Working Paper 2008 | 30 Business cycle accounting for Argentina utilizing capital utilization Tiago Cavalcanti University of Cambridge Pedro Elosegui / George McCandless Emilio Blanco BCRA January, 2008 Banco Central de la República Argentina ie | Investigaciones Económicas January, 2008 ISSN 1850-3977 Electronic Edition Reconquista 266, C1003ABF C.A. de Buenos Aires, Argentina Phone: (5411) 4348-3719/21 Fax: (5411) 4000-1257 Email: investig@bcra.gov.ar Web Page: www.bcra.gov.ar The opinions in this work are an exclusive responsibility of his authors and do not necessarily reflect the position of the Central Bank of Argentina. The Working Papers Series from BCRA is composed by papers published with the intention of stimulating the academic debate and of receiving comments. Papers cannot be referenced without the authorization of their authors. Business cycle accounting for Argentina utilizing capital utilization Tiago V. de V. Cavalcantia Pedro Eloseguib George McCandlessb Emilio Blancob University of Cambridgea Banco Central de la República Argentinab January 31, 2008 Abstract We use a variation on the business cycle accounting method of Chari, Kehoe and McGrattan [3] to study the business cycle in Argentina from 1972 to 2006. We use capital utilization as a household decision variable to be able to better extract the wedge that functions as a tax on capital. Applying the model to Argentina, we …nd that all four wedges are important in explaining the evolution of output over this period (although net exports is the least important). The major political subperiods can be charaterized by the relative importance of each wedge. We compare the results of this technique to the standard narrative. JEL classi…cations: E22, E32, N16 1 Introduction Using a prototype growth model, Chari, Kehoe and McGrattan [3] (henceforth, CKM) propose an accounting procedure for guiding researchers in developing and analyzing quantitative models of economic ‡uctuations. This procedure uses real data together with the equilibrium conditions of a prototype growth model to measure four wedges that are explained by the variables of the model. These wedges can be viewed as “distortions” from a perfectly competitive economy and represent the set of policies and institutions which a¤ect productivity and factors input1 . The authors show that a large class of economic models, including The opinions expressed here are those of the authors and should not be taken to represent those of the Banco Central de la República Argentina. Sabastian Katz and Carlos Zarazaga provided helpful comments. 1 These are wedges in the sense that they measure how di¤erent the economy is from a closed, non-stochastic, perfectly competitive economy. 1 those with various additional features (e.g., …nancial frictions, nominal rigidities, entrepreneur decisions, and monetary shocks) are equivalent to a prototype growth model with time-varying wedges. Institutions, public policies, …scal, monetary, income and labor policies a¤ect the four wedges and therefore a¤ect the allocations of capital and labor, net exports and productivity in the economy. In their standard model for business cycle accounting, CKM use four wedges that are explained by four variables of the model. The period t value for three of the wedges, government expenditures and net exports, total factor productivity, and an implicit labor tax, are found using the model and period t data. The fourth wedge, an implicit capital tax, is found using the intertemporal relations of the model and its aggregate e¤ects on output are found as a residual of the e¤ects of the other three wedges. For the US data used by CKM, the capital wedge is found to be quite small. It is not clear if this is a characteristic of the US data they are using or a result of their calculation technique. Making use of capital utilization data provides an alternative way of …nding the capital wedge. Capital utilization information can be used directly in the production function (since it is utilized capital that creates production) and in the household budget constraint (since it is utilized capital that earns rents). In addition, it is quite natural to think of depreciation as being a function of capital utilization, since the more that capital is utilized in a period, the more likely it is to wear out and need replacement parts or maintaince. Given these characteristics of capital utilization, it becomes a household choice variable, households choose how much capital to send to the market each period, and adds an additional …rst