R&D investment under stress and uncertainty: the case of Argentina

ABSTRACT

Firms in developing economies are subject to macroeconomic fluctuations and policy swings, which generate uncertainty about the future behaviour of key variables that condition their return to innovation. In this paper, we explore how macroeconomic uncertainty and past exposure to macroeconomic shocks affect R&D investments. Using firm-level data on Argentina for the period 1992–2001, we find that firms that have experienced more shocks are less likely to commit to R&D investment and, also, that macroeconomic uncertainty prevents firms’ investment in R&D. We find that this effect is stronger for firms that have experienced more macroeconomic-shocks and weaker for firms with higher levels of foreign ownership.

KEYWORDS:

• Macroeconomic uncertainty
• experience of shocks
• R&D investment
• foreign direct investment (FDI)
• developing countries
• Argentina

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 There is previous macro-level research investigating the impact of different types of uncertainty on several outcomes, including R&D. Results are mixed. Tajaddini and Gholipour (2020) use data for 19 countries during 1996–2015 and show that higher levels of economic political uncertainty are associated positively with higher R&D expenditure per capita and higher levels of innovation output (patent and trademark applications, patent grants). In contrast, Ivus and Wajda (2018) study 30 countries over the period 1982–2012, and find that macro uncertainty fluctuations reduce R&D growth and, especially, business enterprise expenditure on R&D. In addition, some recent theoretical developments show that a lagging industrial structure can promote higher levels of macroeconomic volatility when countries catch up technologically and grow (e.g., Spinola 2020).

2 There was explicit interest in the reactivation of investment decisions. The Ministry for the Economy 1992 Annual Report (p. 2) claims that, the long-term economic recession had resulted in a deteriorated productive infrastructure and, therefore, stimulating investment was the only way to achieve economic growth (quoted in Galiani, Heymann, and Tommasi 2003, 26).

3 Had the economic actors been confident that the currency board regime would not change, interest rates would have remained relatively similar over the period; however, the differences were quite pronounced which reflected this lack of confidence. The pessimists were proved right and the regime crashed in December 2001, marking the beginning of one of the country’s biggest economic crises. Public debt reached US$144.453 million in 2001, around 50% of GDP in 2001 and 5 times Argentina’s exports in 2001 (Argentinean Finance Secretariat 2005); unemployment tripled between the beginning and the end of the Convertibility era, reaching 17.4% in 2001, while the percentage of the population living on less than US$2 per day rose by 148% between 1992 (no data available for 1991) and 2001, when it accounted for 14.3% of the population.

4 Since exports do not react to economic growth at the same speed as imports and, especially, in the context of a national currency appreciation, the current account deficit needs to be financed by net capital inflows to the financial accounts. If this does not happen, there is a balance of payment crisis. These dynamics of growth and abrupt stops due to balance of payment crises have been described as ‘stop & go’ phases and are a recurring phenomenon in Latin America (Moreno-Brid, Caldentey, and Npoles 2004).

5 According to UNCTAD statistics, in the 1990s, Brazil was ranked 1st for FDI inflow value.

6 We built the panel based on firms present in the two IS waves (1992-1996 and 1998-2001). Use of a balanced panel is justified on two grounds: (a) because the volatility measures we use for the analysis (see further in this section) are time-invariant for the whole period and, hence, we need to observe the dependent variables over the whole period, which could not be guaranteed with an unbalanced panel; (b) volatility is measured at the 3-digit level, but firms are linked to their 3-digit industry only in the 1992–1996 IS. In the other cases, firms are classified at the 2-digit level. Hence, a balanced panel allows us to more accurately connect firm industry to volatility. To ensure external validity of the balanced panel, we performed chi-squared tests on the annual distribution of firms across sectors, size categories and domestic/foreign ownership, between the balanced and the unbalanced panels. We found that, for the 1998–2001 period, there were no significant differences in terms of sectors and nationality and only a small difference in terms of size. Since large firms are usually included in all IS waves, we would expect the balanced panel sub-sample to be biased towards large size. Although significant, the absolute differences are not important: for the period 1998-2001, the proportion of large firms in the balanced sub-sample is 27% and in the full sample is 23%, while the proportion of small firms is 51% in the sub-sample versus 56% in the full sample.

7 In both surveys, employment data refer to the extreme years of the period covered by each survey (i.e., 1992, 1996, 1998, and 2001). For all other periods, the data are estimated using the firm-specific employment growth rates for the relevant period.

8 The National Bureau of Economic Research defines a recession as 2 consecutive quarters of real GDP decline. Since we do not have a long series of quarterly data for Argentina, we counted only years with negative GDP growth. We have quarterly data only from 1969, which allowed us to confirm that, in the period 1969-2001, all years with two consecutive quarters of negative GDP growth were also years with negative annual GDP growth.

9 Since the Ministry for the Economy of Argentina data base does not include specific information on profits, rates of growth of sectoral sales are used to proxy for profit growth rates.

10 Original data are in current US dollars deflated using the US wholesale price index; the variable then is expressed in (millions of) constant 1985 US dollars.

11 Sectoral labour productivity was constructed using all the information provided by the IS (unbalanced panel).

12 Although Age shows some variability over time (i.e., it increases by 1 year for all firms), we decided to retain it in the second step since, the inclusion of time FE in the first step, means that its explanatory power is cross-sectional. In addition, since it is correlated to shock experience, we prefer to control for it explicitly when analysing the effect of shock experience on R&D.

13 In both cases, the turning point occurs at the end of the size distribution, around the 99th percentile.

14 Import_Volatility is measured in millions of 1985 US dollars, with a mean of 0.3 and standard deviation of 0.15. Thus, if uncertainty increases by 1 standard deviation (0.15), the probability to invest in R&D decreases by at least 14 percentage points (-0.84*0.15) and R&D_Intensity decreases by at least 16%. (-1.08*0.15).

15 The exception is Profit_Volatility on R&D_Intensity, which is not significant.

16 We thank an anonymous reviewer for this suggestion.

Additional information

Funding

This work was supported by Agencia Nacional de Promoción de la Investigación, el Desarollo Tecnológico y la Innovación [Grant Number PICT 1331/2017].