The neo-classical theory of investment argues that firm’s investment is determined by economic fundamentals, and not by financial variables such as cash flow (Melander et al., 2017). New developments of this theory show that, in the presence of financial frictions and asymmetry of information, investment is inter-alia a function of cash flow (see for example the Hayashi’s (1982) neoclassical investment model). However, the interpretation of the investment – cash flow nexus is controversial (Carpenter and Guariglia, 2008). This relationship is influenced on the one hand by financial constraints and, on the other hand, by investment opportunities. A common solution for the second problem in the early literature was to include the Tobin’s Q (the ratio between the market value and the replacement value of a physical asset) in the empirical regressions explaining the investment determinants. From a theoretical point of view, it is expected that the cash flow have a positive influence on the investment decision, as it represents an internal funding source. However, noteworthy empirical findings report a negative relationship between investment and cash flow. Therefore, researchers try to explain the puzzling empirical findings on investment – cash flow sensitivities, underlining the role of firm’s asset liquidity (Flor and Hirth, 2013), of firm’s financial constraints (Kim, 2014), and of monetary transmission channels (Chatelain et al., 2003). We contribute to the existing literature in several ways. First, using firm-level data, we test the bidirectional causality between investment and cash flow. Previous studies do not focus on the role of investment on cash flow, although the endogeneity issue is sometimes addressed in empirical papers (i.e. Bond and Meghir, 1994). It is well known that investment should be made if and only if the present value of future revenues equals the opportunity cost of capital. Accordingly, an investment should generates a stream of cash flows during the next periods. It is surprising that the empirical literature does not focus on the bidirectional causality between investment and cash flow, which might explain the divergent results reported so far. To show this, we use a panel data vector autoregression method (pVAR) and we investigate the bidirectional Granger causality between investment (the ratio between fixed assets and total assets) and the cash flow level (as percentage of operational revenues) for 739 companies from the agriculture field in Romania (a similar empirical approach is used by Melander et al., 2017). We also search how shocks in cash flow/investment are transmitted to investment/cash flow, using impulse response functions.