order condition to the model. One can use this extra …rst order condition to extract, using period t data, a period t capital wedge. CKM do provide an alternative speci…cation in which they use variable capital utilization to adjust the capital stock. What they do not do, and what we do here, is make capital utilization a choice variable for the households, get the additional …rst order condition and use this extra equation to extract the capital tax and construct the capital wedge. Once the wedges are constructed it is possible to assess what fraction of output ‡uctuation can be attributed to each wedge separately and in combination. In this paper, we apply a model with endogenous capital utilization to the Argentine economy2 from 1972 to the end of 2006. The accounting procedure allows us to analyze the Argentine economy during a time period subject to several structural breaks and where these can be associated with di¤erent economic regimes. The wedge decomposition helps to explain several features that characterize each of the di¤erent economic regimes. Emphasis is given to the evolution of labor and capital wedge as well as factor productivity. In this model, total factor productivity is based on labor and utilized capital. It should be noted that many of the nominal and real frictions that are often added to standard RBC models can be captured with one of our four wedges. 2 Methods similar to those of the basic CKM method have been applied to both the modern Japanese economy and the prewar economy by Hayashi and Prescott [8] and [9] and to the French economy by Prescott [13]. Cole and Ohanan [4] have applied the basic model to the U.S. and U.K. and Cavalcanti [2] to Portugal. 2 CKM make much of this in their paper. Determining which wedges are most important for explaining the Argentine business cycle is a …rst step towards determining which frictions are most important in generating a productive RBC model of Argentina, especially if these models are to be used for forecasting or for analyzing monetary policy and monetary transmission channels. 2 The model with capital utilization We assume that capital depreciation, t , is a function of capital utilization, Let depreciation in period t be given by t = exp (a ( t. )) t where is the average depreciation rate, a measures how strongly depreciation responds to capital utilization, and is the average utilization. This formulation produces the average depreciation rate whenever utilitization is average and makes depreciation an increasing function of utilization. With this addition equation, we write the CKM model as U = Et 1 X i u (ct+i ; ht+i ) i=0 subject to ct + kt+1 1 exp (a ( t h t )) kt = 1 wt ht + 1 k t rt t kt + Tt and a production function of the perfectly competitive …rms of t Yt = At ( t Kt ) Ht 1 : Expected utility for a representative household is a function of a constant discount factor, , and an in-period sub-utility function that depends on the household’s consumption, ct , and on the labor it supplies, ht . The household accumulates capital, kt , and pays lump sum taxes, Tt . In the household budget constraint, the variables ht and kt are the implicit taxes being imposed by policies that are not oberved on labor and capital, respectively. Output of the one good is competitive and depends on the level of technology (total factor productivity), At , the aggregate capital being used, t Kt , and the aggregage supply, Ht , adjusted by the growth in labor augmenting technology, t . The …rst order conditions for the …rms (and the factor markets) are wt = (1 ) t t At ( t Kt ) and rt = At ( t K t ) 1 t Ht Ht 1 : Note that the rental rate calculated here, rt , is that paid on utilized capital and not on total capital, and this is taken into account in the household budget constraint. 3 The …rst order conditions for the households (who, in period t, choose ct , ht , kt+1 , and t ) are uh (ct ; ht ) uc (ct ; ht ) k t 1 rt uc (ct ; ht ) = h t 1 wt = a exp (a ( = )) t Et uc (ct+1 ; ht+1 ) exp a 1 1 t+1 k t+1 rt+1 Three period t shocks (or wedges, are they are called in the literature), ht , and At , can be found in terms of period t observations on the data from the …rst two of these …rst order condition and the production function as k t, uh (Ct ; Ht ) uc (Ct ; Ht ) a exp (a ( t = )) = h t 1 k t 1 t ) Yt t At ( t K t ) 1 At ( t K t ) and At = (1 t Ht 1 ; : 1 ( t Kt ) ( t Ht ) Ht ; (1) (2) (3) In addition, A resource constraint that must hold is Ct + Kt+1 + Gt = Yt + 1 exp (a ( )) Kt : t Notice that with the extra choice variable of capital