RDP 2024-01: Do Monetary Policy and Economic Conditions Impact Innovation? Evidence from Australian Administrative Data 5. Through Which Channels Does Monetary Policy Affect Innovation?
February 2024
- Download the Paper 1.23MB
As discussed in the literature review, there are several channels through which monetary policy and economic conditions could lower the amount of innovative activity in the economy. One is the demand channel: contractionary monetary policy may weaken aggregate demand and therefore lower the returns to innovation. Another is the credit constraint channel: monetary policy may lead to tighter credit conditions and make cash flow constraints more binding for some firms. In turn this may lessen their ability to fund and undertake innovative activity. To consider the importance of these channels we compare outcomes for firms that we would expect to be more exposed to the channels to outcomes for firms that we would expect to be less exposed.
5.1 Demand channel
Changing aggregate demand in the economy is an important channel for monetary policy. As a contractionary monetary policy shock lowers aggregate demand in the economy, it also lowers the potential future profits of a firm. As the likelihood of future profits is the key reason for firms to innovate (Aghion and Howitt 2008), the probability of firms innovating can decrease as aggregate demand falls. The same can be said for expansionary policy, with the directions flipped.
To consider the importance of the demand channel, we examine outcomes for exporting and non-exporting firms. Exporting firms are likely to be less exposed to domestic conditions, as their sales are not determined exclusively by the domestic economy. To do this we interact our shock variable with a dummy variable that takes the value of one if a firm ever exported while in the sample and zero otherwise. We use this to trace out the effect of a shock separately for exporters and non-exporters.
Consistent with the demand channel playing an important role, the negative effect of contractionary monetary policy on innovation is less evident for exporters (Table 2). These results are similar for SMEs and large firms (Table B5), though there is only a significant difference between exporters and non-exporters for the large firms. This suggests that the results do not simply reflect the fact that exporters tend to be larger, and that larger firms tend to increase innovation in response to a contractionary shock.
| Year 0 | Year 1 | Year 2 | Year 3 | |
|---|---|---|---|---|
| Exporters | ||||
| Effect | 0.24 (2.12) |
1.60 (1.96) |
4.29 (2.49) |
8.83* (4.01) |
| R2 | 0.19 | 0.12 | 0.09 | 0.06 |
| No of observations | 17,974 | 14,734 | 11,672 | 8,904 |
| Non-exporters | ||||
| Effect | −1.19 (1.78) |
−5.58*** (1.43) |
−1.14 (1.92) |
4.59** (1.93) |
| R2 | 0.19 | 0.12 | 0.10 | 0.10 |
| No of observations | 27,079 | 20,901 | 14,630 | 8,632 |
| Difference by exporter status significant at: | na | 1 per cent level | 1 per cent level | na |
|
Notes: Significance assessed using T ‐ distribution with t – n degrees of freedom as suggested by Cameron and Miller (2015) to account for small number of clusters, where t is sample length and n is number of coefficients. All regressions include controls for industry, (lag) GDP growth, (lag) inflation, (lag) growth in the exchange rate, (lag) turnover growth and (lag) employment, and lag of the shock and dependent variable. ***, ** and * denote statistical significance at the 1, 5 and 10 per cent levels, respectively. Standard errors are shown in parentheses and are clustered at an annual level. |
||||
It is important to note that there could be other differences between exporting and non-exporting firms that affect these results. For example, exporting firms tend to be more productive and may have better management. To try to isolate the demand effects, we next re-estimate the regression, but use a measure of US monetary policy shocks taken from Choi, Willems and Yoo (2023).[10] If the previous results reflected the demand channel, rather than other inherent differences between exporters and non-exporters, we might expect exporters to respond more strongly to overseas monetary policy shocks as they are more directly exposed to foreign demand than are non-exporting firms. This is the ‘trade channel’ of policy spillovers (Arbatli-Saxegaard et al 2022). This exercise is also valuable as it provides (to our knowledge) the first evidence on spillovers from US monetary policy onto innovation in another country.
