that the overestimation of the CPI takes place in the first part of the sample has to do, in our view, with the massive change occurred in Argentina as a result of the opening up of the economy in the late 80 s and early 90 s. While this time dimension will have to be tested and evaluated in future work, we present here an “illustration” of the effect by showing the change in varieties of commercial retailing in Argentina between the 1980s and the 1990s. In the 1980s, varieties were minimal and the quality relatively poor. We believe that visualizing the difference may help in understanding the magnitude of the potential gain. Figure 4 shows three pictures. One corresponds to the typical grocery store in the 1980s. The shelves show how limited the variety offered was. The two other pictures show a minimarket and a large chain store supermarket (“hipermercado”, as is known in Argentina) in the 2000s. While the change depicts the food component, similar changes were observed throughout this period across all consumption baskets.

Fig. 4

Source Stock photos

Variety in food retailing

One potential criticism of our results is that the food item is composed of products consumed both inside and outside the household. Since goods consumed outside home may include some service component and thus not be entirely subject to the pattern of the typical Engel curve, Table 4 shows the results using only the share of food at home as the dependent variable. As can be seen, the results are similar to those obtained previously.

Table 4 CPI bias robustness check

Table 5 shows the results including the specification suggested by Trebon (2008), which introduces a term to take into account the effect on food shares of household size. A quick inspection of the table, however, reveals that in the case of Argentina, this also does not modify the numbers in any significant manner.

Table 5 Trebon critique

An additional robustness test includes using only the data for the city of Buenos Aires. The results are similar to those estimated previously, and for brevity are not shown here.

Income distribution effects

The Engel curve that we estimate in the parametric version of Eqs. (11) and (12) assumes that the bias is the same across all income levels. If so, the bias is by definition constrained to be neutral from an income distribution point of view. However, this may not be the case. Thus, the more flexible estimation procedure (such as the non-parametric estimation of Yatchew 1997, explained in Income distribution effects Section) allows to test the validity of this assumption, permitting for an estimation of an Engel curve shift that may differ at different income levels.

The result of this more flexible estimation procedure, shown in Figs. 5 and 6, confirms that, in fact, the biases are dramatically different across income levels, being much larger at lower income levels, as shown by the much larger movement in the curve at the lower part of the scale. Figure 5 shows the estimated Engel curves in log terms, whereas Fig. 6 relates the bias to income levels directly.

Fig. 5

Source Own elaboration based on expenditure surveys

Individual effects (log version)

Fig. 6

Source Own elaboration based on expenditure surveys

Individual effects

This result is similar to the one obtained by Carvalho Filho and Chamon (2012) for Brazil.

As we mention in the methodological section, once we compute the bias at different income levels, we can estimate an adjusted income (see Eq. 15). Table 6 shows the basic statistics for the bias in real income measures, at each income level, when comparing the base year with the two following periods.

Table 6 Biases by income level

On average, the bias estimated is fairly similar, though somewhat larger, to that obtained in Tables 3, 4, and 5. However, as can be seen in Table 6, this average hides a large heterogeneity across income levels.

Once we compute the bias, we can correct individual income levels using individual biases. Thus, we re-estimate the corrected income using the formula:

$$RY*_{it} = \frac{{RY_{it} }}{{\left( {1 + E_{it} } \right)}}$$

where $RY_{it} = \frac{{Y_{it} }}{{\left( {1 + \prod_{Gt} } \right)}}$ is the real income, and $RY*_{it}$ is the bias-corrected real income.While we can compute E _it only for the common support area,⁹ we use the minimum (maximum) value of E _it to correct real income in observations at time t that have a real income higher (lower) than the maximum (minimum) real income in the common support area.¹⁰

Figure 7 shows the mean values for income and expenditure deflated after correcting for the bias in the CPI.¹¹ In the figure, we show the numbers taking 85–86 as base years. While the official data show a declining real income, adjusting for real purchasing power shows a significant increase in average real expenditure and income.

Fig. 7

Source Own elaboration based on expenditure surveys

Corrected income levels (mean values). Note: values are obtained taking 1985–1986 as bench mark and adjusting 1996–1997 and 2004–2005 incomes/expenditures by the corrected increase in purchasing power

Figure 8 shows the Gini coefficients both for the official numbers and for those computed using the corrected real income numbers. Again, we take as benchmark the 85–86 values. It is important to notice that we are not making a statement on the actual level of inequality (had we taken the 2005–2006 period as benchmark, the corrected value of the Gini would have coincided with the official numbers for these years), but we are making a statement on the fact that during the 85–2006 period, we find a sizable reduction in income inequality in Argentina, which, again, contrasts starkly with official figures.

Fig. 8

Source Own elaboration based on expenditure surveys

Corrected Gini coefficients

Figure 9 shows Lorenz Curves and the bias-corrected versions for the 1996/1997 (left column) period and the 2004/2005 (right column) period, both for income (first row) and expenditures (second row). We can see that bias-corrected curves strictly dominate not corrected curves, so we can reproduce the same results of Fig. 9 using any inequality index.