utilization, it is possible to extract all four of the time t shocks using just time t data on Ct , Ht , Kt , Gt , and t . Gt come directly from the data on government expenditures and net exports. At is calculated given output, the production function and the data on the capital stock, capital utilization and labor (along with labor productivity growth). The wedges, ht and kt , are then calculated from the two remaining …rst order conditions. In the actual implementation, we take K0 as given and use the data on gross investment, It ; and capital utilization to calculate the sequence of capital stock from Kt+1 = 1 exp (a ( t )) Kt + It : Note that the capital stock path depends crucially on the parameter a of the depreciation function. The sub-utility function3 used for implementation is u (Ct ; Ht ) = log Ct + 3 The log(H Ht ) subutility function used in Kydland and Zarazaga [11] is ct h ht 1 1 ; 1 which give exactly the same marginal conditions when 4 = 1 : t+1 so uh (Ct ; Ht ) = uc (Ct ; Ht ) Ct : H Ht With this Cobb-Douglas type utility function, we …nd the four wedges. To …nd the impact of each of the wedges independently, we …x the values of three of wedges at their average value, use the time series for the capital stock that we found above, and use a nonlinear system solver in Matlab to …nd the values of Ct , Ht , t , and Yt as the zeros for the system Ct + 1 Ht 0 = 0 = a exp (a ( H t h t )) 0 = At 0 = Yt + 1 exp (a ( (1 ) 1 k t t t At ( t K t ) At ( t Kt ) 1 Ht t Ht 1 Yt 1 ( t K t ) ( t Ht ) )) Kt t Ct Kt+1 Gt : In addition, we calculate the time path for output that comes from using the total factor productivity wedge and the capital tax wedge together, keeping the other two wedges at their average value. 3 The data and the wedges The data are quarterly and run from 1972:1 to 2006:4, a relatively long period of time for studies of Argentina. The data for the four macroeconomic variables are quarterly, expressed in per capita terms in prices of 1993, and are shown in Figure 1 and those for utilization and labor (calculated from hours worked per capita) are given in Figure 2. The period has been characterized as one of economic and political turbulence in Baer, Elosegui and Gallo [1]. The …rst decade, from 1972 until 1982, has been characterized as a series of so called "stop and go" policies and the "plata dulce" period, with a seriously overvalued exchange rate, that ended in a debt crisis. The lost decade of the 80s can be characterized by systemic growing in‡ation and a …nal hyperin‡ation period mixed with a number of failed attempts to introduce stabilization programs. The problem of in‡ation was only dominated in 1991, with the introduction of the "Convertibility Plan", a currency board system that characterized the decade until its collapse at the end of 2001. Since 2003, the economy has been recovering from that crisis. The data encompass a number of policy regimes, substantial variation over business cycles and a virtually ‡at long run trend. The construction of the wedges should be able to help us to analyze each of these periods and the impact of their policies on capital, labor and productivity. We use series on the real, per capita value of output, consumption (in this case, the combined private and public consumption), net exports, investment, 5 Y 2000 1500 C 1000 I 500 X-M 0 1970 1975 1980 1985 1990 date 1995 2000 2005 Figure 1: Deseasonalized macro data: Y , C, X 2010 M , and I 0.9 u 0.8 0.7 0.6 0.5 H 0.4 0.3 0.2 1970 1975 1980 1985 1990 date 1995 2000 2005 Figure 2: Data for utilization and labor 6 2010 depreciation 0.017 0.0165 0.016 0.0155 0.015 0.0145 0.014 0.0135 0.013 1970 1975 1980 1985 1990 1995 2000 2005 2010 Figure 3: Depreciation normalized hours worked, and an index for capital utilization4 . The pre-1993 data do not come with private and public consumption separated, so we assume that government consumption enters the utility function mixed together with private consumption. Since government consumption is not available for the whole period, the variable Gt in the above equations only contains net exports. The series for the capital stock is constructed using Kt+1 = Yt + 1 exp (a ( t )) Kt Ct Gt ; using a value of 21490 for the 1980:1 value of the per capita capital stock, calculating backwards to 1972:1 and forward to 2006:4. In the calculations we = :75, and a = 1. The used = :57, = :06174=4, t = 1, = 0:6878, depreciation series that comes from the function (in quarterly terms) is shown in Figure 3. The time series for capital (measured in quarterly, per capita, in prices of 1993 terms) that we get from these calculations is shown in Figure 4. For the years where the two studies overlap, our time path for the capital stock is generally similar to the one in Kydland and Zarazaga [10]. Using this data, we …nd At using equation 3, ht , using equation 1, and kt from equation 2. The net export wedge, Gt , comes directly from the data and are per capita, measured in prices of 1993. The wedges that result from these calculations are shown in Figure 5. The labor and capital wedges are measured as the fraction of income from each source paid as implicit taxes. 