The first thing to note is that contractionary US monetary policy shocks do appear to lower the share of firms innovating in Australia (Table 3). The effects appear larger, compared to the domestic shock. However, it is difficult to compare the magnitudes given the different implied persistence of the shocks.
| Year 0 | Year 1 | Year 2 | Year 3 | |
|---|---|---|---|---|
| All firms | ||||
| Effect | −5.73 (5.40) |
−11.02*** (3.20) |
−8.16 (5.17) |
−7.59 (8.88) |
| R2 | 0.20 | 0.14 | 0.11 | 0.09 |
| No of observations | 45,053 | 35,635 | 26,302 | 17,536 |
| Exporters | ||||
| Effect | −7.38 (6.97) |
−10.86* (5.37) |
−9.76 (6.29) |
−11.28 (10.17) |
| R2 | 0.19 | 0.12 | 0.09 | 0.06 |
| No of observations | 17,974 | 14,734 | 11,672 | 8,904 |
| Non-exporters | ||||
| Effect | −4.53 (5.04) |
−10.61** (3.79) |
−6.71 (4.95) |
4.59 (5.89) |
| R2 | 0.19 | 0.12 | 0.10 | 0.09 |
| No of observations | 27,079 | 20,901 | 14,630 | 8,632 |
| Difference by exporter status significant at: | na | 1 per cent level | 1 per cent level | na |
|
Notes: Significance assessed using T ‐ distribution with t – n degrees of freedom as suggested by Cameron and Miller (2015) to account for small number of clusters, where t is sample length and n is number of coefficients. All regressions include controls for industry, (lag) GDP growth, (lag) inflation, (lag) growth in the exchange rate, (lag) turnover growth and (lag) employment, and lag of the shock and dependent variable. ***, ** and * denote statistical significance at the 1, 5 and 10 per cent levels, respectively. Standard errors are shown in parentheses and are clustered at an annual level. |
||||
The second key finding is that the response for exporters appears if anything larger than for non-exporters, which is the opposite of the findings when we consider a shock to Australian monetary policy. This is observed for both SMEs and large firms (Table B6) and is consistent with other work focusing on the response of firm-level investment and sales in foreign countries following US policy shocks (Arbatli-Saxegaard et al 2022). While the differences are not statistically significant, the greater sensitivity of exporters to foreign shocks, and of non-exporters to domestic shocks, provides further evidence of the importance of the demand channel.
5.2 The credit constraint channel
Broadly defined, we can think of the credit constraint channel of monetary policy as capturing any way in which contractionary monetary policy (or weaker economic conditions) makes it harder for a firm to finance innovative activity. This can reflect: tighter aggregate credit supply; lower asset prices which reduce borrowers' collateral value and therefore borrowing capacity, the ‘financial accelerator channel’ (Bernanke, Gertler and Gilchrist 1999); or lower liquidity due to lower revenue or higher interest payments making existing financing constraints more binding (Jeenas 2023).
To examine the credit constraint channel directly, we explore an additional question in the BCS. This question asks whether a lack of access to additional funds is hampering the business's ability to innovate. We examine whether monetary policy shocks lead to a change in the share of firms reporting that lack of access to additional funds is limiting their innovative activity.[11]
Consistent with the credit constraint channel being important, contractionary monetary policy shocks lead to an increase in the likelihood that firms report that lack of funds is significantly hampering their ability to undertake innovation (Table 4). This is almost entirely driven by SMEs, which is consistent with the evidence that SMEs have a larger decline in innovation following a monetary policy shock. To put the results in context, over the sample around 17 per cent of SMEs note that a lack of funds is hampering their innovation (compared to 8 per cent for large firms). So a 100 basis point shock (which is very large historically) would cause the share of SMEs noting that a lack of funds are hampering innovation to increase by around 15 to 20 per cent. These findings are also consistent with the broader literature that finds that SMEs are more likely to be credit or cash constrained than larger firms – for example, Mancusi and Vezzulli (2010) and Bakhtiari et al (2020).
