Purpose: Portugal's real estate prices recently increased sharply (INE, 2023). Although data from Portugal's central bank shows a decrease in real estate developers' financial debt (Banco de Portugal, 2022), the duration and speed of price increases could threaten financial stability if this trend continues or is accentuated. The surge in global liquidity, especially in advanced economies, has been pointed out as one of the main sources of real estate price increases (Chien & Liu, 2023). Recent research shows that, in Portugal, real estate developers seem not to gain abnormal returns from house price increases (Cunha et al., 2023). This finding suggests that other factors determine real estate developers' investment decisions. According to Tobin (1969), the corporate investment rate depends on the ratio between the firm's market value and its replacement cost (Tobin's Q), where a ratio higher than one means that investors believe there are investment opportunities for the company to grow. However, to invest, the company needs funding, and if it relies on external funding, it incurs additional costs due to agency costs and information asymmetry (Myers & Majluf, 1984). This motivates managers to use internal resources, such as cash flow, to finance their investments (Pinegar & Wilbricht, 1989). Fazzari et al. (1988) categorized firms as financially constrained or unconstrained, demonstrating that cash flow significantly impacts investment decisions, particularly in more constrained firms. Some recent studies show that investment sensitivity to cash flows is decreasing (Chen & Chen, 2012; Larkin et al., 2018), coinciding with a low real interest rates period. Verona (2020) also found that investment's sensitivity to Tobin's Q declines over time, while investment's sensitivity to cash flow declines with business cycles but remains largely stable over the medium-to-long run. Studies based on Fazzari et al. 's (1988) investment sensitivity to cash flows model rely on Tobin's Q. The study of non-listed companies' sensitivity is scarce or inexistent. Moreover, there are no studies about the real estate industry's investment sensitivity to cash flow. This study analyzes the relationship between cash flows and Tobin's Q of Portuguese real estate development non-listed companies and their investment levels to find if those companies have their investment decisions dependent on their internal resources or if other factors such as Tobin's Q determine their investment levels. Methodology: The following equation (1) describes Fazzari et al. (1988) model:... I represent the investment of the real estate development company i in the calendar year t, measured by the difference between invested capital (K) in year t and year t-1. K represents the company's stock of invested capital, measured by non-current assets plus inventories plus customers minus suppliers. CF stands for operating cash flow, with the subscript t-1 referring to the previous year; EV represents the company's market value (enterprise value), and EV/K represents Tobin's Q. v represents the error term. In this model, the dependent variable is the investment rate of each company, the independent variables are the cash flow rate, and Tobin's Q. f and g are the coefficients to estimate. We acknowledge that the model carries a risk of endogeneity, but this is the original specification used in several studies throughout the years to study stock-listed companies' investment sensitivity. We intended to apply this equation to non-listed companies which do not have a market value. We will generate a proxy valuation for each company to be used in the equation as the EV variable, thus allowing for the estimation of Tobin's Q. When valuing stock-listed companies, financial analysts use multi-period income approaches such as the discounted cash flow (DCF) and the residual income (RI) methods, single period market approach such as the market multiples (MM) method, and the hybrid approach, where the two approaches are combined to find an average valuation (Erkilet et al., 2022). The weights of each method depend on the industry specificities and whether the financial statements properly capture the value of the firm's tangible and intangible assets (Demirakos et al., 2004), with industries that have predominantly tangible assets (such as the real estate development industry) showing some preference for multi-period income approaches (DCF and RI), and using the MM to obtain confirmation of the accuracy of their valuation (Ali & Khalidi, 2020). As we do not have previous real estate development industry research suggesting the weights for each method, we will estimate each company's EV as the unweighted average of three valuation methods: DCF, RI, and MM. We will not consider future growth in cash flows in any of the companies, as we will estimate the EV for all companies under the same conditions. To calculate the EV using the DCF method we use the cash flows of the year t+1, divided by the weighted average cost of capital (WACC), as shown in equation 2:... NOPAT is the net operating profit after taxes, and invested capital is K. We find the Enterprise Value by adding the invested capital to the present value of the EVA in perpetuity, calculated by the ratio EVA/WACC. To calculate the EV using the MM method, we used Damodaran's database and took the market multiple of the EV/Sales for the real estate developers' industry and each year studied. We multiply it by each company's sales, thus obtaining the EV for each company. The average of the three estimated EVs for each company will be inserted in Equation 1. Regarding the source of the data, we collected the sample series of accounting data from SABI, a Moody's Analytical database, from 2012 (first date available) to 2020, according to the following criteria: First, companies with main activity in the Nomenclature of Economic Activities (NACE) code 411 (real estate development). Second, companies with at least one employee and financial information for the entire analysis period. This process resulted in a sample of 415 companies, with 9 annual accounts, resulting in a sample with 3320 observations from which we developed panel data with the necessary accounting information for estimating equation 1 f and g coefficients. Results: The tests show the presence of fixed effects. We estimated equation 1 coefficients using the ordinary least squares with fixed effects estimator, with the following results: f = 2,4 (p-value <0.01), g without statistical significance. We conclude that real estate developers invest according to last year's cash flows. This confirms the literature according to which financially constrained companies rely on internal funding for investment. Portugal's central bank also confirmed that the indebtedness of companies has been decreasing, meaning that companies resort less to external funding. The data also shows that invested capital has been decreasing, which could be one of the causes of last decade's real estate price increases. Research limitations: We estimate Tobin's Q with the unweighted average of three subjective valuation models as a proxy for the enterprise's market value, and the results are limited to Portugal's real estate development companies. Originality: This is the first study that uses accounting information to create a proxy of companies' market value to apply Tobin's Q to non-listed companies. [ABSTRACT FROM AUTHOR]