4 The index on capital utilizaton comes from FIEL (Fundación de Investigaciones Económicas Latinoamericanas). The other data comes from INDEC (Argentina’s Instituto Nacional de Estadística y Censos) except that output is de…ned from the national income accounting identity, Y = C + I + G + netX. Real variables are in terms of 1993 prices and were linked to earlier real series by averaging over overlapping dates. 7 2.3 x 10 4 2.2 2.1 2 1.9 1.8 1.7 1.6 1970 1975 1980 1985 1990 1995 2000 2005 2010 Figure 4: Calculated capital stock XM A 150 14 100 13 50 12 0 11 -50 -100 1970 10 1980 1990 τ 2000 2010 1970 1980 1990 τ k 0.7 2000 2010 2000 2010 h 0.4 0.6 0.3 0.5 0.2 0.4 1970 1980 1990 2000 0.1 1970 2010 1980 1990 Figure 5: Four wedges for Argentina 8 Efect of XM on Y 2000 1800 1600 1970 1975 1980 1985 1990 1995 2000 2005 2010 2000 2005 2010 2000 2005 2010 2000 2005 2010 Efect of A on Y 2000 1500 1000 1970 1975 1980 1985 1990 1995 Efect of τ k on Y 2200 2000 1800 1600 1400 1970 1975 1980 1985 1990 1995 Efect of τ h on Y 2000 1800 1600 1400 1970 1975 1980 1985 1990 1995 Figure 6: E¤ects of the four wedges on output Taking each of these wedges separately and using the mean value for the remaining three, we calculate the importance each of the wedges has had in producing cycles in the Argentine economy. The contribution to Argentine output of each of the four wedges over the last 35 years are shown in Figure 6 and are quarterly, per capita, in prices of 1993. In Figure 6, one can see that total factor productivity wedge, At , captures much of the long term cycle that is comprised of the decline of the 1980’s, the growth of the 1990’s and some of the crash and recovery of the 2000’s.5 It is interesting to note that there is a peak in the total factor productivity factor in 2002, since the decline in labor and capital utilization meant that the factors 5 Using this method on data with the seasonals left in puts most of the seasonal variation in total factor productivity. The biggest seasonal variation for Argentina is a sharp fall in output caused by vacations during the …rst quarter of each year. Net exports explain a bit of the seasonals caused by primary product exports. 9 still being used were highly productive. Total factor productivity "explains" a deeper cycle than is seen in the data and is mostly compensated by the e¤ects of the capital tax. The part of the business cycle explained by net exports is quite small, although it does capture a slow decline until around 1993 and a gradual rise thereafter. The relatively small contribution that net exports makes in explaining the cycles is at odds with claims that real international shocks are important for explaining the Argentine business cycle. The bottom two graphs in Figure 6 show the e¤ects of kt and ht . Neither of these explains much of the long term cycle and the part of output that is explained by the capital wedge is negatively correlated with that cycle for most of the period. The capital wedge captures a number of important events in Argentine economic history. Capital clearly su¤ered from the hyperin‡ations of 1989 and 1991 as well as the in‡ation leading up to the Austral Plan in 1985 and the debt crisis of 1981-2. Interestingly, capital did not seem to su¤er during the Tequila crisis around 1995, where most of the changes seem to be re‡ected through the labor wedge. Capital also su¤ered from the 2002 crisis but recovered to a level similar to that before the crisis. This last suggests that the output growth in the last few years of the period under study does not seem to have come from policies that bene…t capital. The labor wedge indicates that the part of output explained by labor had a long slow decline during the 1980’s, although it was much less sensitive to business cycles than capital, su¤ered in the hyperin‡ations at the beginning of the 1990’s, recovered a bit after that only to be heavily hit by the Tequila crisis of 1995. This is consistent with the