4. Structural Breaks

From the graph of inflation (figure 3) the level of inflation potentially changes around 1970, tending to a higher level. I will therefore test a null hypothesis of a structural break in 1970Q3 against the alternative of parameter consistency.

Clearly, deciding on a potential structural break is open to individual interpretation¹⁴. The two equations follow, for 1956Q3 to 1970Q3 and 1970Q4 to 1982Q2 respectively.

p_t^A = 1.0177 + 0.46012p_t* - 6.2752x_1t + 6.1785x_1t-1 - 0.00019x_2t + 0.0000417x_2t-1 + 0.000236x_2t-2 + 0.00009x_2t-3 + 0.000349x_2t-4 + 0.51361x_3t - 0.44079x_3t-1 - 0.47381x_3t-2 + 0.5345x_3t-3 + 0.95067x_3t-4 (5)

p_t^B = 2.2808 + 0.87387p_t* + 4.5815x_1t + 3.6053x_1t-1 - 0.00016x_2t + 0.000412x_2t-1 - 0.000186x_2t-2 + 0.000143x_2t-3 + 0.000126x2t-4 + 0.72987x3t - 0.67389x_3t-1 - 1.7056x_3t-2 - 0.04807x_3t-3 + 0.6699x_3t-4 (6)

Most noticeably, in equation (5), the coefficient on core inflation is 0.46, as opposed to being almost one in equation (3). This suggests that inflation was caused by factors other than its previous level and that consumers did not place as much weight on past events. Also, in a marked change from the whole sample, import prices have a massively significant effect, in both sub-models. The lower significance of imports in the second period is surprising, given the events of the 1970s¹⁵. Again, the demand shock is insignificant. The formal test for structural breaks is the Chow test, detailed previously, which offers no evidence to support the null of structural break and so we conclude that the causes of inflation do not change significantly over time¹⁶.

5. Subsets

As explained, I will use 10 year blocks, as opposed to actual decades, which are selected by the years and not by the events that occurred. The periods are 1956 Q3 to 1965 Q4, 1966 Q1 to 1975 Q4 and 1976 Q1 to 1983 Q4. The OLS of the first period produced the following result, using the previous notation.

p_t^A = 2.0902 + 0.020675p_t* - 13.590x_1t + 9.98766x_1t-1 - 0.00052x_2t + 0.0003x_2t-1 + 0.00043x_2t-2 + 0.00013x_2t-3 + 0.00072x_2t-4 + 1.6836x_3t + 0.61869x_3t-1 - 0.50539x_3t-2 - 0.2845x_3t-3 - 0.13436x_3t-4 (7)

As we can see, the core inflation term has virtually no effect on the current inflation term. This could be as a result of the historically low inflation rates throughout this period, which causes low expectations of inflation, which is one of the theories behind core inflation. The effect of the relative price of imports is highly significant, although with a different sign on the two lags, which is unexpected. Demand shocks again have no significance and the incomes policy dummy has only a small impact, although decreasing through lags, as expected.

The OLS of 1966 Q1 to 1975 Q4:

p_t^B = 3.6423 + 0.81005p_t* + 7.9024x_1t + 8.0463x_1t-1 - 0.00022x_2t + 0.00006x2t-1 + 0.00006x_2t-2 + 0.00017x_2t-3 + 0.00005x_2t-4 + 1.7678x_3t - 1.3972x3t-1 - 1.4483x_3t-2 - 0.2785x_3t-3 - 0.3526x_3t-4

(8)

Similarly to the first block, demand shocks are insignificant and relative import prices are important, owing to the start of the OPEC crisis but, in this case, both signs are positive, as is to be expected. The core inflation term has changed from having little importance in the first equation to having a significant effect. The absolute incomes policy effect weakens through all the lags. And the OLS of 1976 Q1 to 1983 Q4:

p_t^C = -0.42788 + 0.7844p_t* + 22.839x_1t - 33.238x_1t-1 - 0.00044x_2t + 0.00003x_2t-1 - 0.00039x_2t-2 + 0.0005x_2t-3 - 0.00005x_2t-4 - 3.8325x_3t + 1.4029x_3t-1 + 2.1544x_3t-2 - 1.5131x_3t-3 + 6.2695x_3t-4

(9)

Most noticeably, the relative import price has a massive effect, as is to be expected as the major OPEC oil crisis occurred within this period. Once more demand shocks are insignificant. During the first two periods incomes policy has a significant but stable effect, whereas in this period the effect is much larger, especially for the lagged terms. Core inflation remains at a significant level.

6. OPEC Period

Of the period covered by the data, one of the obvious causes of higher inflation is the rising oil prices experienced in the 1970s. Therefore, I have modelled the period of 1973 Q1 to 1979 Q4, and will compare the determinants to see if, as one might expect, the relative price of imports is more significant during this period than for the entire sample. The OLS regression produces

p_t = 10.264 + 0.66282p_t* + 8.9699x_1t + 6.1102x_1t-1 -0.00061x_2t + 0.00015x_2t-1 - 0.00014x_2t-2 + 0.00080x_2t-3 - 0.00027x_2t-4 - 1.7396x_3t - 1.4376x_3t-1 - 0.9719x_3t-2 - 1.2232x_3t-3 - 2.2935x_3t-4

(10)

As predicted, relative import prices have a strong effect, especially relative to the other variables. This is to be expected, as the OPEC crisis led to higher import prices and so depreciated sterling, therefore imported factors of production led to higher costs and consumable imports created inflation directly. Incomes policies of some description were in force for most of this period, undoubtedly as a response to the high inflation caused by oil prices. The incomes policy dummy is exclusively negative in this period, which suggests it was achieving its objective of limiting inflation.

Notes

14. See E. Zivot & D. W. K. Andrews, "Further evidence in the great crash, the oil-price shock, and the unit-root hypothesis", Journal of Business and Economic Statistics 10, 1992, for a discussion of endogenous versus exogenous choices of structural break points.
15. See subsequent section on OPEC Period.
16. See Appendix 4 for workings.

Go to I. Introduction

Go to Estimation and Results part 1

Go to III. Conclusions

Return to project index.

Titlepage