Regression Results

According to the Augmented Dickey-Fuller test results (Appendix 2), prices, monetary aggregates, credit to government and exchange rate appear to be I(1) processes with trend and drift, though the government balance seems to be I(0) i.e. stationary in levels. Thus we differentiated once all series except GBAL, which is left in levels.

Pairwise Granger causality tests (Granger, 1969) are shown in Appendix 3. It becomes clear NBU financing may directly affect the prices. However, we have not enough evidence to reject the null hypothesis of no causality from budget balance to prices.

The regression results are shown in Appendix 4, whereas Appendix 7 covers the impulse response functions (IRF), a short-run dynamic one and a cumulative one for the steady state[14], and the variance decomposition. Appendix 5 assures that stability of the underlying AR scheme in VAR holds. All roots lie inside the unit circle that guarantees the lag polynomial stability. We chose the lag length of 2 after the concurrent testing and examination of various cross-correlation relationships (see Appendix 6).

The discovered VAR scheme is dynamically stable and rapidly converge to steady state. The GBAL equation is the weakest part of the model, however not weak enough to ignore GBAL effect. Some of the regressors are correlated, so it’s hard to analyze the corresponding t-statistics. Nevertheless, the estimation method generates consistent unbiased estimates for the model coefficients. Chow forecast test couldn’t reject the hypothesis of no structural break for financial crisis 1998.

The largest cumulative effect on inflation rate has the inflation itself. A 1% increase in inflation generates 2.5% more inflation over the next year (2% over the 6 months).

Inflationary impacts of the monetization of the budget deficit and the budget deficit itself are of same magnitude. Dynamic cumulative elasticity[15] of the price level to the NBU claims to government equals 0.4. A 1% increase in the stock of the credit to government leads to 0.4% of inflation over the year. The biggest impact is on 3-5 month after the shock.

On average, an increase in the deficit of 1% GDP[16] leads to 0.38% inflation over the next year. As the average monthly GDP and the stock of NBU credit to government are very close amounts, same percentage change in both leads to roughly equal inflationary response.

Somewhat surprisingly, regression results show exchange rate and monetary base dynamics have very low, if any, influence on prices[17]. We tend to explain this finding by supposedly the same monetary nature of the monetary base, the claims to government, and the exchange rate. Thus, the model finds the most linked variable, diminishing the marginal contribution of the others.

Forecast-error decomposition illustrates that steady state configuration of CPI inflation variance is distributed 10% due to growth of the NBU credit. Exchange rate and budget balance variations yield approximately 3% explanation to inflation each. The monetary base growth explains less than 1% of the inflation variability.

Date: 2015-02-28; view: 746