Long run determinants and short run dynamics of inflation in Tunisia

Abstract

The essential idea of this study is to analyse the origins of inflation at short and long runs in Tunisia relying on annual data during the period 1962 to 2003. We also suggest a model that has a structure determined by monetary and structural factors, and estimated by Johansen's cointegration technique. The empirical results show that inflation is explained by mixed factors: monetary ones such as money supply, the interest rate and the real effective exchange rate; and structural ones like the nominal average annual wage rate, the import prices and the real output. The analysis aims at pointing out the long run determinants of inflation and studying its short run dynamics.

Notes

¹ Friedman (1976), Saini (1982), Darrat (1986, 1994), Dhakal and Kandil (1993), Beltas and Jones (1993), Deme and Fayissa (1995), Lui and Adedeji (2000), El-Sakka and Ghali (2005), Kandil (2005) and Pelipas (2006).

² Olivera (1964), Argy (1970), Bruno (1978) and Gordon (1988).

³ Mehra (1991), Darrat (1994), Kim (2001) and Boschi and Girardi (2005).

⁴ See Saini (1982), Darrat (1986, 1994, 1997), Siddig (1993), Kim (2001), Boschi and Girardi (2005), Kandil (2005), El-Sakka and Ghali (2005) and Pelipas (2006).

⁵ See Darrat (1986, 1997) and Deme and Fayissa (1995).

⁶ See Kandil (2005) and Boschi and Girardi (2005).

⁷ See Gordon (1988), Mehra (1991, 1994), Darrat (1994), Kim (2001) and Boschi and Girardi (2005).

⁸ See Deme and Fayissa (1995) and Darrat (1997).

⁹ See Deme and Fayissa (1995), Darrat (1997), Lui and Adedeji (2000), Boschi and Girardi (2005), El-Sakka and Ghali (2005) and Pelipas (2006).

¹⁰ The lag length on each variable is determined by Hendry method on the basis of statistical significance.

¹¹ All the variables are taken in natural logarithms.

¹² See Fig. A1 (Appendix).

¹³ We measure general level of prices by wholesale prices (PG_t ).

¹⁴ Following the adaptive-expectations theory, we approximate expected inflation for any future period by a weighted-average of the current and past inflation rates. The formula is presented by . The number of retards is determined by Akaike Information Criteria (AIC).

¹⁵ See Fig. A2 (Appendix).

¹⁶ The regression which used in testing for a unit root with a changing slope and intercept (Perron, 1989) is where ΔΔy_t  = Δy_t  − Δy_{t− 1}; Δy_t  = y_t  – y_{t− 1}; μ, γ₁, γ₂, α, ci are the parameters of regression; DU_t  = 1 (t > T _B) and DR_t  = t (t > T_B ) are the dummies; k is the number of lags in regression; T_B is the point of structural break; ϵ _t is an error term.

¹⁷ The expected inflation is excluded from the model (3) because it is not of the same integration order as the other variables.

¹⁸ The specificity of the model of Deme and Fayissa (1995) is that it is close to the monetarist model. The inflation rate is a function of the money supply, the real income, the domestic nominal interest rate, foreign interest rate, expected inflation rate and exchange rate relying on the use of annual data during the period 1964 to 1993 for Egypt, Morocco and Tunisia. They showed that the monetary variables have no impacts on the inflation process in these countries.

¹⁹ Darrat (1997) tried to analyse the causes of inflation for Morocco, Pakistan, Egypt and Turkey during the period 1960 to 1993, using the cointegration method. He showed that the monetary variables have no impact on the general levels of prices.

²⁰ Siddig (1993) studied the inflation determinants in petrol exporting developing countries during the period 1970 to 1990, using quarterly data. His model estimation was carried out by the least squares method. The results confirm that money supply is an important determinant of domestic inflation in these countries.

²¹ Abdel Mahmoud (1997) used the technique of error correction model and quarterly data to Sudan during the period 1970 to 1994. The results obtained showed that money supply affected the level of prices.