big jump in unemployment observed in Argentina in the mid-1990’s. The 2001 2 crisis initially had as big an impact on labor as it did on capital but the output e¤ects of the labor wedge continue to show improvement up to the last observations available. Figure 7 shows how well the combined e¤ects of the total factor productivity wedge, At , and the capital tax wedge, kt , explain output. As can be seen in the center two graphs of Figure 6, the e¤ects of total factor productivity alone is a very large long cycle and that of the capital tax alone is a similar long cycle but going in the opposite direction. When the two wedges are taken together, they capture a very large fraction of the movement in output. Given how our accounting method is constructed, the little that is left to explain must be captured by the other two wedges. A second way of studying how well each wedge explains output is to look at the correlation coe¢cient between output and the part of the business cycle explained by each wedge separately. Table 1 shows these correlations coe¢cients. The upper number in each box give the correlation coe¢cient between the variable to the left and that above from the data in levels. Since we cannot reject the hypothesis that output in Argentina has a unit root6 , the lower number in each box shows the correlation coe¢cients for the …rst di¤erences of the same variables. 6 Using an augumented Dickey-Fuller unit root test. 10 Y 2200 Y explained by A and 2100 τ k wedg es combined 2000 1900 1800 1700 1600 1500 Y 1400 1300 1200 1970 1975 1980 1985 1990 date 1995 2000 2005 2010 Figure 7: The combined e¤ects of the total factor productivity wedge and the capital tax wedge Y YXM YA Y k Y h Y 1:0000 1:0000 0:6111 0:5288 0:7971 0:5641 0:2439 0:0946 0:3282 0:4616 YXM 0:6111 0:5288 1:0000 1:0000 0:6663 0:4685 0:2582 0:0796 0:0941 0:0138 YA 0:7971 0:5641 0:6663 0:4685 1:0000 1:0000 0:7052 0:7729 0:1702 0:0483 Y k 0:2439 0:0946 0:2582 0:0796 0:7052 0:7729 1:0000 1:0000 0:4088 0:0794 Y h 0:3282 0:4616 0:0941 0:0138 0:1702 0:0483 0:4088 0:0794 1:0000 1:0000 Table 1: Correlations of output and the part explained by each wedge 11 One of the most interesting features of Table 1 is the correlation between the part of output explained by total factor productivity and that explained by the capital tax. This is strongly negative, whether measured in levels or in …rst di¤erences. As we showed above, these two wedges combined explain a large fraction of the movement in output. The path of output that results from the combined e¤ects of the total factor productivity wedge and the capital tax wedge has a 90:17% correlation with the output data when measured in levels and a 81:04% correlation when measured in …rst di¤erences. This result in the table suggests that in Argentina, whenever there is an increase in total factor productivity, it is met with increase in policies that operate as a tax on capital7 . Looking at the wedges themselves in Figure 5, the tax on capital moves very much like total factor productivity except during the hyperin‡ations of 1989 and 1991 when the tax on capital rises without a compensating rise in total factor productivity and in the Tequila crisis of 1995 when the tax on capital falls dramatically with only a small decline in total factor productivity. The e¤ects of each of the wedges on C, , and H are shown in Figure 8. As one might expect, given the importance of output in determining consumption, the e¤ect of each wedge on explaining consumption is very similar to its e¤ect on explaining output. The e¤ect of net exports, the unlabeled line in each graph, is small in all cases, although it is a bit more important, and seasonal, in explaining labor movements. In explaining capital utilization, it is not surprising that the capital wedge captures some important movements, especially the drop in 2002. Technology, A, captures long trends in the capital utilization as well as some seasonals. The labor tax wedge does not explain much of utilization, although it does reinforce the decline in the 2002 crisis. Most of the movement in employment is explained by the labor tax wedge. The net export wedge provides a bit of a seasonal and a bit of long term movement, however, this is small. The impact of the technology (A) wedge and the capital wedge on labor are very similar, so much so that they cannot be very well separated in the graphs. 