An Empirical Test of the Post-Kaleckian Model applied to functional income distribution and long-run growth regime in Brazil

Appendix

Table A1. KPSS test

Series	Band	Test	Specification	Result
π	5	0.1273	Constant	I(0)
g	4	0.4218	Constant	I(0)
u	4	0.2420	Constant	I(0)

Notes: In this test is used Barlett kernel and Newey-West bandwidth.

The critical value (5% level) is 0.463.

Table A2. VAR lag order selection criteria

Lags	AIC	SC	HQ
0	-12.70755	-12.57424	-12.66153
1	-14.96855*	-14.43528*	-14.78446*
2	-14.63514	-13.70193	-14.31299
3	-14.67012	-13.33697	-14.20992
4	-14.76003	-13.02693	-14.16177

* Indicates lag order selected by the criterion.

Fig A1. Inverse root of AR characteristic polynomial

Table A3. VAR residual serial correlation LM test

Lags	LM-Stat	Prob
1	7.855255	0.5488
2	12.99147	0.1630
3	8.155983	0.5185
4	11.81306	0.2241

Fig. A2: Response of g and g to π shocks.

Fig. A3: Response of u to π shocks.

Fig. A4: Response of I/K to π shocks.

Fig. A5: Response of Y/Y^* to π shocks.

Fig. A6: Response of g to u shocks.

1^ Economic inequality for instance, has become a prominent topic after the publication of Piketty’s (2014) book Capital in the Twenty-First Century.

2^ In the light of what was reported by Blecker (2016) due to the relative closeness of the Brazilian economy we would expect a wage-led regime. However, the VAR approach is essentially a short-run methodology and, according to Blecker’s rationale it is more prone to yield profit-led results. So we cannot say beforehand the growth regime for the Brazilian economy.

3^ In fact these authors have adopted a structural VAR approach (SVAR) in which they impose contemporaneous restrictions on the interaction of the variables.

4^ We are using the nomenclature Post-Kaleckian growth models in reference to the third generation of KGM. In this generation, that follow the Bhaduri-Marglin’s approach, both profit-led and wage-led outcomes are possible. In the Neo-Kaleckian growth models, however, just the wage-led outcome is possible.

5^ From the structural approach, many different papers, such as Naastepad (2006), Naastepad and Storm (2007), Ederer and Stockhammer (2007), Hein and Vogel (2008) and Stockhammer, Onaran and Ederer (2009), employed this empirical strategy. Hein and Vogel (2008) and Stockhammer (2016) provide comparative boxes of results.

6^ Baer reports that “there is also considerable evidence that the real wages declined at first in the second half of the 1960’s and then rose at a rate substantially smaller than the rate of productivity increases.”

7^ The estimates are performed using data up to 2008 because, to the best of our knowledge, there is no series of functional income distribution that start at least in 1970 and cover a more recent period.

8^ According to Enders (2014, p. 290) the OLS estimates are consistent and asymptotically efficient.

9^ The results remained virtually unchanged when we used the Normal-Wishart prior.

10^ Capital accumulation estimates can be found in Marquetti and Porsse (2014).

11^ Estimates to potential output were based on Shaikh-Moudud methodology, available in Bruno (2015).

12^ This series was obtained from Institute of Applied Economic Research (IPEA) database.

13^ The main impulse-response functions of these alternative models are available in the Appendix.