This provides fairly strong evidence that the credit constraint channel is important. As a further test, but one which relies less heavily on self-reported data, we examine whether the results differ for foreign- and domestically owned firms. Foreign-owned firms may be better able to access credit from overseas markets or from their overseas parent (Dahlquist and Robertsson 2001). Thus, they may be more sensitive to the global, not the domestic, cost of capital. As such, they may be less affected by domestic credit conditions.[12]
To look at whether foreign ownership mitigates the effect of monetary policy on innovation, we interact the monetary policy shock variable with a dummy variable for foreign ownership and then trace out the effects for domestically- and foreign-owned firms. If a firm indicates it has any level of foreign ownership (for the period that it is visible in our data), it is categorised as foreign owned. For this analysis we focus on large firms only, given the number of foreign-owned SMEs is small.
There is some evidence that contractionary monetary policy has less of a negative effect on foreign-owned firm innovation, compared to domestically owned firms (Table 5). While this evidence is less direct and could reflect factors other than the ability to access overseas financing, it is consistent with the credit supply channel playing an important role in the transmission of monetary policy to innovative activity and reinforces the above, more direct results.
| Year 0 | Year 1 | Year 2 | Year 3 | |
|---|---|---|---|---|
| All firms | ||||
| Effect | 1.40** (0.63) |
2.06*** (0.63) |
1.44 (0.99) |
2.37*** (0.54) |
| R2 | 0.28 | 0.32 | 0.35 | 0.39 |
| No of observations | 42,138 | 32,901 | 23,826 | 15,245 |
| SMEs | ||||
| Effect | 2.02* (1.00) |
2.92*** (0.70) |
2.45* (1.30) |
3.67*** (0.65) |
| R2 | 0.28 | 0.32 | 0.34 | 0.37 |
| No of observations | 28,216 | 20,966 | 13,518 | 6,630 |
| Large firms | ||||
| Effect | 0.16 (0.49) |
0.63 (1.41) |
0.25 (1.19) |
1.10 (0.83) |
| R2 | 0.28 | 0.34 | 0.39 | 0.42 |
| No of observations | 13,922 | 11,935 | 10,308 | 8,885 |
| Difference by size significant at: | na | na | 10 per cent level | 10 per cent level |
|
Notes: Significance assessed using T ‐ distribution with t – n degrees of freedom as suggested by Cameron and Miller (2015) to account for small number of clusters, where t is sample length and n is number of coefficients. All regressions include controls for industry, (lag) GDP growth, (lag) inflation, (lag) growth in the exchange rate, (lag) turnover growth and (lag) employment, and lag of the shock and dependent variable. ***, ** and * denote statistical significance at the 1, 5 and 10 per cent levels, respectively. Standard errors are shown in parentheses and are clustered at an annual level. |
||||
| Year 0 | Year 1 | Year 2 | Year 3 | |
|---|---|---|---|---|
| Foreign owned | ||||
| Effect | 6.78** (2.96) |
6.89** (3.09) |
7.58** (3.34) |
9.52* (4.80) |
| R2 | 0.20 | 0.11 | 0.08 | 0.07 |
| No of observations | 9,556 | 8,402 | 7,382 | 6,528 |
| Domestically owned | ||||
| Effect | −1.73 (2.22) |
−4.10 (3.18) |
0.81 (3.02) |
9.99 (5.79) |
| R2 | 0.19 | 0.10 | 0.07 | 0.05 |
| No of observations | 5,943 | 5,048 | 4,304 | 3,717 |
| Difference by size significant at: | 1 per cent level | 10 per cent level | 1 per cent level | na |
|
Notes: Significance assessed using T ‐ distribution with t – n degrees of freedom as suggested by Cameron and Miller (2015) to account for small number of clusters, where t is sample length and n is number of coefficients. All regressions include controls for industry, (lag) GDP growth, (lag) inflation, (lag) growth in the exchange rate, (lag) turnover growth and (lag) employment, and lag of the shock and dependent variable. ***, ** and * denote statistical significance at the 1, 5 and 10 per cent levels, respectively. Standard errors are shown in parentheses and are clustered at an annual level. |
||||
Footnotes
This measure is based on high-frequency changes in bond yields. We use this measure as it is available for a wide range of countries, which could facilitate future analysis. [10]
The survey question asks firms about what factors are hampering their innovative activity. As well as a lack of access to additional funds, firms can note: lack of skilled workers; cost of development; regulations and compliance; or uncertain demand for new goods/services. Firms can select more than one factor. [11]
That said, the global cost of capital may affect domestic costs. [12]