4 More detailed historical analysis of the growth accounting exercise for Argentina Major crises have occurred with certain regularity in the period under consideration. A serious in‡ation in 1975-6 was followed by a military coup (a potential regime change8 ). A period of an overvalued peso (known as the Tablita) was followed by a foreign exchange crisis. The return to democracy witnessed high and rising in‡ation that in 1985 was met with the Austral plan. The transition from the 1980s to the 1990s su¤ered two hyperin‡ations. A banking crisis in 1995 resulted in the closing or consolidation of much of the banking system. 7 An alternative hypothesis is that whenever policies go against capital, …rms respond by increasing total factor productivity. 8 Zablotsky [14] looks at military coups in Argentina as regime changes with democratic periods raising taxes on land and military regimes lowering them. 12 C 2000 τk A 1500 1000 1970 1975 1980 1985 1990 1995 2000 2005 2010 2000 2005 2010 2000 2005 2010 u τk 1 A 0.8 0.6 0.4 1970 1975 1980 1985 1990 1995 H 0.4 0.35 A τh 0.3 τk 1970 1975 1980 1985 1990 1995 Figure 8: E¤ect of each wedge separately on C, , and H 13 The recession that began in late 1998 became a full blown banking and exchange crisis in 2002. Each of these crisis events can be noted in Figure 6, with greater or lesser short term impact on output. What is of interest is how these crisis di¤ered and what our wedges can tell us about how they were di¤erent. The return of Peron to power in 1973 is marked by a sharp decline in the labor tax and a relatively small decline in total factor productivity as seen in Figure 5. The labor tax returns to its original level with the military coup of 1976 and stays high for most of the time the military is in government (it falls at the end of the period as the military is leaving government). There is a steady fall in the capital tax over the military period that is accompanied by a secular decline in total factor productivity. The overvalued peso of the Tablita permitted capital accumulation and is observed in the positive e¤ect of net exports on output in the period just prior to 1980. This capital accumulation is also indicated in the increase in output from the capital tax but does not show up in total factor productivity. The low level of the tax on labor continues throughout all of the Alfonsin presidency. During the …rst part of this period, total factor productivity continues to fall and then stays low for the second half of the 1980’s (the so called "lost decade"). The tax on capital rises for most of this period, although with substantial variation. The e¤ects of net exports on output are relatively constant over the Alfonsin period. The hyperin‡ation of 1989 generates rises in net exports (as imports fall) and increases in the capital tax, but has little e¤ects on the labor tax or on total factor productivity. The hyperin‡ation in 1991, at the beginning of the Menem period, was somewhat di¤erent from that of 1989. The e¤ects of both the capital and labor wedge on output are negative and the tax on capital was particularly high. It is also one of the peaks of the net export wedge. The beginning of the convertibility period9 is marked by a fall in the tax on labor (and a increase in output from the labor wedge) and a secular rise in total factor productivity. The Tequila crisis of 1995 hit labor hard but had almost no e¤ect on the other wedges. This is consistent with the large increases in unemployment that accompanied and followed that crisis. The general trend of the recession of 1999 to 2002 and the recuperation after is captured in the net export wedge. The crisis of 2002 hits both the labor and capital wedges hard, with sharp increases in the wedges and sharp declines in the part of the business cycle explained by these wedges. These two e¤ects are so sharp that total factor productivity actually rises during this crisis. The recovery during the Kirchner years comes from a large reduction in the tax on labor, an increase in total factor productivity, and a decline in net exports. The tax on capital increases to a level similar to the Peron period at the beginning of the sample (the mid-1970’s). In Figure 6, the e¤ect of both the labor wedge and factor productivity are the highest over the sample and the e¤ect of net exports is close to the maximum. The e¤ect of the capital wedge is a small decline after the initial recuperation from the 2002 crisis. These results are 9 See Baer, Elosegui and Gallo [1] for a detailed explanation of the convertibility plan. 14 consistent with the perception that the Kirchner government looks to bene…t workers with policies that tend to increase wages, promote labor formalization and