The Determinants of Price Inflation in the United States: A Multivariate Dynamic Cointegration and Causal Analysis

Although Classical theory postulates that there is a short-run trade-off between inflation and unemployment rates, numerous studies have shown that the relation between inflation and unemployment rates vary greatly across different sample periods, countries, estimation techniques and data sources. Due to the contrasting results obtained in previous studies, in this paper we investigate the short-run and long-run effects of unemployment rate, long-term interest rate, trade openness, budget deficit, money supply, economic growth rate and exchange rate on U.S. inflation rate over the period 1978-2014. The Augmented Dickey Fuller unit root test is performed. The variables are found to be integrated of order one. We apply the bounds testing approach proposed by Pesaran et. al (2001). The Johansen and Juselius (1990) and the Engle-Granger (1987) tests show that the variables are cointegrated. The Granger F-test is applied to the vector error correction model to investigate causal relationships between different pairs of variables. Results indicate short-run unidirectional causalities from interest rate, trade openness, economic growth rate and exchange rate to inflation rate, from interest rate to unemployment rate, from economic growth rate to trade openness, and from unemployment rate, trade openness, budget deficit, economic growth rate and exchange rate to money supply. Long-run causality is observed when inflation rate is an endogenous variable. In cointegration analysis, the adjustment coefficient is found to be statistically insignificant; thus, the results are considered plausible only in the short-run. Long-term interest rate and trade openness are found to have significant positive short-run effects on inflation rate. No significant short-run trade-off is observed between inflation and unemployment rates. The results of the cumulative sum (CUSUM) and cumulative sum of squares (CUSUMSQ) tests proposed by Borensztein et. al (1998) are within the critical bounds; thus all the coefficients in the error correction model are found to be stable. The diagnostic tests indicate that there is no evidence of serial correlation, and no problem of heteroscedasticity. The autoregressive conditional heteroscedasticity is not present in the short-run model. The test results also indicate that there is no problem of normality in error distribution. Based on the cointegration and causal analysis, we find no evidence to support a short-run Phillips curve for the U.S. economy for the period 1978-2014. A rise in inflation rate, due to a monetary expansion, is not found to have any significant short-run impact on unemployment rate in the U.S. economy for the sample period under study.