The purpose of this section is to empirically examine inflationary response to budget deficits, if any, when related variables, such as base money growth, exchange rate, domestic budget financing by NBU are taken into account. We test whether the price level is fiscal-dominantly determined[10]. We are also going to explore which part of the inflation variability is explained mostly by budget financing needs and which is influenced by other factors.
Rejection of the fiscal-dominancy hypothesis may be only implicit, because of possible cross-relationships that neutralize the direct effect of budget financing through monetary measures. Similarly, finding that the government balance affects prices itself does not directly imply that the impact of the monetary factors is lower or absent.
We chose the class of non-structural Vector Auto Regression (VAR) models for this analysis. Attempts to identify the components separately either in AR or structural specifications have failed in the sense that we obtain both poor econometric properties and weak economic inference. So we have tried to capture the mutual effect of variables associated with inflation to determine their net effect.
It may be interesting to impose additional restrictions and consider structural VAR, but it is hard to identify whether fiscal or monetary disturbances in the Blanchard-Quah (1989) decomposition should be considered permanent[11]. Another candidate for the model specification, Vector Error Correction (VEC) model, was not supported by the cointegration tests[12].
Table 2. Data Used for the Estimation
Mnemonics[13]
Description
Unit
Source
CPI
Consumer Price Index
State Statistics Committee
MBASE
Monetary Base
Mln UAH
National Bank of Ukraine
EXRATE
Official Exchange Rate, period average
UAH/ USD
National Bank of Ukraine
MA_CLG
Net claims on Government
Mln. UAH
International Financial Statistics
GBAL
General Government Balance
Mln UAH
State Treasury
We have examined specifications with the following features:
· The price level tends to be the most endogenous variable while the others are considered more exogenous. The last exogenous, and hopefully the most powerful, is the fiscal deficit series or its transformations;
· The specification includes 5 variables: consumer prices (CPI), exchange rate (EXRATE), monetary base (MBASE), NBU net credit to government (MA_CLG), and budget balance itself (GBAL). The ordering of the variables is important in VAR analysis, so we sort out variables as follows: CPI, MBASE, EXRATE, MA_CLG, GBAL;
· The general VAR specification is:
where pt is the transformed variables vector , pt-i , are the lagged variables vectors, Xt are the explanatory variables, matrices Ai, B to be estimated.
We have examined the sample of monthly data for 1995:1- 2000:6. The results of quarterly regressions have shown significant loss in explanatory power, so we remained with the volatile but more responsive monthly frequency. Table 2 reports the sources of the data, whereas summary statistics of the original and transformed VAR series are presented in Appendix 1. Government balance is the general government balance, using national definition, and is negative for a deficit.