increase labor union powers while imposing increased tax pressures on the owners of capital and land. In a model as simple as this one, what we call the capital stock is comprised of physical capital and land. 5 On the importance of the capital stock path The calculation of the time path for capital is one of the more di¢cult aspects of this analysis. The capital path is important in determining two of the wedges, that of total factor productivity and the capital tax, and is important in determining the time path of output that comes from each of the wedges. Data on the capital stock is not good and capital stock series are created from the investment series, an estimate for some initial capital stock, the average depreciation rate, and the function of the capital utilization rate. The time path for capital that comes from the investment series is sensitive to the choice of an average depreciation rate and to the initial capital stock, although the importance of the initial capital stock diminishes with time. Figure 9 shows the calculated time paths for capital for three values of and of K=Y1980 . The top-most graph uses a depreciation rate of 6.17% per year and a capital stock for 1980 of 2:85 times output. This series for capital is consistent with a literature that gives estimates of the capital stock for Argentina that are high by international standards and where this capital stock is attributed to a mix of protectionism and government regulations that make labor expensive and di¢cult to change. In a regulatory environment in which labor dismissal costs go up with length of service and can become quite large, optimizing …rms tend to keep their labor force small and use relatively high quantities of capital. The second and third graphs use depreciation rates and output capital ratios more in keeping with world averages. The second graph uses the values from Kydland and Zarazaga [10]. What is very di¤erent in this graph is the …rst decade of the period: there is much less growth in the capital stock during the 1970s. The third shows how a slightly lower depreciation rate and initial capital stock result in a time path not much di¤erent from the …rst, although at a lower output-capital ratio. The set of wedges that come from using the …rst capital series are shown in Figure 10, those from the second capital series in Figure 11, and those from the third capital series in Figure 12. Notice that the capital stock has no e¤ect on the wedges related to net exports and the labor tax. The only e¤ects are on the total factor productivity (A), and the tax on capital. For the …rst and third capital series, the changes are very minor. The relative fall in total factor productivity during the period from 1970 through 1990 is somewhat large with the third capital stock and the rise in TFP in the 1990’s is somewhat smaller. The capital tax series is very similar for these two as well. For the second capital series (the one that is closest to that of Kydland and Zarazaga [10]), the fall in TFP during the …rst 20 years is smaller and the 15 2.4 x 10 δ = .0617 K/Y 4 1980 = 2.85 K 2.2 2 1.8 1.6 1970 1.6 x 10 1975 1980 1985 1990 δ = .113 K/Y 4 1995 1980 2000 2005 2010 2000 2005 2010 2000 2005 2010 =2 K 1.4 1.2 1 1970 x 10 1975 1980 1985 1990 δ = .1 K/Y 4 1980 1995 = 1.8 K 1.4 1.2 1 0.8 1970 1975 1980 1985 1990 date 1995 Figure 9: Three di¤erent calculations of the capital stock 16 XM A 150 14 100 13 50 12 0 11 -50 -100 1970 10 1980 1990 τ 2000 9 1970 2010 1980 1990 τ k 0.7 2000 2010 2000 2010 h 0.4 0.6 0.3 0.5 0.2 0.4 1970 1980 1990 2000 0.1 1970 2010 Figure 10: Wedges when 1980 1990 = :0617 and K=Y1980 = 2:85 growth during the 1990’s is substantially larger than those that result from the other two capital stock series. The capital tax associated with this capital stock series implies a small drop in the tax in the 1980’s and a more secular upward trend thereafter (with similar variations, however). All three of the capital stock series shown here produce what Kydland and Zarazaga [11] refer to as Argentina’s capital shallowing. They note that there was relatively little recovery in the capital stock in the boom of the 1990s after the long period of capital stock rundown that was the 1980’s. Their model predicts a capital stock at least 20% higher than the observed one. The output boom of the 1990s seems to have been more motivated by increases in total factor productivity and by early declines in the capital tax than by capital accumulation. There was substantial new startup activity in the …rst half of the 1990s, concentrated mostly in service and commercial sectors, and many of the older import substituting manufacturing plants were closed (see Escudé, et al [7]). Since the service and commercial sectors tend to be less capital intensive than manufacturing, this shift in production away from manufacturing may explain at least part of the capital shallowing observed. 