News media occasionally report anecdotal evidence that currently depressed levels of interest rates do encourage higher household saving. Usually, it is then claimed to reveal a dysfunctionality of current expansionary monetary policy including ultra-low key policy rates and large-scale asset purchases ("QE") by central banks. Low interest rates discourage consumption as households discount lower savings until retirement and, hence, make up by saving more in order to reach a certain retirement wealth. Less private consumption leads to less output growth and eventually lowers inflation. Hence, lower interest rates do not lead to higher inflation but to lower inflation and, in that sense, it is suggested that central bank’s policies are kind of fundamentally flawed.
In Austria, the savings rate averaged 7.6 % since 2010 and has been below its long-term average of 10 % since the rapid fall following the 2008/09 financial crisis. However, in recent years, the savings rate gradually increased from its trough at 5.5 % in Q2 2013 to 7.8 % in Q2 2016. During the three years period, the interest rate paid on household overnight deposits more than halved from 0.78 % to 0.31 % (Figure 1).
Still, Austrian private households prefer deposits to other means of saving and investment. The OeNB reported recently that households paid 48 bn EUR into overnight deposits since 2010 implying a strong liquidity preference.
Do lower interest rates lead to higher savings rates?
No – according to parts of the existing literature. The Austrian National Bank (OeNB) conducted a survey in spring 2015 to assess household interest rate perceptions and their impact on savings decisions. The authors found that the very low interest rates were only one of several determinants in savings decisions and have had only a small effect so far. Low real returns have also only marginally influenced the savings and investment behavior of German households. Factors relating to income and wealth, demographics and the institutional framework, are likely to have been far more important determinants. The persistent strong risk aversion among German households has been further diminishing the significance of returns as a determinant of investment behavior. Household savings behavior has various determinants. Traditional explanatory variables include income and wealth levels, demographics (for example, the old age dependency ratio), the real interest rate, uncertainty (inflation, unemployment rate), fiscal policy, the pension system, financial market performance (money stock, private sector credit) or international financial integration (current account balance). In the relevant literature, the impact of real interest rates seems ambiguous, as some authors do not find any effect and some authors find a positive effect or a negative effect.
In theory, there is an income effect and a substitution effect: A drop in interest rates will initially result in savers receiving less income from their savings than previously expected (income effect) forcing them to save more and consume less today if they wish to maintain future consumption at a previously targeted level. At the same time, an interest rate reduction can drive up current consumption at the expense of future consumption (substitution effect) as lower income from saving makes current consumption less "expensive". In addition, a wealth effect acts when a reduction in interest rates lifts the prices of a household’s securities holdings, theoretically lifting the household’s consumption options.
Empirical exercise
We explore the empirical relationship between the nominal overnight deposit rate and the savings rate. Hence, it is assumed that households predominately observe changes in interest rate paid on their check account or term deposits. In comparison, financial investors would typically look at longer maturities, for example, ten year government bond yields. Employing nominal interest rates instead of real (inflation adjusted) interest – as usually done in the literature, implies a sort of household money illusion. Households care about nominal returns, as is also often suggested in the public discussion and as real deposit rates have eventually been very low for long. The savings rate is the percentage rate of household disposable income that is not spend on consumption during the period. We use annual data from Eurostat and the ECB for the period between 2000 and 2015. The sample covers 13 Euro Area countries including some core countries (Austria, Belgium, Finland, France, Germany and Netherlands) and some peripheral countries (Estonia, Greece, Italy, Ireland, Latvia, Portugal, Spain). This allows us to exploit cross-country and time-variation in the variables (panel data) and use a statistically viable sample size. The regression model takes account of country-specific effects, which could include unobserved factors like, for example, culturally induced differences in risk attitudes. Real GDP growth is included as a control variable. The small model falls short of including all determinants described above, hence, we hope that GDP growth controls for a lot of other "stuff" (income, uncertainty, etc.).
We would not have expected that.
Some interesting results arise (Table 1). As suggested in parts of the literature (see above), over the total period between 2000 and 2015, there existed a statistically significant and positive link between bank deposit interest rates and the savings rate. An increase in the interest rate by 1 percentage points is associated with an increase in the savings rate by 1.4 percentage points. The result also holds basically true when only the peripheral countries are in the regression. It appears as a bit of a surprise, that real GDP growth affects the savings rate negatively and statistically significant across most of the specifications (regressions 1-6). If GDP growth controls effectively for uncertainty, than it could imply a precautionary savings motive: Households save more in times of low growth or recessions. Lower (expected) income leads to higher saving.
However, the outcome changes strikingly once only the "ultra-low" interest rate period between 2013 and 2015 is regarded (regressions 4-6). The sign of the interest rate effect changes and the coefficient becomes quite large: A decrease in the interest rate by 0.1 % is statistically significant associated with an increase in the savings rate by more than 1 %. Households save (very much) more observing a small decrease in the nominal interest rate on their bank account ("income effect"). The result remains valid even when including only the Euro periphery (and the coefficient rises), although the sample becomes critically small. The regressions explain about 30 % of the variation in the savings rate. The results remain basically valid, though less significant, when we use the term deposit rates (agreed maturity up to two years) instead of the overnight deposit rates. Arguably, the small-scale exercise is simplistic evidence, but it leaves a discomforting aftertaste.