Fig. 9

Source Own elaboration based on expenditure surveys

Original and modified Lorenz curves

To complete this presentation of our findings, Fig. 10 mimics the same graphs, but for the distribution of income and expenditure levels (left and right columns, respectively), comparing the original data with the bias-corrected data (upper and lower rows, respectively).

Fig. 10

Source Own elaboration based on expenditure surveys

Income distribution. Note: values are obtained taking 1985–1986 as bench mark and adjusting 1996–1997 and 2004–2005 incomes by the corrected increase in purchasing power

Conclusions

This paper has estimated the CPI measurement bias for Argentina during its recent democratic period. While we use a methodology that unveils the bias from the inconsistencies between the assumption of stable Engel curves and the evolution of the share of food in expenditures, we innovate in that we obtain identification from individual differences in the consumption bundles and price indexes at the household level, thus being able to estimate the bias with data from only one region, something that had not been done in previous work.

The findings are striking. Argentina’s democracy has experienced a larger (much larger) raise in real expenditure levels than previously thought, and has improved its income distribution.

The bias in expenditure levels arises primarily between 84/85 and 96/97. It is difficult with further data to estimate when the bias may be originating. The years 84/85 were a time of very high inflation; thus, the data may be underestimating the level of regressivity in the income distribution those years. In addition, the late 80s and early 90s showed a period of significant opening up of the economy that led to a significant increase in income levels. Because openness comes with large changes in the quantity and quality of available products, it is not surprising that during this period, we may have experienced substantial increases in economic well-being not fully reflected in the standard statistics.

The second period is a bit more puzzling. While the data suggest an overestimation of the CPI, the level of this overestimation appears to be small. However, the bias in income distribution appears to be larger. This is puzzling, because the later period sees a rising inflation, indicating, a priori, that there should be deterioration in the income distribution levels.

All in all, our conclusion, however, is that Argentina’s democracy has allowed for a much brighter performance in economic terms than it is usually credited for. Far from the typical pessimism that permeates the recollection of Argentina’s history and Argentina’s present, we provide an optimistic view of the last 25 years, which we hope will be the beginning of a brighter XXIst century for the country and the region.

Appendices

Appendix A: The data

To run our estimations, we use the individual data points for the (EGH 85/68), (ENGH 96/97), and (ENGH 04/05) constructed by the Instituto Nacional de Estadísticas y Censos (INDEC). The EGH 85/86 covers only the city of Buenos Aires and its metropolitan area. As a result, we only considered the same region for the ENGH 96/97. For the ENGH 04/05, we only had access to the data for the city of Buenos Aires. This appears to have no fundamental effect on our estimations. Running all the estimates just for data from the city of Buenos Aires gives virtually identical results.

The price index used is the CPI for the greater Buenos Aires area, 1999 = 100.

The EGH 85/86, ENGH 96/97, and ENGH 04/05 provide data for 2717, 4907 years 2841 households each,¹² reporting income and expenditures (itemized by groups) as well as the typical demographic characteristics.

Because the INDEC does not provide information about inconsistent observations in the survey, we keep out of the analysis a few observations that seem to be inconsistent in expenditure. We take out households that:

• Do not report total expenditure or report a negative value (1 in EGH 85/86, 6 in ENGH 96/97 and 10 in ENGH 04/05).

• Report a very low total expenditure (lower than 100 pesos of 1999) and a share of food lower than 50% (19 in ENGH 96/97 and 3 in ENGH 04/05).

• Do not report expenditures in food (26 in EGH 85/86, 49 in ENGH 96/97 and 31 in ENGH 04/05).

In addition, we found 58 households in ENGH 96/97 and 93 households in ENGH 04/05, with negative consumption in at least one expenditure group. We have set at zero the level corresponding to negative expenditure.

Needless to say, these obvious mistakes are numerically insignificant, and do not change the main results.

In the ENGH 96/97 and the ENGH 04/05, there is information about households with imputed income and expenditure,¹³ but not in the EGH 85/86, as a consequence. we will assume that the imputation method used by the INDEC is valid and similar across surveys.

The EGH 85/86 was conducted between July 1985 and June 1986. The base indicates the quarter in which each household has been surveyed. Based on this information, we have paired the data with the corresponding CPI level (and its categories) corresponding to the average for each quarter.

ENGH 96/97 took place between February 1996 and March 1997, but numbers have been taken nominal values relative to the average CPI during the period, as there is no information as to the specific quarter in which the survey was conducted. Fortunately, this is a very low inflation period, and therefore, whatever mistake arises from this must necessarily be minimal.¹⁴

ENGH 04/05 took place between October 2004 and December 2005. The base indicates the quarter in which each household was surveyed, and therefore, the procedure followed is similar that used for EGH 85/86.

Appendix B: Additional tables

See Tables 7, 8, 9, and 10.

Table 7 Basic statistics of additional variables used for regressions 4–6 (Tables 3, 4 and 5)

Table 8 Table 3 coefficients

Table 9 Table 4 coefficients

Table 10 Table 5 coefficients