Post n°13. In the household portfolio in France, the weight of life insurance stabilised at a high level in late 2015. Its previous growth came at the expense of regulated savings and money market mutual funds. This distribution appears to be mainly linked to supply effects, such as financial innovation and tax incentives, beyond the crises, even though there are some breaks in trends around 2008.
Six categories of assets with contrasted dynamics
Households’ financial assets are broken down into six categories (see chart): means of payment (sight deposits, for instance), passbooks with regulated rates (A passbooks, etc.), money market mutual funds (MMMFs) and term deposits, Housing Savings Plans (HSPs), “securities” (shares, bonds and non-money market mutual funds), life insurance and Popular Savings Plans (PSPs). These categories are defined in relation to the similarity of their characteristics (i.e. their respective degrees of liquidity, capital risk, etc.). In this post, we examine the shares of these different categories (i.e. the ratio of each category to the total portfolio).
The share of life insurance quadrupled from 11% of the total in 1980 to 45% in 2015. Conversely, passbooks with regulated rates declined by almost half (32% to 17%). The share of HSPs almost doubled to 10% in 2000, before converging to their level observed in the 1980s. The share of securities, and especially of MMMFs, has gradually fallen over time. Finally, the share of the means of payment has fluctuated more or less significantly around their 1990 levels (16% in 1990 compared with 15% in 2015). But most of these developments took place up to the end of the 1990s, and the global financial crisis and European sovereign debt crisis do not appear to have resulted in major upheavals. A closer analysis of trend breaks nevertheless reveals disruptions around 2008 for the least liquid assets.
Developments related to innovation, regulation, tax incentives
Four factors are at the root of the fall in the share of means of payment: the increase in wealth, the sharp fall in inflation since the late 1970s (formerly, households had to permanently rebuild their cash balances so as to maintain their purchasing power, the “Pigou effect”), the development of modern means of payment, and the spreading of financial innovations which have met the need for diversification, including liquid assets.
The changes in the shares of passbooks with regulated rates are attributable to the setting of rates by the authorities, and to the authorisation granted to all banks to distribute certain regulated products such as the A passbook. The HSP is an asset whose developments are dictated by regulation and reforms enacted by the government: deposit and loan ceilings, subsidies and taxes, interest rates, etc. Developments in MMMFs and term deposits closely mirror those of the money market.
The fluctuations of “securities” can mainly be ascribed to the valuation effects of the portfolio components, especially for shares. Finally, tax incentives, demography and the fall in long-term interest rates are three factors which account for the exceptional performance of life insurance over the whole period (see Monfront et al., 2005).
Least liquid assets: most recent break in 2008
The statistical tests (“Bai-Perron test”) carried out over the period 1978-2015 make it possible to isolate several breaks in the dynamics of the shares of the different assets (see Avouyi-Dovi et al., 2016, “A Revisited Almost Ideal Demand System: An Application to the French Households’ Financial Portfolio”, mimeo).
In fact, three breaks appear in the developments of aggregate assets (see table). They occured mostly in the 1980s and 1990s. They can be attributed to discontinuities in savings policies. The 2000s also feature a break for each asset category. These breaks occured either at the beginning of the period (means of payment, MMMFs and term deposits), or at the beginning of the financial crisis (passbooks with regulated rates, life insurance and securities of which the cumulative shares exceeded 75% of the total of the portfolio in 2015 Q4).
HSPs are apart with a break at the end of 2006, when all interests accrued on HSPs opened more than 12 years before became taxable, leading to a fall in their share.
| Financial Assets | Break 1 | Break 2 | Break 3 | Break 4 |
|---|---|---|---|---|
| Means of payment | 1984 T1 | 1989 T3 | 1996 T4 | 2002 T3 |
| Passbooks with regulated rates | 1986 T1 | 1991 T3 | 2008 T3 | ND |
| Money market mutual funds (MMMFs) and term | 1984 T4 | 1989 T2 | 1996 T4 | 2003 T4 |
| Housing Savings Plans | 1986 T3 | 1995 T3 | 2006 T3 | ND |
| Securities | 1985 T3 | 1992 T1 | 2002 T1 | 2008 T4 |
| Life insurance, and Popular Savings Plans | 1987 T3 | 1992 T1 | 1998 T3 | 2008 T1 |
Instability of the relationships between the portfolio structure and its explanatory factors
The model used here show the relationship of the shares of financial assets to their yields and to households’ wealth (see Deaton and Muellbauer, 1980 et Avouyi-Dovi et al., 2016. It also allows for substitution or complementarity effects between assets. We added indicators of uncertainty, real activity and life cycles. Finally, we used constraints to ensure the consistency of the system.
The breaks are captured through the instability of the explanatory relationships between the assets and their determinants (yields, wealth, etc.). The equations are estimated continuously on sub-samples in which the oldest observations are gradually replaced by the most recent observations. These regressions make it possible to check whether the variations in the information set change the coefficients significantly.
The results show instability in the relationships to the extent that the elasticities between the asset shares and their determinants change over time irrespective of the asset considered. This observation is valid for the fundamental explanatory factors (yields, wealth, etc.) as well as for the additional factors (indicators of uncertainty). In particular, the indicators estimated over the first third of the period under review, which is marked by financial deregulation and the crises of the European Monetary System, are very volatile. Finally, depending on the assets, some breaks in the trends of the coefficients occurred in the late of the 1990s and early 2000s.
To summarise, the model is not called into question but its coefficients are not stable over time. Indeed, the explanatory power of the model estimated over the period 1999-2015 remains high (over 90%) despite the breaks in 2008.
The occurrence of crises and the persistence of low interest rates have not impacted the structure of the French households’ portfolio so far. Indeed, changes in this structure can still be attributed to supply effects, relating to financial innovation and tax incentives. However, in the wake of the financial crisis, some breaks have occurred in the trends of the least liquid assets (life insurance, securities, etc.). As data are not available over a sufficiently long period, we cannot deduce that this predominance of life insurance will not continue in the near future.