Two closely watched indicators of economic performance are inflation and unemployment. This study empirically analyzes the causality between inflation and unemployment in Indonesia during 1984 to 2017. The data were gathered from the Indonesian Central Bureau of Statistics. Methodologically, this study employed the Granger Causality test and Vector Autoregression to determine the causality between inflation and unemployment. The results show that there is a one-way causality between inflation and unemployment. The findings imply that unemployment causes inflation, but not vice versa. Next inflation and unemployment are also closely related to other determining factors, such as season, household income, and the decisions to attend school or to perform the housekeeping.

Post global financial crises (2008) have forced countries to adopt expansionary and stimulating macroeconomic policies aiming to reduce unemployment. Some countries, such as United Kingdom, Germany, and the United States of America have become successful in lowering the unemployment in their labor markets. However, Spain and Italy are stuck at high rates of unemployment with rigid labor markets (

Short-term economic problems, such as inflation and unemployment are among the most notable macroeconomic problems all the time (

The high inflation rate in 1965 also caused a high unemployment rate (read: stagflation). Since 1965, the unemployment rate has increased by 5–6% per year. However, similar to the inflation rate, the Indonesian government managed to reduce the unemployment rate to less than 10% (Bank Indonesia 2004). Every government closely monitor inflation and unemployment as the two main economic performance indicators. Statisticians combine inflation and unemployment data to develop the misery index that aims to measure the health of an economy. One of the economic principles is the short-term trade-off between inflation and unemployment. If fiscal and monetary policymakers increase aggregate demands and economy along the short-run aggregate demand curve, they can reduce unemployment temporarily, albeit with an increase in inflation rate. On the other hand, if monetary and fiscal policymakers reduce aggregate demands and economy along the short-run aggregate demand curve, they can curb inflation but also increase unemployment temporarily (

This study aims to investigate the trade-off between inflation and unemployment as found by

Several studies have investigated the relationship between inflation and unemployment.

Besides studies that show no causal relationship between inflation and unemployment,

Further,

Besides the negative results, other studies find the positive relationship. For example, using the Egyptian data,

Other studies demonstrate the two-way causality between inflation and unemployment. For example,

Overall, these studies show varying results such as one-way causality, two-way causality, and no causal relationship between inflation and unemployment. Further, these studies also use different analytical models, such as Granger Causality, Johansen Cointegration, Autoregressive Distributive Lag, Error Correction Model, Vector Error Correction Model, Panel Data, Vector Autoregression, and etc. It can be concluded from the previous discussion that there is an uncertain relationship between inflation and unemployment of different economies in the certain period.

This study uses the secondary data from the central bureau of statistics and the world bank publication. More specifically, the study relied on the time-series data from 1984 to 2017. Further Granger Causality and Vector Autoregression used to analyze the data. Before running the Granger Causality and Vector Autoregression model, this study initially ran the stationary and the lag length test. The following are the models for the stationary test and the test statistic (

Δ

After running the stationary test, this study ran the lag length test. There are various approaches to select the optimal lag length, such as Likelihood Ratio, Final Prediction Error, Akaike Information Criterion and Schwarz Information Criterion (

The above equation indicates that _{t}_{t}

1. If Σ

2. If Σ

3. If Σ

4. If Σ

Further, this study ran the Vector Autoregression after completing the Granger Causality test. The Vector Autoregression (VAR) is commonly used for forecasting systems of interrelated time series and for analyzing the dynamic impact of random disturbances on the system of variables. The reduced form VAR approach sidesteps the need for structural modeling by treating every endogenous variable in the system as a function of _{t}_{1}_{t} , y_{2}_{t} , …,y_{kt}

_{t}_{1}_{t}_{–1} + … + _{p}y_{t–p}_{t}_{t}

where:

_{t}_{1}_{t} , y_{2}_{t} ,…,y_{kt}

_{t}_{1}_{t} , x_{2}_{t} ,…,x_{dt}

_{1}, …, _{p}

∈_{t}_{1}_{t}_{2}_{t}_{kt}_{t}_{t}_{t}_{t}_{s}

Table

Table

Stationarity test

Variable | Conclusion | |

Inflation | 0.0000* | I(0) |

Unemployment | 0.3012 | the series is not stationary |

DUnemployment** | 0.0000* | I(1) |

*indicates the rejection of the null hypothesis at 5% of significance level. ** DUnemployment implies that Unemployment at the first difference [I(1)].

