Does Gold Price Matter for Divorce Rate in Iran?

Abstract

The increasing divorce rate has become a major social concern for policy makers in the Islamic government of Iran. The price of gold coin is an important factor in cost-benefit analysis for individuals in their marriage and divorce decisions in Iran. Dowries (Mehrieh) are usually in the form of gold coin and a wife has a legal right to request them from her husband upon both parties signing the marriage contract. Increasing the price of gold coin may intensify the internal stress and struggles within families, leading to a higher probability of divorce. We investigated the long-run relationship between real price of gold coin and divorce rate for the case of Iran over the period 1980–2014. Controlling for other factors such as women’s education, social globalization, economic growth rate, and the war period with Iraq, our regression results showed that there is a positive and significant long-run relationship between real price of gold coin (as well as unanticipated changes in real price of gold coin) and marital instability.

Appendix

The VAR Model: Dynamic Response of Divorce to Gold Coin Price Shocks

The VAR provides a multivariate framework relating changes in a particular variable (e.g., divorce rates) to changes in its own lags and to changes in (the lags of) other variables (e.g., gold coin prices). The reduced form VAR model is presented as follows:

$${y_t}={A_1}{y_{t - 1}}+ \ldots +{A_p}{y_{t - p}}+B{x_t}+{\varepsilon _t}$$

where ${y_t}~$ is a vector of k endogenous variables; ${x_t}$ is a vector of d exogenous variables; ${A_1}$,…, Ap and B are matrices of coefficients to be estimated; and ${\varepsilon _t}$ is a vector of innovations. The variables are affecting each other with some optimum lags and therefore they are all treated as endogenous variables. While changes in macroeconomic conditions, such as the gold market, can influence family stability, the changes in divorce rate can also influence the former. Our endogenous variables in the VAR were logarithm of divorce rate, logarithm of gold coin prices, and logarithm of women’s education, social globalization index and GDP per capita growth. Descriptions of used variables are mentioned in the main body of the analysis. We also controlled for the war period with Iraq with a dummy variable.

Our aim was to apply impulse response function in order to simulate the response of divorce rate to a positive shock in gold coin prices in Iran. Farzanegan and Gholipour (2016) also used this method to study the response of divorce rate to positive shocks in housing prices of Iran. To estimate response of divorce rate in Iran to positive shocks to gold coin prices, we employed the generalized impulse-response function (IRF) on the basis of our estimated unrestricted VAR model. The IRF shows the direction, size, and statistical significance of responses following an initial shock to gold coin prices. It is important to check for stability and residual autocorrelation of the estimated VAR model before using IRF tool. We used 2 years as the optimum lag in estimation of the VAR model on the basis of different lag order selection indicators as are shown in Table 5.

Table 5 Optimum lag length

The next step was to check the stability of the estimated VAR model. In the case of instability of the estimated VAR, we could not rely on the confidence intervals of simulated responses in IRF analysis (Farzanegan and Krieger 2017; Lütkepohl 1991). Table 6 shows that our estimated VAR was stable. In other words, the influence of a shock on all variables was decreasing over time.

Table 6 Stability of VAR model

In addition, we needed to check the residual autocorrelation of the estimated VAR model. The null hypothesis implied “no serial correlation at a specific order of lag.” In case of rejection of null hypothesis, we had to revise our VAR specification or had to use a higher lag lengths. Table 7 shows the residual serial correlation test results which does not show a concern in our case.

Table 7 Residual serial correlation

The ordering of variables in the estimated VAR model can be also important for subsequent IRF analysis. We followed Pesaran and Shin (1998) and used the Generalized IRF, which was not sensible to specific ordering of variables in the VAR. Figure 3 shows the IRF results. We can see that an unexpected positive change in gold coin prices led to an increasing response of divorce rate following initial shocks. The positive response of divorce reached its peak point within 3 years after shock. The response was statistically significant for the first 3 years following shock. A one standard deviation increase in log (gold coin prices) led to 0.06 unit increase in log (divorce rate) within the 3 years. This was equivalent to 6% increase in divorce rate as a response to positive shock in gold coin prices, controlling for other channels such as women’s education, social globalization and economic growth.

Fig. 3

Response of logarithm of (divorce rate) to logarithm of (gold coin price) positive shock. The dotted lines represent ± 2 standard deviation (at 95% confidence intervals). The deviation from the baseline scenario of no shocks is on the vertical axis; the periods (years) after the shock are on the horizontal axis. The vertical axis shows the magnitude of the responses