17 XM A 150 18 100 16 50 0 14 -50 -100 1970 1980 1990 τ 2000 12 1970 2010 1980 1990 τ k 2000 2010 2000 2010 h 0.4 0.6 0.3 0.4 0.2 1970 0.2 1980 1990 2000 0.1 1970 2010 Figure 11: Wedges when 1980 1990 = :113 and K=Y1980 = 2 XM A 150 20 100 18 50 16 0 14 -50 -100 1970 1980 1990 τ 2000 12 1970 2010 1980 1990 τ k 0.7 2000 2010 2000 2010 h 0.4 0.6 0.3 0.5 0.2 0.4 1970 1980 1990 2000 0.1 1970 2010 Figure 12: Wedges when 18 1980 1990 = :1 and K=Y1980 = 1:8 6 Conclusions The growth accounting technology provides another window through which we can decompose the economic history of a country. The narrative of economic history frequently points out that particular policies were favorable to one or another factor or that much of the evolution of the period was based on Solow residuals or total factor productivity10 . The growth accounting technique allows us to decompose the business cycle and growth of Argentina into a net export component, a total factor productivity component, and components that functions as taxes on labor and on capital. This paper makes two contributions to the literature. First, we provide a method for extracting the wedge that functions as a capital tax by adding to the model capital utilization as a household decision variable and then applying the data on capital utilization to the wedge extraction process. With this method, the wedge for the tax on capital makes a substantial contribution in the explaining the business cycle of Argentina (while the earlier method resulted in very little explanatory power for the capital wedge for the United States). This result may come from the method or may come from the greater importance that the capital wedge has in Argentina. Notable is the large negative correlation between total factor productivity wedge and the capital tax wedge. Further, cross-country studies on the importance of taxes on capital over the business cycle and its relationship to growth is suggested. The second contribution is applying this method to Argentina and comparing the results of the growth accounting technique to the narrative history. References [1] Baer, W., P. Elosegui and A. Gallo, (2002): "The Achievements and Failures of Argentina´s Neo-liberal Economic Policies". Oxford Development Studies. Volume 30, Number 1, February. [2] Cavalcanti, T., (2007): "Business Cycle and Level Accounting: The Case of Portugal," Portuguese Economic Journal, 6(1), 6:4764. [3] Chari, V. V., Patrick J. Kehoe and E. McGrattan (2007): "Business Cycle Accounting," Econometrica, 75(3), 781-836. [4] Cole, H., and L. Ohanain, (2002): "The U.S. and U.K. Great Depressions through the Lens of Neoclassical Growth Theory," American Economic Review, 92(2), 28-32. [5] de Pablo, Juan Carlos (2005): La Economía Argentina en la Segunda Mitad del Siglo XX, Volumes I and II, La Ley, Buenos Aires. [6] Elias, V., (1992): Sources of Growth: A Study of Seven Latin American Economies, ICS Press, San Francisco. 10 A particularly …ne work on total factor productivity in Latin America is by Elias [6]. 19 [7] Escudé, G., T. Burdisso, M. Catena, L. D’Amato, G. McCandless, and T. Murphy (2001): "Las MIPyMES y el mercado de crédito en Argentina," Working paper, Banco Central de la República Argentina. [8] Hayashi, F., and E. Prescott (2002): "The 1990s in Japan: A Lost Decade," Review of Economic Dynamics, 5, 206-235. [9] Hayashi, F., and E. Prescott (2006): "The Depressing E¤ect of Agricultural Institutions on the Pre-war Japanese Economy," Working Paper, University of Tokyo. [10] Kydland, F., and C. E. J. M. Zarazaga (2002): "Argentina’s Lost Decade," Review of Economic Dynamics, 5(1), 152-165. [11] Kydland, F., and C. E. J. M. Zarazaga (2002): "Argentinas’s Recovery and Excess Capital Shallowing of the 1990’2," Estudios de Economia, 29(1), June, pp.35-45. [12] Neumeyer, P., and F. Perri (2005): "Business Cycles in Emerging Economies: The Role of Interest Rates," Journal of Monetary Economics, 52(2), 345-380. [13] Prescott, E., (2002): "Prosperity and Depression," Amercan Economic Review, 92(2), 1-15. [14] Zablotsky, E., (1992): "A Public Choice Approach to Military Coups d’Etat," CEMA Working Papers: Serie Documentos de Trabajo. 85, Universidad del CEMA 20