Lag length test

Lag | LogL | LR | FPE | AIC | SC | HQ |

0 | –151.4651 | NA | 197.5714 | 10.96179 | 11.05695 | 10.99088 |

1 | –150.5083 | 1.708562 | 245.9338 | 11.17917 | 11.46464 | 11.26644 |

2 | –145.1985 | 8.723246 | 225.3365 | 11.08561 | 11.56140 | 11.23106 |

3 | –125.0476 | 30.22635* | 72.08667* | 9.931972* | 10.59807* | 10.13561* |

4 | –121.2335 | 5.176270 | 74.94396 | 9.945251 | 10.80167 | 10.20707 |

5 | –120.3520 | 1.070367 | 97.66239 | 10.16800 | 11.21473 | 10.48800 |

*indicates the optimal lag.

Table

Granger causality test

Pairwise Granger Causality Tests | |||

Sample: 1984 2017 | |||

Lags: 3 | |||

Null Hypothesis: | Obs | F-Statistic | Prob. |

Inflation does not Granger Cause DUnemployment | 30 | 0.72869 | 0.5454 |

Dunemployment does not Granger Cause Inflation | 33.0657 | 2.E-08 |

Table

VAR Model of inflation and DUnemployment for Indonesia (1984–2017)

DUnemployment | Inflation | |||

Coefficients | t-prob | Coefficients | t-prob | |

DUnemployment(–1) | –0.135523 | 0.20747 | 0.709939 | 1.30180 |

DUnemployment(–2) | –0.181229 | 0.20753 | –6.937364 | 1.30223 |

DUnemployment(–3) | 0.217161 | 0.24175 | 11.60203 | 1.51692 |

Inflation(–1) | 0.024945 | 0.01742 | 0.421936 | 0.10934 |

Inflation(–2) | 0.003714 | 0.01524 | 0.042180 | 0.09564 |

Inflation(–3) | –0.006079 | 0.01513 | 0.021545 | 0.09495 |

Constant | –0.186910 | 0.34132 | 4.400302 | 2.14168 |

R^{2} |
0.124570 | 0.813392 | ||

F-statistic | 0.545466 | 16.70883 | ||

Log-likehood | –39.33832 | –94.43436 | ||

AIC | 3.089221 | 6.762290 | ||

Swarz SC | 3.416167 | 7.089236 |

Table

Impulse Response Functions (IRFs) was calculated for DUnemployment and inflation to address the reaction of the economy to external changes (shocks). The results of the IRFs analysis show that there is a trade-off between inflation and DUnemployment as shown by the IRFs of DUnemployment to inflation. Overall, estimate results of Granger Causality, Vector Autoregression, and Impulse Response Functions (IRFs) prove that the DUnemployment is more instrumental to explain inflation in Indonesia.

Impulse responses to DUnemployment and inflation shocks

Further, this result is supported by

Thus, when inflation does not support DUnemployment, it is necessary to analyze factors that affect DUnemployment. Table

Table

Further, another factor that affects the number of unemployment is the industry in which the population work. Table

The Population 15 Years of age or over by the main employment status (2001–2017) (source: Badan Pusat Statistik (2017a), processed))

Number | Main Employment Status | 2001 | 2017 | ||

Amount | % | Amount | % | ||

1 | Self-employed | 17,451,704 | 19.22 | 21,849,573 | 17.54 |

2 | Employer Assisted by Temporary/ Unpaid Worker | 20,329,073 | 22.39 | 21,275,899 | 17.08 |

3 | Employer Assisted by Permanent/ Paid Worker | 2,788,878 | 3.07 | 4,446,024 | 3.57 |

4 | Employee | 26,579,000 | 29.27 | 47,420,633 | 38.08 |

5 | Casual Agricultural Worker | 3,633,126 | 4.00 | 5,360,306 | 4.30 |

6 | Casual Non-Agricultural Worker | 2,439,035 | 2.69 | 6,021,760 | 4.84 |

7 | Family/ Unpaid Worker | 17,586,601 | 19.37 | 18,164,654 | 14.59 |

8 | No Answer | – | – | – | – |

Total | 90,807,417 | 100.00 | 124,538,849 | 100.00 |

Population 15 years of age or over who worked by main industry (1991–2016) (source: Badan Pusat Statistik (2017b), processed))

Number | Main Industry | 1991 | 2016 | ||

Amount | % | Amount | % | ||

1 | Agriculture, Plantation, Forestry, Hunting, and Fisheries | 39,385,946 | 53.29 | 38,291,111 | 31.74 |

2 | Mining and Quarrying | 551,581 | 0.75 | 1,311,834 | 1.09 |

3 | Manufacturing Industry | 7,712,468 | 1.43 | 15,975,086 | 1.24 |

4 | Electricity, Gas, and Water | 148,480 | 0.20 | 403,824 | 0.33 |

5 | Construction | 2,415,002 | 3.27 | 7,707,297 | 6.39 |

6 | Trade, Restaurants, and Accomodation Services | 11,190,391 | 1.14 | 28,495,436 | 23.62 |

7 | Transportation, Warehousing, and Communication | 2,475,803 | 3.35 | 5,192,491 | 4.30 |

8 | Financial, Real Estate, and Business Services | 515,401 | 0.70 | 3,481,598 | 2.89 |

9 | Community, Social, and Personal Services | 9,377,036 | 12.69 | 19,789,020 | 1.40 |

10 | Undefined | 139,516 | 0.19 | – | – |

Total | 73,911,624 | 100.00 | 120,647,697 | 100.00 |

The proportion of the population above 15 years who worked at the primary sectors (agriculture, plantation, forestry, hunting, and fishery) declined sharply. In 1991, 53.29% of the working population worked in the primary sectors, and proportion declined to 31.74% in 2016. The agricultural sector dominates the primary sectors because of most population work in this sector (

Independent Sample T Test for Unemployment on February and August (1986–2017)

Levene’s Test for Equality of Variances | t-test for Equality of Means | |||||||||

F | Sig. | t | Df | Sig. (2-tailed) | Mean Difference | Std. Error Difference | 95% Confidence Interval of the Difference | |||

Lower | Upper | |||||||||

Unmpl | Equal variances assumed | .002 | .967 | .000 | 24 | 1.000 | –268.00000 | 6.06551E5 | –1.25213E6 | 1.25159E6 |

Equal variances not assumed | .000 | 23.945 | 1.000 | –268.00000 | 6.06551E5 | –1.25228E6 | 1.25174E6 |

The significance value (2 tailed) of 1.000 is bigger than the tolerance value of 5% (0.05) implies that there is no difference between February and August unemployment rate. The more developed irrigation system reduce the farmers’ dependence on rainfall and eventually on the season. Next, the Indonesian working-age population who attended school or performed the housekeeping increased. The number of economically inactive women due to housekeeping increased both in absolute and relative terms (

Independent Sample T Test for Working Age Between Attend School and Housekeeping (1986–2017)

Levene’s Test for Equality of Variances | t-test for Equality of Means | |||||||||

F | Sig. | T | df | Sig. (2-tailed) | Mean Difference | Std. Error Difference | 95% Confidence Interval of the Difference | |||

Lower | Upper | |||||||||

WorkAge | Equal variances assumed | 4.160 | .046 | –48.096 | 60 | .000 | –9.88226 | .20547 | –10.29326 | –9.47126 |

Equal variances not assumed | –48.096 | 51.170 | .000 | –9.88226 | .20547 | –10.29472 | –9.46980 |

The significance value (2 tailed) of 0.000 is lower than the tolerance value of 5% (0.05) implies that there is a difference between the Indonesian working-age population who attended school and performed the housekeeping.

Further, the number of the population 15 years of age or over who attended school increased from 9,147,830 in 2005 to 15,244,852 in 2017. However, the proportion of the working-age population who attended school decreased from 9.04% in 2005 to 7.94% in 2017. Higher school attendance decreases the number of unemployment.

This study suggests the one-way relationship between inflation and DUnemployment. More specifically, the Granger Causality, Vector Autoregression, and Impulse Response Functions (IRFs) model show that from 1984 to 2017, DUnemployment causes inflation, but not vice versa. The results imply that the Phillips model (

Various factors affect the Indonesian unemployment rate, such as: (1) The season factor significantly affects unemployment, albeit with the declining magnitude, because the agricultural sector still absorbs a significant portion of the Indonesian labor force; (2) Increased income encourages young labor force (15–19 years) to delay entering the labor market but to continue their studies; and (3) Better economic condition also increases the number of non-labor force. More specifically, women prefer becoming housewives (caring for their households) in entering the labor market because caring for households is also a productive activity (

Inflation is a less effective policy instrument to overcome the unemployment problem in Indonesia. This argument implies that increasing the inflation rate is ineffective to reduce the unemployment rate. Numerous facts indicate that other variables affect the Indonesian unemployment rate. However, it is viable to increase the unemployment rate to control inflation, although this policy has to be implemented carefully. Further, the Indonesian geographical condition that consists of thousands of islands likely causes the implementation of macro policies to take a longer time because of the greater needs to adjust for the inter-region differences. Thus, the use of the panel data model likely accounts for the possible inter-region variances better.